Diefolio
English

GeForce RTX 5090 vs GeForce RTX 4090

GeForce RTX 5090 vs GeForce RTX 4090: GeForce RTX 5090 has 32GB of dedicated VRAM and 21,760 CUDA cores. Its published memory bandwidth is 1,792GB/s. / GeForce RTX 4090 has 24GB of dedicated VRAM and 16,384 CUDA cores.

Comparison notes

Compare the documented configurations and measurement conditions. VRAM, bandwidth and compute configuration are separate specifications. AI TOPS is not converted into game FPS or model generation speed. Check measurements with the software, driver and precision required by the workload. 24GB of VRAM does not guarantee compatibility with every AI model. Check the model, quantization, context length and runtime; CUDA core counts across generations are not measured performance ratios.

Reviewed: 2026-10-11

Explore workloads and conditions →

Evidence at a glance · 20

Benchmarks and test conditions

3DMark Time Spy · Graphics

GeForce RTX 5090 45,432 points

Source aggregate · Median · Source sample size: 7

Time Spy (build unknown) · GPU

GeForce RTX 4090 36,671 points

Source aggregate · Median · Source sample size: 39

Time Spy (build unknown) · GPU

Inspect full results for conditions and sources. Conditions can differ per product.

3DMark Fire Strike · Graphics

GeForce RTX 5090 72,390 points

Source aggregate · Median · Source sample size: 5

Fire Strike (build unknown) · GPU

GeForce RTX 4090 75,641 points

Source aggregate · Median · Source sample size: 37

Fire Strike (build unknown) · GPU

Inspect full results for conditions and sources. Conditions can differ per product.

Geekbench GPU 6.7 OpenCL

GeForce RTX 5090 383,833 points

Source aggregate · Median · Source sample size: 3

6.7 · GPU

GeForce RTX 4090 328,983 points

Source aggregate · Median · Source sample size: 21

6.7 · GPU

Inspect full results for conditions and sources. Conditions can differ per product.

All measurement results (28)

Up to 24 results with representative scores are shown per product, prioritizing shared test conditions. Explore the comparison tool for all results and individual submissions. Results with matching tests, versions and environments appear in the same row. An unavailable result is missing evidence in this selection, not a zero score. These results do not establish an OS effect or an overall winner.

Parameter and conditionsGeForce RTX 5090GeForce RTX 4090
Both products have resultsBlender · 5.1.1 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "5.1.1"
}

Blender scene rendering

3,962.6515418482463 samples/min

Median and quartilesn = 31

5.1.1 · classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
3,903.37
P50 · Median
3,962.65
P75
4,012.76

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,421.65 / 4,443.72
Measurement dates
2026-04-19 – 2026-08-27
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 31

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3962.6515418482463,
    "observed_max": 4443.7167485473665,
    "observed_min": 3421.645786574691,
    "percentiles": {
      "p25": 3903.372832365536,
      "p50": 3962.6515418482463,
      "p75": 4012.7554096067256
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 31,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "ce5f2f2334eadd2cab98",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f5d03ce8-270c-4afe-8785-677346d60ac7",
    "1575b8b7-3554-4f66-922b-b55cc55380fa",
    "1febfee0-ae79-4a79-b7cb-944e2d6e70c6"
  ],
  "measured_from": "2026-04-19T16:10:28.777075+00:00",
  "measured_to": "2026-08-27T00:45:55.864618+00:00",
  "median": 3962.6515418482463,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3903.372832365536,
    "p50": 3962.6515418482463,
    "p75": 4012.7554096067256
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 31,
  "sample_status": "quartiles"
}

Blender scene rendering

2,761.484129730943 samples/min

Provisional median · Limited submissionsn = 25

5.1.1 · classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

2,761.48 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,430.37 / 3,084.14
Measurement dates
2026-05-06 – 2026-07-23
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 25

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2761.484129730943,
    "observed_max": 3084.144109959965,
    "observed_min": 2430.3673355088586,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 25,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "0a8a06589960539c89a4",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "a04f0044-12b7-4fbd-8558-eb901a78347b",
    "b5b6c3f8-ec31-4adf-80f8-38a72648601a",
    "4969a621-990d-4dc0-8068-f87afad862d1"
  ],
  "measured_from": "2026-05-06T16:17:53.364298+00:00",
  "measured_to": "2026-07-23T20:25:01.159256+00:00",
  "median": 2761.484129730943,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 25,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "5.1.1"
}

Blender complete benchmark

16,242.08735939134 samples/min

Median and quartilesn = 31

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
16,040.67
P50 · Median
16,242.09
P75
16,358.36

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
15,296.8 / 17,448.64
Measurement dates
2026-04-19 – 2026-08-27
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 31

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 16242.08735939134,
    "observed_max": 17448.644549686185,
    "observed_min": 15296.801454181732,
    "percentiles": {
      "p25": 16040.668141912362,
      "p50": 16242.08735939134,
      "p75": 16358.356666781438
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 31,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "eef3dc1b47858cfc1d1c",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f5d03ce8-270c-4afe-8785-677346d60ac7",
    "1575b8b7-3554-4f66-922b-b55cc55380fa",
    "1febfee0-ae79-4a79-b7cb-944e2d6e70c6"
  ],
  "measured_from": "2026-04-19T16:10:28.777075+00:00",
  "measured_to": "2026-08-27T00:45:55.864618+00:00",
  "median": 16242.08735939134,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 16040.668141912362,
    "p50": 16242.08735939134,
    "p75": 16358.356666781438
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 31,
  "sample_status": "quartiles"
}

Blender complete benchmark

11,321.40509884421 samples/min

Provisional median · Limited submissionsn = 25

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

11,321.41 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
9,764.59 / 12,347.31
Measurement dates
2026-05-06 – 2026-07-23
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 25

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 11321.40509884421,
    "observed_max": 12347.309495451687,
    "observed_min": 9764.590394450453,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 25,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "1dab7407dccf99b0d037",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "a04f0044-12b7-4fbd-8558-eb901a78347b",
    "b5b6c3f8-ec31-4adf-80f8-38a72648601a",
    "4969a621-990d-4dc0-8068-f87afad862d1"
  ],
  "measured_from": "2026-05-06T16:17:53.364298+00:00",
  "measured_to": "2026-07-23T20:25:01.159256+00:00",
  "median": 11321.40509884421,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 25,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "5.1.1"
}

Blender scene rendering

5,139.655460779394 samples/min

Median and quartilesn = 31

5.1.1 · junkshop · OPTIX · Windows · 1P · 8 cores · 16 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
5,057.18
P50 · Median
5,139.66
P75
5,205.85

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,769 / 5,431.9
Measurement dates
2026-04-19 – 2026-08-27
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 31

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5139.655460779394,
    "observed_max": 5431.89853691969,
    "observed_min": 4769.000639249682,
    "percentiles": {
      "p25": 5057.175929401931,
      "p50": 5139.655460779394,
      "p75": 5205.851399810038
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 31,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "1c7853391cd194a835be",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f5d03ce8-270c-4afe-8785-677346d60ac7",
    "1575b8b7-3554-4f66-922b-b55cc55380fa",
    "1febfee0-ae79-4a79-b7cb-944e2d6e70c6"
  ],
  "measured_from": "2026-04-19T16:09:51.603794+00:00",
  "measured_to": "2026-08-27T00:45:18.062243+00:00",
  "median": 5139.655460779394,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 5057.175929401931,
    "p50": 5139.655460779394,
    "p75": 5205.851399810038
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 31,
  "sample_status": "quartiles"
}

Blender scene rendering

3,484.2355275484206 samples/min

Provisional median · Limited submissionsn = 25

5.1.1 · junkshop · OPTIX · Windows · 1P · 8 cores · 16 threads

3,484.24 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,892.11 / 3,800.05
Measurement dates
2026-05-06 – 2026-07-23
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 25

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3484.2355275484206,
    "observed_max": 3800.048123315248,
    "observed_min": 2892.1067313454882,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 25,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "5104ae6c61bfdde48b65",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "a04f0044-12b7-4fbd-8558-eb901a78347b",
    "b5b6c3f8-ec31-4adf-80f8-38a72648601a",
    "4969a621-990d-4dc0-8068-f87afad862d1"
  ],
  "measured_from": "2026-05-06T16:17:18.190645+00:00",
  "measured_to": "2026-07-23T20:24:19.447929+00:00",
  "median": 3484.2355275484206,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 25,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "5.1.1"
}

Blender scene rendering

7,158.3192931880185 samples/min

Median and quartilesn = 31

5.1.1 · monster · OPTIX · Windows · 1P · 8 cores · 16 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
7,014.87
P50 · Median
7,158.32
P75
7,201.66

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
6,714.54 / 7,625.63
Measurement dates
2026-04-19 – 2026-08-27
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 31

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 7158.3192931880185,
    "observed_max": 7625.627621309495,
    "observed_min": 6714.541225775251,
    "percentiles": {
      "p25": 7014.865764840695,
      "p50": 7158.3192931880185,
      "p75": 7201.655169505284
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 31,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "3b375e39eaceb3532bae",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f5d03ce8-270c-4afe-8785-677346d60ac7",
    "1575b8b7-3554-4f66-922b-b55cc55380fa",
    "1febfee0-ae79-4a79-b7cb-944e2d6e70c6"
  ],
  "measured_from": "2026-04-19T16:09:10.365282+00:00",
  "measured_to": "2026-08-27T00:44:31.762485+00:00",
  "median": 7158.3192931880185,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 7014.865764840695,
    "p50": 7158.3192931880185,
    "p75": 7201.655169505284
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 31,
  "sample_status": "quartiles"
}

Blender scene rendering

5,022.506326099379 samples/min

Provisional median · Limited submissionsn = 25

5.1.1 · monster · OPTIX · Windows · 1P · 8 cores · 16 threads

5,022.51 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,169.53 / 5,463.12
Measurement dates
2026-05-06 – 2026-07-23
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 25

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5022.506326099379,
    "observed_max": 5463.117262176474,
    "observed_min": 4169.525838246759,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 25,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "6d0b21353e9d9c7c546f",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "a04f0044-12b7-4fbd-8558-eb901a78347b",
    "b5b6c3f8-ec31-4adf-80f8-38a72648601a",
    "4969a621-990d-4dc0-8068-f87afad862d1"
  ],
  "measured_from": "2026-05-06T16:16:38.529015+00:00",
  "measured_to": "2026-07-23T20:23:27.180147+00:00",
  "median": 5022.506326099379,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 25,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.0 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "5.1.0"
}

Blender scene rendering

3,969.2070125345476 samples/min

Median and quartilesn = 42

5.1.0 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
3,838.15
P50 · Median
3,969.21
P75
4,020.9

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,274.14 / 4,163.8
Measurement dates
2026-03-20 – 2026-08-04
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 42

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3969.2070125345476,
    "observed_max": 4163.800785661752,
    "observed_min": 2274.1416257693186,
    "percentiles": {
      "p25": 3838.1459649543735,
      "p50": 3969.2070125345476,
      "p75": 4020.90244268857
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 42,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.0"
  },
  "condition_id": "7595c878c188459d00e9",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "54052f79-2e91-44ab-91d0-c9d2272e2a90",
    "dc23181c-1e3d-449d-b71b-bc1f3136693b",
    "33a43f9d-7b32-4709-a46d-d2d64c28df1f"
  ],
  "measured_from": "2026-03-20T01:47:12.971877+00:00",
  "measured_to": "2026-08-04T15:06:09.335172+00:00",
  "median": 3969.2070125345476,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3838.1459649543735,
    "p50": 3969.2070125345476,
    "p75": 4020.90244268857
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 42,
  "sample_status": "quartiles"
}

Blender scene rendering

2,937.947846296585 samples/min

Provisional median · Limited submissionsn = 19

5.1.0 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

2,937.95 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
1,824.16 / 3,060.46
Measurement dates
2026-03-18 – 2026-07-21
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 19

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2937.947846296585,
    "observed_max": 3060.460580450145,
    "observed_min": 1824.1593984296032,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 19,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.0"
  },
  "condition_id": "9982f2186b6664130476",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f78e5c54-4815-4744-8084-6dda984ef82f",
    "46b8ffd2-fd78-4819-b632-7e80b6c67a99",
    "909bc081-c347-47d3-a11e-1c458eae0ee1"
  ],
  "measured_from": "2026-03-18T13:40:03.968766+00:00",
  "measured_to": "2026-07-21T21:16:17.132057+00:00",
  "median": 2937.947846296585,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 19,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.0 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "5.1.0"
}

Blender complete benchmark

16,216.410888936864 samples/min

Median and quartilesn = 42

5.1.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
15,779.77
P50 · Median
16,216.41
P75
16,426.66

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
9,064.24 / 17,141.83
Measurement dates
2026-03-20 – 2026-08-04
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 42

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 16216.410888936864,
    "observed_max": 17141.829035112198,
    "observed_min": 9064.238909742378,
    "percentiles": {
      "p25": 15779.769088240348,
      "p50": 16216.410888936864,
      "p75": 16426.664812530544
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 42,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.0"
  },
  "condition_id": "6171859c596784655304",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "54052f79-2e91-44ab-91d0-c9d2272e2a90",
    "dc23181c-1e3d-449d-b71b-bc1f3136693b",
    "33a43f9d-7b32-4709-a46d-d2d64c28df1f"
  ],
  "measured_from": "2026-03-20T01:47:12.971877+00:00",
  "measured_to": "2026-08-04T15:06:09.335172+00:00",
  "median": 16216.410888936864,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 15779.769088240348,
    "p50": 16216.410888936864,
    "p75": 16426.664812530544
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 42,
  "sample_status": "quartiles"
}

Blender complete benchmark

11,821.022054183262 samples/min

Provisional median · Limited submissionsn = 19

5.1.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

11,821.02 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
7,001.82 / 12,446.68
Measurement dates
2026-03-18 – 2026-07-21
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 19

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 11821.022054183262,
    "observed_max": 12446.681943580783,
    "observed_min": 7001.824467061057,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 19,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.0"
  },
  "condition_id": "6858edcfe93a7bf240c2",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f78e5c54-4815-4744-8084-6dda984ef82f",
    "46b8ffd2-fd78-4819-b632-7e80b6c67a99",
    "909bc081-c347-47d3-a11e-1c458eae0ee1"
  ],
  "measured_from": "2026-03-18T13:40:03.968766+00:00",
  "measured_to": "2026-07-21T21:16:17.132057+00:00",
  "median": 11821.022054183262,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 19,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.0 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "5.1.0"
}

Blender scene rendering

5,150.494727709769 samples/min

Median and quartilesn = 42

5.1.0 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
4,992.85
P50 · Median
5,150.49
P75
5,231.15

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,774.15 / 5,476.18
Measurement dates
2026-03-20 – 2026-08-04
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 42

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5150.494727709769,
    "observed_max": 5476.178815207121,
    "observed_min": 2774.151490029494,
    "percentiles": {
      "p25": 4992.8488090009405,
      "p50": 5150.494727709769,
      "p75": 5231.15143819951
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 42,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.0"
  },
  "condition_id": "e6b7f01f355caf28e560",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "54052f79-2e91-44ab-91d0-c9d2272e2a90",
    "dc23181c-1e3d-449d-b71b-bc1f3136693b",
    "33a43f9d-7b32-4709-a46d-d2d64c28df1f"
  ],
  "measured_from": "2026-03-20T01:46:36.387336+00:00",
  "measured_to": "2026-08-04T15:05:33.278236+00:00",
  "median": 5150.494727709769,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 4992.8488090009405,
    "p50": 5150.494727709769,
    "p75": 5231.15143819951
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 42,
  "sample_status": "quartiles"
}

Blender scene rendering

3,655.1840746715984 samples/min

Provisional median · Limited submissionsn = 19

5.1.0 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

3,655.18 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,317.15 / 3,895.26
Measurement dates
2026-03-18 – 2026-07-21
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 19

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3655.1840746715984,
    "observed_max": 3895.257680610732,
    "observed_min": 2317.1457227641736,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 19,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.0"
  },
  "condition_id": "1742c0e99a6fa99f36bf",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f78e5c54-4815-4744-8084-6dda984ef82f",
    "46b8ffd2-fd78-4819-b632-7e80b6c67a99",
    "909bc081-c347-47d3-a11e-1c458eae0ee1"
  ],
  "measured_from": "2026-03-18T13:39:26.141907+00:00",
  "measured_to": "2026-07-21T21:15:39.999755+00:00",
  "median": 3655.1840746715984,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 19,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.0 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "5.1.0"
}

Blender scene rendering

7,111.864149801046 samples/min

Median and quartilesn = 42

5.1.0 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
6,910.4
P50 · Median
7,111.86
P75
7,187.59

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,015.95 / 7,584.72
Measurement dates
2026-03-20 – 2026-08-04
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 42

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 7111.864149801046,
    "observed_max": 7584.720793195193,
    "observed_min": 4015.9457939435656,
    "percentiles": {
      "p25": 6910.403818486027,
      "p50": 7111.864149801046,
      "p75": 7187.593224322071
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 42,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.0"
  },
  "condition_id": "33083f6f41edd7ac7aa4",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "54052f79-2e91-44ab-91d0-c9d2272e2a90",
    "dc23181c-1e3d-449d-b71b-bc1f3136693b",
    "33a43f9d-7b32-4709-a46d-d2d64c28df1f"
  ],
  "measured_from": "2026-03-20T01:45:48.318882+00:00",
  "measured_to": "2026-08-04T15:04:49.539221+00:00",
  "median": 7111.864149801046,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 6910.403818486027,
    "p50": 7111.864149801046,
    "p75": 7187.593224322071
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 42,
  "sample_status": "quartiles"
}

Blender scene rendering

5,229.7214275624865 samples/min

Provisional median · Limited submissionsn = 19

5.1.0 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

5,229.72 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,860.52 / 5,527.17
Measurement dates
2026-03-18 – 2026-07-21
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 19

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5229.7214275624865,
    "observed_max": 5527.174751261398,
    "observed_min": 2860.51934586728,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 19,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.0"
  },
  "condition_id": "d741e0f6400baabf972a",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f78e5c54-4815-4744-8084-6dda984ef82f",
    "46b8ffd2-fd78-4819-b632-7e80b6c67a99",
    "909bc081-c347-47d3-a11e-1c458eae0ee1"
  ],
  "measured_from": "2026-03-18T13:38:41.686417+00:00",
  "measured_to": "2026-07-21T21:14:57.861374+00:00",
  "median": 5229.7214275624865,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 19,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "5.1.1"
}

Blender scene rendering

3,979.2911139199123 samples/min

Submitted benchmark distributionn = 101

5.1.1 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

2,287.48–2,402.86: 3 Submissions2,402.86–2,518.24: 2 Submissions2,518.24–2,633.61: 1 Submissions2,633.61–2,748.99: 1 Submissions2,748.99–2,864.36: 2 Submissions2,864.36–2,979.74: 0 Submissions2,979.74–3,095.11: 0 Submissions3,095.11–3,210.49: 1 Submissions3,210.49–3,325.87: 0 Submissions3,325.87–3,441.24: 1 Submissions3,441.24–3,556.62: 0 Submissions3,556.62–3,671.99: 1 Submissions3,671.99–3,787.37: 12 Submissions3,787.37–3,902.74: 13 Submissions3,902.74–4,018.12: 24 Submissions4,018.12–4,133.5: 22 Submissions4,133.5–4,248.87: 13 Submissions4,248.87–4,364.25: 3 Submissions4,364.25–4,479.62: 0 Submissions4,479.62–4,595: 2 Submissionsn = 242,287.484,595

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
2,625.51
P25
3,813.76
P50 · Median
3,979.29
P75
4,077.18
P95
4,211.68

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,287.48 / 4,595
Measurement dates
2026-04-15 – 2026-07-20
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 101

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 3,
        "from": 2287.4842400958523,
        "to": 2402.85998918607
      },
      {
        "count": 2,
        "from": 2402.85998918607,
        "to": 2518.2357382762875
      },
      {
        "count": 1,
        "from": 2518.2357382762875,
        "to": 2633.611487366505
      },
      {
        "count": 1,
        "from": 2633.611487366505,
        "to": 2748.9872364567227
      },
      {
        "count": 2,
        "from": 2748.9872364567227,
        "to": 2864.36298554694
      },
      {
        "count": 0,
        "from": 2864.36298554694,
        "to": 2979.738734637158
      },
      {
        "count": 0,
        "from": 2979.738734637158,
        "to": 3095.114483727375
      },
      {
        "count": 1,
        "from": 3095.114483727375,
        "to": 3210.4902328175926
      },
      {
        "count": 0,
        "from": 3210.4902328175926,
        "to": 3325.8659819078102
      },
      {
        "count": 1,
        "from": 3325.8659819078102,
        "to": 3441.241730998028
      },
      {
        "count": 0,
        "from": 3441.241730998028,
        "to": 3556.6174800882454
      },
      {
        "count": 1,
        "from": 3556.6174800882454,
        "to": 3671.993229178463
      },
      {
        "count": 12,
        "from": 3671.993229178463,
        "to": 3787.3689782686806
      },
      {
        "count": 13,
        "from": 3787.3689782686806,
        "to": 3902.7447273588978
      },
      {
        "count": 24,
        "from": 3902.7447273588978,
        "to": 4018.120476449116
      },
      {
        "count": 22,
        "from": 4018.120476449116,
        "to": 4133.496225539333
      },
      {
        "count": 13,
        "from": 4133.496225539333,
        "to": 4248.871974629551
      },
      {
        "count": 3,
        "from": 4248.871974629551,
        "to": 4364.247723719768
      },
      {
        "count": 0,
        "from": 4364.247723719768,
        "to": 4479.623472809986
      },
      {
        "count": 2,
        "from": 4479.623472809986,
        "to": 4594.999221900203
      }
    ],
    "independent_system_count": null,
    "median": 3979.2911139199123,
    "observed_max": 4594.999221900203,
    "observed_min": 2287.4842400958523,
    "percentiles": {
      "p25": 3813.7562940240455,
      "p5": 2625.509204869577,
      "p50": 3979.2911139199123,
      "p75": 4077.1842182555874,
      "p95": 4211.67954363406
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 101,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "6c783c8ec68b5af499e5",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f95d1091-6784-4fd0-9ad9-44cc26bf2875",
    "6f168d57-efcb-4c3b-9f65-e2cc349c4f09",
    "84912c43-1fbd-4f0e-932f-b58e88f7f64b"
  ],
  "measured_from": "2026-04-15T13:45:00.180174+00:00",
  "measured_to": "2026-07-20T11:55:36.157512+00:00",
  "median": 3979.2911139199123,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3813.7562940240455,
    "p5": 2625.509204869577,
    "p50": 3979.2911139199123,
    "p75": 4077.1842182555874,
    "p95": 4211.67954363406
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 101,
  "sample_status": "distribution"
}

Blender scene rendering

2,811.779618677315 samples/min

Median and quartilesn = 35

5.1.1 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
2,753.67
P50 · Median
2,811.78
P75
2,942.94

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,586.42 / 3,107.12
Measurement dates
2026-04-22 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 35

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2811.779618677315,
    "observed_max": 3107.1174277503605,
    "observed_min": 2586.4224656060783,
    "percentiles": {
      "p25": 2753.6740092129885,
      "p50": 2811.779618677315,
      "p75": 2942.9382404260896
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 35,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "a4f55de942eaf1be6feb",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "653847f4-19be-45de-b5bf-b9268163d711",
    "0065b816-541b-4b2f-9e81-89f78d039617",
    "978c67fb-d600-4f0f-9296-94aecf86675d"
  ],
  "measured_from": "2026-04-22T20:48:37.845051+00:00",
  "measured_to": "2026-07-11T08:06:47.871900+00:00",
  "median": 2811.779618677315,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 2753.6740092129885,
    "p50": 2811.779618677315,
    "p75": 2942.9382404260896
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 35,
  "sample_status": "quartiles"
}
Both products have resultsBlender · 5.1.1 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "5.1.1"
}

Blender complete benchmark

16,312.99010666296 samples/min

Submitted benchmark distributionn = 101

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

9,124.55–9,607.81: 2 Submissions9,607.81–10,091.07: 3 Submissions10,091.07–10,574.33: 0 Submissions10,574.33–11,057.59: 2 Submissions11,057.59–11,540.85: 1 Submissions11,540.85–12,024.11: 1 Submissions12,024.11–12,507.37: 0 Submissions12,507.37–12,990.62: 1 Submissions12,990.62–13,473.88: 0 Submissions13,473.88–13,957.14: 1 Submissions13,957.14–14,440.4: 0 Submissions14,440.4–14,923.66: 0 Submissions14,923.66–15,406.92: 8 Submissions15,406.92–15,890.18: 16 Submissions15,890.18–16,373.44: 20 Submissions16,373.44–16,856.7: 27 Submissions16,856.7–17,339.96: 11 Submissions17,339.96–17,823.22: 6 Submissions17,823.22–18,306.48: 0 Submissions18,306.48–18,789.74: 2 Submissionsn = 279,124.5518,789.74

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
10,765.92
P25
15,561.42
P50 · Median
16,312.99
P75
16,621.23
P95
17,465.51

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
9,124.55 / 18,789.74
Measurement dates
2026-04-15 – 2026-07-20
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 101

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 2,
        "from": 9124.547691298067,
        "to": 9607.807322656448
      },
      {
        "count": 3,
        "from": 9607.807322656448,
        "to": 10091.066954014828
      },
      {
        "count": 0,
        "from": 10091.066954014828,
        "to": 10574.326585373208
      },
      {
        "count": 2,
        "from": 10574.326585373208,
        "to": 11057.58621673159
      },
      {
        "count": 1,
        "from": 11057.58621673159,
        "to": 11540.845848089972
      },
      {
        "count": 1,
        "from": 11540.845848089972,
        "to": 12024.105479448352
      },
      {
        "count": 0,
        "from": 12024.105479448352,
        "to": 12507.365110806732
      },
      {
        "count": 1,
        "from": 12507.365110806732,
        "to": 12990.624742165113
      },
      {
        "count": 0,
        "from": 12990.624742165113,
        "to": 13473.884373523495
      },
      {
        "count": 1,
        "from": 13473.884373523495,
        "to": 13957.144004881875
      },
      {
        "count": 0,
        "from": 13957.144004881875,
        "to": 14440.403636240255
      },
      {
        "count": 0,
        "from": 14440.403636240255,
        "to": 14923.663267598637
      },
      {
        "count": 8,
        "from": 14923.663267598637,
        "to": 15406.922898957018
      },
      {
        "count": 16,
        "from": 15406.922898957018,
        "to": 15890.182530315398
      },
      {
        "count": 20,
        "from": 15890.182530315398,
        "to": 16373.442161673778
      },
      {
        "count": 27,
        "from": 16373.442161673778,
        "to": 16856.70179303216
      },
      {
        "count": 11,
        "from": 16856.70179303216,
        "to": 17339.96142439054
      },
      {
        "count": 6,
        "from": 17339.96142439054,
        "to": 17823.221055748923
      },
      {
        "count": 0,
        "from": 17823.221055748923,
        "to": 18306.4806871073
      },
      {
        "count": 2,
        "from": 18306.4806871073,
        "to": 18789.740318465683
      }
    ],
    "independent_system_count": null,
    "median": 16312.99010666296,
    "observed_max": 18789.740318465683,
    "observed_min": 9124.547691298067,
    "percentiles": {
      "p25": 15561.41863358362,
      "p5": 10765.922955830834,
      "p50": 16312.99010666296,
      "p75": 16621.23271106766,
      "p95": 17465.514659768003
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 101,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "6602478ad36db145cc9f",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f95d1091-6784-4fd0-9ad9-44cc26bf2875",
    "6f168d57-efcb-4c3b-9f65-e2cc349c4f09",
    "84912c43-1fbd-4f0e-932f-b58e88f7f64b"
  ],
  "measured_from": "2026-04-15T13:45:00.180174+00:00",
  "measured_to": "2026-07-20T11:55:36.157512+00:00",
  "median": 16312.99010666296,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 15561.41863358362,
    "p5": 10765.922955830834,
    "p50": 16312.99010666296,
    "p75": 16621.23271106766,
    "p95": 17465.514659768003
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 101,
  "sample_status": "distribution"
}

Blender complete benchmark

11,547.421308874416 samples/min

Median and quartilesn = 35

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
11,046.6
P50 · Median
11,547.42
P75
11,906.08

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
10,419.18 / 12,689.08
Measurement dates
2026-04-22 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 35

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 11547.421308874416,
    "observed_max": 12689.083635054987,
    "observed_min": 10419.17639272871,
    "percentiles": {
      "p25": 11046.595246198933,
      "p50": 11547.421308874416,
      "p75": 11906.081226192664
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 35,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "bb1c47d416cb4320879b",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "653847f4-19be-45de-b5bf-b9268163d711",
    "0065b816-541b-4b2f-9e81-89f78d039617",
    "978c67fb-d600-4f0f-9296-94aecf86675d"
  ],
  "measured_from": "2026-04-22T20:48:37.845051+00:00",
  "measured_to": "2026-07-11T08:06:47.871900+00:00",
  "median": 11547.421308874416,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 11046.595246198933,
    "p50": 11547.421308874416,
    "p75": 11906.081226192664
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 35,
  "sample_status": "quartiles"
}
Both products have resultsBlender · 5.1.1 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "5.1.1"
}

Blender scene rendering

5,193.347704092749 samples/min

Submitted benchmark distributionn = 101

5.1.1 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

2,790.59–2,951.69: 1 Submissions2,951.69–3,112.79: 1 Submissions3,112.79–3,273.89: 3 Submissions3,273.89–3,434.99: 0 Submissions3,434.99–3,596.09: 2 Submissions3,596.09–3,757.2: 1 Submissions3,757.2–3,918.3: 1 Submissions3,918.3–4,079.4: 1 Submissions4,079.4–4,240.5: 0 Submissions4,240.5–4,401.6: 0 Submissions4,401.6–4,562.71: 1 Submissions4,562.71–4,723.81: 1 Submissions4,723.81–4,884.91: 10 Submissions4,884.91–5,046.01: 12 Submissions5,046.01–5,207.11: 25 Submissions5,207.11–5,368.21: 24 Submissions5,368.21–5,529.32: 9 Submissions5,529.32–5,690.42: 7 Submissions5,690.42–5,851.52: 0 Submissions5,851.52–6,012.62: 2 Submissionsn = 252,790.596,012.62

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
3,549.03
P25
4,929.6
P50 · Median
5,193.35
P75
5,311.39
P95
5,568.35

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,790.59 / 6,012.62
Measurement dates
2026-04-15 – 2026-07-20
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 101

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 1,
        "from": 2790.5851068996535,
        "to": 2951.6869420531075
      },
      {
        "count": 1,
        "from": 2951.6869420531075,
        "to": 3112.788777206562
      },
      {
        "count": 3,
        "from": 3112.788777206562,
        "to": 3273.890612360016
      },
      {
        "count": 0,
        "from": 3273.890612360016,
        "to": 3434.9924475134703
      },
      {
        "count": 2,
        "from": 3434.9924475134703,
        "to": 3596.0942826669243
      },
      {
        "count": 1,
        "from": 3596.0942826669243,
        "to": 3757.1961178203783
      },
      {
        "count": 1,
        "from": 3757.1961178203783,
        "to": 3918.2979529738323
      },
      {
        "count": 1,
        "from": 3918.2979529738323,
        "to": 4079.3997881272867
      },
      {
        "count": 0,
        "from": 4079.3997881272867,
        "to": 4240.501623280741
      },
      {
        "count": 0,
        "from": 4240.501623280741,
        "to": 4401.603458434195
      },
      {
        "count": 1,
        "from": 4401.603458434195,
        "to": 4562.705293587649
      },
      {
        "count": 1,
        "from": 4562.705293587649,
        "to": 4723.807128741103
      },
      {
        "count": 10,
        "from": 4723.807128741103,
        "to": 4884.908963894557
      },
      {
        "count": 12,
        "from": 4884.908963894557,
        "to": 5046.010799048012
      },
      {
        "count": 25,
        "from": 5046.010799048012,
        "to": 5207.112634201466
      },
      {
        "count": 24,
        "from": 5207.112634201466,
        "to": 5368.21446935492
      },
      {
        "count": 9,
        "from": 5368.21446935492,
        "to": 5529.316304508374
      },
      {
        "count": 7,
        "from": 5529.316304508374,
        "to": 5690.418139661828
      },
      {
        "count": 0,
        "from": 5690.418139661828,
        "to": 5851.519974815283
      },
      {
        "count": 2,
        "from": 5851.519974815283,
        "to": 6012.621809968737
      }
    ],
    "independent_system_count": null,
    "median": 5193.347704092749,
    "observed_max": 6012.621809968737,
    "observed_min": 2790.5851068996535,
    "percentiles": {
      "p25": 4929.601435339788,
      "p5": 3549.032215731278,
      "p50": 5193.347704092749,
      "p75": 5311.391947655567,
      "p95": 5568.3496191503145
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 101,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "e6d20fb567dffa268d2c",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f95d1091-6784-4fd0-9ad9-44cc26bf2875",
    "6f168d57-efcb-4c3b-9f65-e2cc349c4f09",
    "84912c43-1fbd-4f0e-932f-b58e88f7f64b"
  ],
  "measured_from": "2026-04-15T13:44:19.931143+00:00",
  "measured_to": "2026-07-20T11:54:57.255967+00:00",
  "median": 5193.347704092749,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 4929.601435339788,
    "p5": 3549.032215731278,
    "p50": 5193.347704092749,
    "p75": 5311.391947655567,
    "p95": 5568.3496191503145
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 101,
  "sample_status": "distribution"
}

Blender scene rendering

3,527.6691022519612 samples/min

Median and quartilesn = 35

5.1.1 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
3,404.49
P50 · Median
3,527.67
P75
3,681.01

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,136.78 / 3,958.46
Measurement dates
2026-04-22 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 35

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3527.6691022519612,
    "observed_max": 3958.462609625652,
    "observed_min": 3136.7803417816754,
    "percentiles": {
      "p25": 3404.4942846563827,
      "p50": 3527.6691022519612,
      "p75": 3681.0114049346275
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 35,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "fb233bc3c2f4bb6a78b9",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "653847f4-19be-45de-b5bf-b9268163d711",
    "0065b816-541b-4b2f-9e81-89f78d039617",
    "978c67fb-d600-4f0f-9296-94aecf86675d"
  ],
  "measured_from": "2026-04-22T20:48:00.131130+00:00",
  "measured_to": "2026-07-11T08:06:07.967139+00:00",
  "median": 3527.6691022519612,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3404.4942846563827,
    "p50": 3527.6691022519612,
    "p75": 3681.0114049346275
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 35,
  "sample_status": "quartiles"
}
Both products have resultsBlender · 5.1.1 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "5.1.1"
}

Blender scene rendering

7,135.885235591367 samples/min

Submitted benchmark distributionn = 101

5.1.1 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

3,969.4–4,180.04: 2 Submissions4,180.04–4,390.68: 3 Submissions4,390.68–4,601.31: 1 Submissions4,601.31–4,811.95: 1 Submissions4,811.95–5,022.58: 2 Submissions5,022.58–5,233.22: 0 Submissions5,233.22–5,443.85: 0 Submissions5,443.85–5,654.49: 0 Submissions5,654.49–5,865.13: 1 Submissions5,865.13–6,075.76: 1 Submissions6,075.76–6,286.4: 0 Submissions6,286.4–6,497.03: 1 Submissions6,497.03–6,707.67: 8 Submissions6,707.67–6,918.3: 11 Submissions6,918.3–7,128.94: 18 Submissions7,128.94–7,339.58: 30 Submissions7,339.58–7,550.21: 11 Submissions7,550.21–7,760.85: 8 Submissions7,760.85–7,971.48: 1 Submissions7,971.48–8,182.12: 2 Submissionsn = 303,969.48,182.12

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
4,591.38
P25
6,825.33
P50 · Median
7,135.89
P75
7,259.48
P95
7,685.49

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,969.4 / 8,182.12
Measurement dates
2026-04-15 – 2026-07-20
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 101

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 2,
        "from": 3969.4045247466247,
        "to": 4180.04026283913
      },
      {
        "count": 3,
        "from": 4180.04026283913,
        "to": 4390.676000931637
      },
      {
        "count": 1,
        "from": 4390.676000931637,
        "to": 4601.311739024142
      },
      {
        "count": 1,
        "from": 4601.311739024142,
        "to": 4811.947477116648
      },
      {
        "count": 2,
        "from": 4811.947477116648,
        "to": 5022.583215209154
      },
      {
        "count": 0,
        "from": 5022.583215209154,
        "to": 5233.21895330166
      },
      {
        "count": 0,
        "from": 5233.21895330166,
        "to": 5443.854691394166
      },
      {
        "count": 0,
        "from": 5443.854691394166,
        "to": 5654.490429486672
      },
      {
        "count": 1,
        "from": 5654.490429486672,
        "to": 5865.1261675791775
      },
      {
        "count": 1,
        "from": 5865.1261675791775,
        "to": 6075.761905671683
      },
      {
        "count": 0,
        "from": 6075.761905671683,
        "to": 6286.3976437641895
      },
      {
        "count": 1,
        "from": 6286.3976437641895,
        "to": 6497.033381856695
      },
      {
        "count": 8,
        "from": 6497.033381856695,
        "to": 6707.669119949202
      },
      {
        "count": 11,
        "from": 6707.669119949202,
        "to": 6918.304858041707
      },
      {
        "count": 18,
        "from": 6918.304858041707,
        "to": 7128.940596134213
      },
      {
        "count": 30,
        "from": 7128.940596134213,
        "to": 7339.576334226718
      },
      {
        "count": 11,
        "from": 7339.576334226718,
        "to": 7550.212072319225
      },
      {
        "count": 8,
        "from": 7550.212072319225,
        "to": 7760.84781041173
      },
      {
        "count": 1,
        "from": 7760.84781041173,
        "to": 7971.483548504237
      },
      {
        "count": 2,
        "from": 7971.483548504237,
        "to": 8182.119286596742
      }
    ],
    "independent_system_count": null,
    "median": 7135.885235591367,
    "observed_max": 8182.119286596742,
    "observed_min": 3969.4045247466247,
    "percentiles": {
      "p25": 6825.329133591204,
      "p5": 4591.381535229979,
      "p50": 7135.885235591367,
      "p75": 7259.483440634439,
      "p95": 7685.48549698363
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 101,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "d9ccd2fca541aa8ba2bf",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "f95d1091-6784-4fd0-9ad9-44cc26bf2875",
    "6f168d57-efcb-4c3b-9f65-e2cc349c4f09",
    "84912c43-1fbd-4f0e-932f-b58e88f7f64b"
  ],
  "measured_from": "2026-04-15T13:43:35.286254+00:00",
  "measured_to": "2026-07-20T11:54:09.345083+00:00",
  "median": 7135.885235591367,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 6825.329133591204,
    "p5": 4591.381535229979,
    "p50": 7135.885235591367,
    "p75": 7259.483440634439,
    "p95": 7685.48549698363
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 101,
  "sample_status": "distribution"
}

Blender scene rendering

5,133.220555612203 samples/min

Median and quartilesn = 35

5.1.1 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
4,942.9
P50 · Median
5,133.22
P75
5,284.26

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,657 / 5,623.5
Measurement dates
2026-04-22 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 35

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5133.220555612203,
    "observed_max": 5623.503597678974,
    "observed_min": 4657.001579876338,
    "percentiles": {
      "p25": 4942.8982814612955,
      "p50": 5133.220555612203,
      "p75": 5284.260823077693
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 35,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "d39ad2debfb7a4e48ff2",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "653847f4-19be-45de-b5bf-b9268163d711",
    "0065b816-541b-4b2f-9e81-89f78d039617",
    "978c67fb-d600-4f0f-9296-94aecf86675d"
  ],
  "measured_from": "2026-04-22T20:47:16.437072+00:00",
  "measured_to": "2026-07-11T08:05:25.791575+00:00",
  "median": 5133.220555612203,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 4942.8982814612955,
    "p50": 5133.220555612203,
    "p75": 5284.260823077693
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 35,
  "sample_status": "quartiles"
}
Both products have resultsBlender · 5.1.1 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 24,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "5.1.1"
}

Blender scene rendering

3,933.753341122553 samples/min

Provisional median · Limited submissionsn = 11

5.1.1 · classroom · OPTIX · Windows · 1P · 24 cores · 32 threads

3,933.75 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,457.4 / 4,203.67
Measurement dates
2026-04-16 – 2026-07-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 11

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3933.753341122553,
    "observed_max": 4203.670378099665,
    "observed_min": 3457.3987127279934,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 11,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "e831a476b128d1ffdb99",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "42ffbf56-fa5c-4fae-a587-249954177305",
    "27802395-99a5-41f3-8d10-abbe445939a6",
    "5b381b61-1e7b-4beb-9c2b-1807dcea7b1b"
  ],
  "measured_from": "2026-04-16T14:23:25.258640+00:00",
  "measured_to": "2026-07-05T22:05:19.604480+00:00",
  "median": 3933.753341122553,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 11,
  "sample_status": "provisional"
}

Blender scene rendering

2,909.501335440469 samples/min

Provisional median · Limited submissionsn = 22

5.1.1 · classroom · OPTIX · Windows · 1P · 24 cores · 32 threads

2,909.5 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
1,902.81 / 3,062
Measurement dates
2026-04-15 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2909.501335440469,
    "observed_max": 3062.0039954032286,
    "observed_min": 1902.808031050362,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "24403ee9e682f2797aaa",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "123b3331-7e31-44f8-993b-6a2b4a2e1a59",
    "7bbc4fc3-b7ce-4dfc-8258-c9bdf4cfd54e",
    "fc73677d-6df6-4e7b-8f0b-48d90a71dfdf"
  ],
  "measured_from": "2026-04-15T13:12:25.034592+00:00",
  "measured_to": "2026-07-11T15:14:39.458810+00:00",
  "median": 2909.501335440469,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 22,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 24,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "5.1.1"
}

Blender complete benchmark

16,065.179184618442 samples/min

Provisional median · Limited submissionsn = 11

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 24 cores · 32 threads

16,065.18 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
14,068.02 / 17,197.23
Measurement dates
2026-04-16 – 2026-07-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 11

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 16065.179184618442,
    "observed_max": 17197.232800952363,
    "observed_min": 14068.016575251535,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 11,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "7af1dea04652a2d55552",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "42ffbf56-fa5c-4fae-a587-249954177305",
    "27802395-99a5-41f3-8d10-abbe445939a6",
    "5b381b61-1e7b-4beb-9c2b-1807dcea7b1b"
  ],
  "measured_from": "2026-04-16T14:23:25.258640+00:00",
  "measured_to": "2026-07-05T22:05:19.604480+00:00",
  "median": 16065.179184618442,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 11,
  "sample_status": "provisional"
}

Blender complete benchmark

11,729.595238256596 samples/min

Provisional median · Limited submissionsn = 22

5.1.1 · monster + junkshop + classroom · OPTIX · Windows · 1P · 24 cores · 32 threads

11,729.6 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
7,625.37 / 12,329.75
Measurement dates
2026-04-15 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 11729.595238256596,
    "observed_max": 12329.754201089476,
    "observed_min": 7625.372563518345,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "0c3f758acf0c90de0542",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "123b3331-7e31-44f8-993b-6a2b4a2e1a59",
    "7bbc4fc3-b7ce-4dfc-8258-c9bdf4cfd54e",
    "fc73677d-6df6-4e7b-8f0b-48d90a71dfdf"
  ],
  "measured_from": "2026-04-15T13:12:25.034592+00:00",
  "measured_to": "2026-07-11T15:14:39.458810+00:00",
  "median": 11729.595238256596,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 22,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 24,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "5.1.1"
}

Blender scene rendering

5,059.052569765606 samples/min

Provisional median · Limited submissionsn = 11

5.1.1 · junkshop · OPTIX · Windows · 1P · 24 cores · 32 threads

5,059.05 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,448.92 / 5,545.84
Measurement dates
2026-04-16 – 2026-07-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 11

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5059.052569765606,
    "observed_max": 5545.837589914577,
    "observed_min": 4448.919421574462,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 11,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "583177253833fd0d148c",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "42ffbf56-fa5c-4fae-a587-249954177305",
    "27802395-99a5-41f3-8d10-abbe445939a6",
    "5b381b61-1e7b-4beb-9c2b-1807dcea7b1b"
  ],
  "measured_from": "2026-04-16T14:22:46.915212+00:00",
  "measured_to": "2026-07-05T22:04:43.825686+00:00",
  "median": 5059.052569765606,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 11,
  "sample_status": "provisional"
}

Blender scene rendering

3,646.3823998290673 samples/min

Provisional median · Limited submissionsn = 22

5.1.1 · junkshop · OPTIX · Windows · 1P · 24 cores · 32 threads

3,646.38 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,506.78 / 3,795.82
Measurement dates
2026-04-15 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3646.3823998290673,
    "observed_max": 3795.821952917358,
    "observed_min": 2506.7812902141727,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "f06f45a87cbc2a20bdfb",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "123b3331-7e31-44f8-993b-6a2b4a2e1a59",
    "7bbc4fc3-b7ce-4dfc-8258-c9bdf4cfd54e",
    "fc73677d-6df6-4e7b-8f0b-48d90a71dfdf"
  ],
  "measured_from": "2026-04-15T13:11:45.128123+00:00",
  "measured_to": "2026-07-11T15:14:01.425761+00:00",
  "median": 3646.3823998290673,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 22,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.1.1 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 24,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "5.1.1"
}

Blender scene rendering

7,182.8868170389815 samples/min

Provisional median · Limited submissionsn = 11

5.1.1 · monster · OPTIX · Windows · 1P · 24 cores · 32 threads

7,182.89 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
6,161.7 / 7,525.3
Measurement dates
2026-04-16 – 2026-07-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 11

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 7182.8868170389815,
    "observed_max": 7525.29882705476,
    "observed_min": 6161.69844094908,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 11,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "f1db5811f98f74eb72a8",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "42ffbf56-fa5c-4fae-a587-249954177305",
    "27802395-99a5-41f3-8d10-abbe445939a6",
    "5b381b61-1e7b-4beb-9c2b-1807dcea7b1b"
  ],
  "measured_from": "2026-04-16T14:22:03.576559+00:00",
  "measured_to": "2026-07-05T22:04:01.019733+00:00",
  "median": 7182.8868170389815,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 11,
  "sample_status": "provisional"
}

Blender scene rendering

5,202.592819670859 samples/min

Provisional median · Limited submissionsn = 22

5.1.1 · monster · OPTIX · Windows · 1P · 24 cores · 32 threads

5,202.59 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,215.78 / 5,488.57
Measurement dates
2026-04-15 – 2026-07-11
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5202.592819670859,
    "observed_max": 5488.565533124183,
    "observed_min": 3215.78324225381,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 24,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "d274c2e972658ee73458",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "123b3331-7e31-44f8-993b-6a2b4a2e1a59",
    "7bbc4fc3-b7ce-4dfc-8258-c9bdf4cfd54e",
    "fc73677d-6df6-4e7b-8f0b-48d90a71dfdf"
  ],
  "measured_from": "2026-04-15T13:10:58.766492+00:00",
  "measured_to": "2026-07-11T15:13:21.101902+00:00",
  "median": 5202.592819670859,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 22,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.0.0 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "5.0.0"
}

Blender scene rendering

3,324.1654706222985 samples/min

Provisional median · Limited submissionsn = 22

5.0.0 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

3,324.17 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,788.67 / 4,101.76
Measurement dates
2026-03-11 – 2026-05-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3324.1654706222985,
    "observed_max": 4101.763324548365,
    "observed_min": 2788.673353515482,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.0.0"
  },
  "condition_id": "2946e9a33b4a38f9697e",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "cf57eefb-e672-4dcd-b82c-6adaccce4d38",
    "71ff1c90-1651-4be3-a312-41196ff54277",
    "cd3212e1-6c29-484a-b911-3539f966866a"
  ],
  "measured_from": "2026-03-11T14:31:33.670017+00:00",
  "measured_to": "2026-05-05T20:20:01.335528+00:00",
  "median": 3324.1654706222985,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 22,
  "sample_status": "provisional"
}

Blender scene rendering

2,532.57652592184 samples/min

Provisional median · Limited submissionsn = 13

5.0.0 · classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

2,532.58 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,385.75 / 3,021.78
Measurement dates
2026-03-11 – 2026-03-19
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 13

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2532.57652592184,
    "observed_max": 3021.7838767875,
    "observed_min": 2385.7530449860333,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 13,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.0.0"
  },
  "condition_id": "b65a72d4cc6e15602746",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "8acd4949-a722-48e4-a15e-b1fa534563c1",
    "b865628c-2a74-4b37-b0a8-347fcac781b1",
    "1ff33fd0-acb0-416e-9e88-0f95be6207fd"
  ],
  "measured_from": "2026-03-11T17:04:31.327018+00:00",
  "measured_to": "2026-03-19T13:38:27.850085+00:00",
  "median": 2532.57652592184,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 13,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.0.0 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "5.0.0"
}

Blender complete benchmark

14,366.707101456755 samples/min

Provisional median · Limited submissionsn = 22

5.0.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

14,366.71 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
12,058.63 / 16,115.54
Measurement dates
2026-03-11 – 2026-05-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 14366.707101456755,
    "observed_max": 16115.543416378812,
    "observed_min": 12058.632954052144,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.0.0"
  },
  "condition_id": "0421448890d89feab054",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "cf57eefb-e672-4dcd-b82c-6adaccce4d38",
    "71ff1c90-1651-4be3-a312-41196ff54277",
    "cd3212e1-6c29-484a-b911-3539f966866a"
  ],
  "measured_from": "2026-03-11T14:31:33.670017+00:00",
  "measured_to": "2026-05-05T20:20:01.335528+00:00",
  "median": 14366.707101456755,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 22,
  "sample_status": "provisional"
}

Blender complete benchmark

10,596.064888742825 samples/min

Provisional median · Limited submissionsn = 13

5.0.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 32 threads

10,596.06 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
9,924.61 / 11,715.62
Measurement dates
2026-03-11 – 2026-03-19
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 13

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 10596.064888742825,
    "observed_max": 11715.616795662812,
    "observed_min": 9924.611347860622,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 13,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.0.0"
  },
  "condition_id": "2cf5688c968474ea6b6b",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "8acd4949-a722-48e4-a15e-b1fa534563c1",
    "b865628c-2a74-4b37-b0a8-347fcac781b1",
    "1ff33fd0-acb0-416e-9e88-0f95be6207fd"
  ],
  "measured_from": "2026-03-11T17:04:31.327018+00:00",
  "measured_to": "2026-03-19T13:38:27.850085+00:00",
  "median": 10596.064888742825,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 13,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.0.0 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "5.0.0"
}

Blender scene rendering

4,309.127731741826 samples/min

Provisional median · Limited submissionsn = 22

5.0.0 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

4,309.13 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
3,643.27 / 4,876.39
Measurement dates
2026-03-11 – 2026-05-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 4309.127731741826,
    "observed_max": 4876.390823397917,
    "observed_min": 3643.2660880093044,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.0.0"
  },
  "condition_id": "ad0a8897c6a1d996bf09",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "cf57eefb-e672-4dcd-b82c-6adaccce4d38",
    "71ff1c90-1651-4be3-a312-41196ff54277",
    "cd3212e1-6c29-484a-b911-3539f966866a"
  ],
  "measured_from": "2026-03-11T14:30:57.677878+00:00",
  "measured_to": "2026-05-05T20:19:22.146002+00:00",
  "median": 4309.127731741826,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 22,
  "sample_status": "provisional"
}

Blender scene rendering

3,066.9412302957257 samples/min

Provisional median · Limited submissionsn = 13

5.0.0 · junkshop · OPTIX · Windows · 1P · 16 cores · 32 threads

3,066.94 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,870.14 / 3,373.08
Measurement dates
2026-03-11 – 2026-03-19
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 13

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 3066.9412302957257,
    "observed_max": 3373.0768624680272,
    "observed_min": 2870.1361431744167,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 13,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.0.0"
  },
  "condition_id": "aa134bfcdec18318fc22",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "8acd4949-a722-48e4-a15e-b1fa534563c1",
    "b865628c-2a74-4b37-b0a8-347fcac781b1",
    "1ff33fd0-acb0-416e-9e88-0f95be6207fd"
  ],
  "measured_from": "2026-03-11T17:03:53.465686+00:00",
  "measured_to": "2026-03-19T13:37:47.181246+00:00",
  "median": 3066.9412302957257,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 13,
  "sample_status": "provisional"
}
Both products have resultsBlender · 5.0.0 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.2",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 32,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "5.0.0"
}

Blender scene rendering

6,709.031540608203 samples/min

Provisional median · Limited submissionsn = 22

5.0.0 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

6,709.03 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
5,626.69 / 7,137.39
Measurement dates
2026-03-11 – 2026-05-05
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 22

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 6709.031540608203,
    "observed_max": 7137.389268432531,
    "observed_min": 5626.693512527358,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 22,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.0.0"
  },
  "condition_id": "bcfd95e08f2962248a42",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "cf57eefb-e672-4dcd-b82c-6adaccce4d38",
    "71ff1c90-1651-4be3-a312-41196ff54277",
    "cd3212e1-6c29-484a-b911-3539f966866a"
  ],
  "measured_from": "2026-03-11T14:30:15.872350+00:00",
  "measured_to": "2026-05-05T20:18:39.093488+00:00",
  "median": 6709.031540608203,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 22,
  "sample_status": "provisional"
}

Blender scene rendering

5,035.0617702728 samples/min

Provisional median · Limited submissionsn = 13

5.0.0 · monster · OPTIX · Windows · 1P · 16 cores · 32 threads

5,035.06 Provisional median

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
4,668.72 / 5,321.41
Measurement dates
2026-03-11 – 2026-03-19
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 13

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5035.0617702728,
    "observed_max": 5321.408852084835,
    "observed_min": 4668.722159700172,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 13,
    "sample_status": "provisional",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.2",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 32,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.0.0"
  },
  "condition_id": "91cb1928edbaa5993df8",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "8acd4949-a722-48e4-a15e-b1fa534563c1",
    "b865628c-2a74-4b37-b0a8-347fcac781b1",
    "1ff33fd0-acb0-416e-9e88-0f95be6207fd"
  ],
  "measured_from": "2026-03-11T17:03:08.617813+00:00",
  "measured_to": "2026-03-19T13:36:58.201161+00:00",
  "median": 5035.0617702728,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 13,
  "sample_status": "provisional"
}
One-sided evidenceBlender · 4.5.0 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 5090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "4.5.0"
}

Blender scene rendering

3,715.7736556764166 samples/min

Submitted benchmark distributionn = 216

4.5.0 · classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

2,711.78–2,787.74: 1 Submissions2,787.74–2,863.7: 0 Submissions2,863.7–2,939.66: 0 Submissions2,939.66–3,015.62: 1 Submissions3,015.62–3,091.58: 3 Submissions3,091.58–3,167.53: 5 Submissions3,167.53–3,243.49: 5 Submissions3,243.49–3,319.45: 2 Submissions3,319.45–3,395.41: 5 Submissions3,395.41–3,471.37: 11 Submissions3,471.37–3,547.33: 7 Submissions3,547.33–3,623.29: 24 Submissions3,623.29–3,699.24: 37 Submissions3,699.24–3,775.2: 64 Submissions3,775.2–3,851.16: 38 Submissions3,851.16–3,927.12: 3 Submissions3,927.12–4,003.08: 2 Submissions4,003.08–4,079.04: 4 Submissions4,079.04–4,155: 2 Submissions4,155–4,230.95: 2 Submissionsn = 642,711.784,230.95

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
3,171.51
P25
3,597.25
P50 · Median
3,715.77
P75
3,766.46
P95
3,920.85

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,711.78 / 4,230.95
Measurement dates
2025-07-17 – 2026-03-10
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 216

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 1,
        "from": 2711.7821110420964,
        "to": 2787.740712183224
      },
      {
        "count": 0,
        "from": 2787.740712183224,
        "to": 2863.699313324351
      },
      {
        "count": 0,
        "from": 2863.699313324351,
        "to": 2939.6579144654784
      },
      {
        "count": 1,
        "from": 2939.6579144654784,
        "to": 3015.616515606606
      },
      {
        "count": 3,
        "from": 3015.616515606606,
        "to": 3091.575116747733
      },
      {
        "count": 5,
        "from": 3091.575116747733,
        "to": 3167.5337178888603
      },
      {
        "count": 5,
        "from": 3167.5337178888603,
        "to": 3243.492319029988
      },
      {
        "count": 2,
        "from": 3243.492319029988,
        "to": 3319.4509201711153
      },
      {
        "count": 5,
        "from": 3319.4509201711153,
        "to": 3395.4095213122423
      },
      {
        "count": 11,
        "from": 3395.4095213122423,
        "to": 3471.36812245337
      },
      {
        "count": 7,
        "from": 3471.36812245337,
        "to": 3547.326723594497
      },
      {
        "count": 24,
        "from": 3547.326723594497,
        "to": 3623.2853247356243
      },
      {
        "count": 37,
        "from": 3623.2853247356243,
        "to": 3699.2439258767517
      },
      {
        "count": 64,
        "from": 3699.2439258767517,
        "to": 3775.202527017879
      },
      {
        "count": 38,
        "from": 3775.202527017879,
        "to": 3851.1611281590067
      },
      {
        "count": 3,
        "from": 3851.1611281590067,
        "to": 3927.1197293001337
      },
      {
        "count": 2,
        "from": 3927.1197293001337,
        "to": 4003.0783304412607
      },
      {
        "count": 4,
        "from": 4003.0783304412607,
        "to": 4079.036931582388
      },
      {
        "count": 2,
        "from": 4079.036931582388,
        "to": 4154.995532723516
      },
      {
        "count": 2,
        "from": 4154.995532723516,
        "to": 4230.954133864643
      }
    ],
    "independent_system_count": null,
    "median": 3715.7736556764166,
    "observed_max": 4230.954133864643,
    "observed_min": 2711.7821110420964,
    "percentiles": {
      "p25": 3597.252197949314,
      "p5": 3171.5143214763225,
      "p50": 3715.7736556764166,
      "p75": 3766.4566504741097,
      "p95": 3920.8464197159124
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 216,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "4.5.0"
  },
  "condition_id": "de55e0f8c0e2859b962b",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "6ca31088-6948-4bfe-ac9b-5885414c36ec",
    "69f02af8-3106-46a9-b4f8-085291b7619e",
    "3ed9f735-885c-448c-925d-64b3672d3cba"
  ],
  "measured_from": "2025-07-17T12:51:53.950224+00:00",
  "measured_to": "2026-03-10T17:00:02.771486+00:00",
  "median": 3715.7736556764166,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3597.252197949314,
    "p5": 3171.5143214763225,
    "p50": 3715.7736556764166,
    "p75": 3766.4566504741097,
    "p95": 3920.8464197159124
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 216,
  "sample_status": "distribution"
}
No comparable data
One-sided evidenceBlender · 4.5.0 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 5090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "4.5.0"
}

Blender complete benchmark

14,953.908481170562 samples/min

Submitted benchmark distributionn = 216

4.5.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 8 cores · 16 threads

10,732.04–11,045.25: 1 Submissions11,045.25–11,358.46: 0 Submissions11,358.46–11,671.68: 0 Submissions11,671.68–11,984.89: 0 Submissions11,984.89–12,298.1: 3 Submissions12,298.1–12,611.31: 1 Submissions12,611.31–12,924.52: 4 Submissions12,924.52–13,237.73: 5 Submissions13,237.73–13,550.94: 7 Submissions13,550.94–13,864.15: 2 Submissions13,864.15–14,177.36: 14 Submissions14,177.36–14,490.57: 15 Submissions14,490.57–14,803.78: 31 Submissions14,803.78–15,116.99: 75 Submissions15,116.99–15,430.2: 39 Submissions15,430.2–15,743.42: 9 Submissions15,743.42–16,056.63: 2 Submissions16,056.63–16,369.84: 3 Submissions16,369.84–16,683.05: 2 Submissions16,683.05–16,996.26: 3 Submissionsn = 7510,732.0416,996.26

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
13,004.49
P25
14,545.77
P50 · Median
14,953.91
P75
15,126.82
P95
15,707.87

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
10,732.04 / 16,996.26
Measurement dates
2025-07-17 – 2026-03-10
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 216

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 1,
        "from": 10732.043022443115,
        "to": 11045.253788770882
      },
      {
        "count": 0,
        "from": 11045.253788770882,
        "to": 11358.464555098652
      },
      {
        "count": 0,
        "from": 11358.464555098652,
        "to": 11671.675321426419
      },
      {
        "count": 0,
        "from": 11671.675321426419,
        "to": 11984.886087754187
      },
      {
        "count": 3,
        "from": 11984.886087754187,
        "to": 12298.096854081956
      },
      {
        "count": 1,
        "from": 12298.096854081956,
        "to": 12611.307620409723
      },
      {
        "count": 4,
        "from": 12611.307620409723,
        "to": 12924.51838673749
      },
      {
        "count": 5,
        "from": 12924.51838673749,
        "to": 13237.729153065258
      },
      {
        "count": 7,
        "from": 13237.729153065258,
        "to": 13550.939919393028
      },
      {
        "count": 2,
        "from": 13550.939919393028,
        "to": 13864.150685720795
      },
      {
        "count": 14,
        "from": 13864.150685720795,
        "to": 14177.361452048564
      },
      {
        "count": 15,
        "from": 14177.361452048564,
        "to": 14490.572218376332
      },
      {
        "count": 31,
        "from": 14490.572218376332,
        "to": 14803.7829847041
      },
      {
        "count": 75,
        "from": 14803.7829847041,
        "to": 15116.993751031867
      },
      {
        "count": 39,
        "from": 15116.993751031867,
        "to": 15430.204517359634
      },
      {
        "count": 9,
        "from": 15430.204517359634,
        "to": 15743.415283687404
      },
      {
        "count": 2,
        "from": 15743.415283687404,
        "to": 16056.626050015173
      },
      {
        "count": 3,
        "from": 16056.626050015173,
        "to": 16369.83681634294
      },
      {
        "count": 2,
        "from": 16369.83681634294,
        "to": 16683.047582670708
      },
      {
        "count": 3,
        "from": 16683.047582670708,
        "to": 16996.258348998475
      }
    ],
    "independent_system_count": null,
    "median": 14953.908481170562,
    "observed_max": 16996.258348998475,
    "observed_min": 10732.043022443115,
    "percentiles": {
      "p25": 14545.771067775535,
      "p5": 13004.48595773311,
      "p50": 14953.908481170562,
      "p75": 15126.822498048714,
      "p95": 15707.870211418434
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 216,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "4.5.0"
  },
  "condition_id": "fe44df2b0bb29882a1f5",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "6ca31088-6948-4bfe-ac9b-5885414c36ec",
    "69f02af8-3106-46a9-b4f8-085291b7619e",
    "3ed9f735-885c-448c-925d-64b3672d3cba"
  ],
  "measured_from": "2025-07-17T12:51:53.950224+00:00",
  "measured_to": "2026-03-10T17:00:02.771486+00:00",
  "median": 14953.908481170562,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 14545.771067775535,
    "p5": 13004.48595773311,
    "p50": 14953.908481170562,
    "p75": 15126.822498048714,
    "p95": 15707.870211418434
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 216,
  "sample_status": "distribution"
}
No comparable data
One-sided evidenceBlender · 4.5.0 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 5090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "4.5.0"
}

Blender scene rendering

3,933.8884699257205 samples/min

Submitted benchmark distributionn = 216

4.5.0 · junkshop · OPTIX · Windows · 1P · 8 cores · 16 threads

2,967.14–3,045.18: 1 Submissions3,045.18–3,123.21: 0 Submissions3,123.21–3,201.25: 1 Submissions3,201.25–3,279.29: 2 Submissions3,279.29–3,357.32: 4 Submissions3,357.32–3,435.36: 2 Submissions3,435.36–3,513.39: 3 Submissions3,513.39–3,591.43: 8 Submissions3,591.43–3,669.46: 5 Submissions3,669.46–3,747.5: 12 Submissions3,747.5–3,825.54: 16 Submissions3,825.54–3,903.57: 23 Submissions3,903.57–3,981.61: 89 Submissions3,981.61–4,059.64: 26 Submissions4,059.64–4,137.68: 11 Submissions4,137.68–4,215.71: 4 Submissions4,215.71–4,293.75: 1 Submissions4,293.75–4,371.79: 3 Submissions4,371.79–4,449.82: 1 Submissions4,449.82–4,527.86: 4 Submissionsn = 892,967.144,527.86

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
3,487.17
P25
3,825.81
P50 · Median
3,933.89
P75
3,978.38
P95
4,185.39

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,967.14 / 4,527.86
Measurement dates
2025-07-17 – 2026-03-10
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 216

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 1,
        "from": 2967.1429936091886,
        "to": 3045.1787076066016
      },
      {
        "count": 0,
        "from": 3045.1787076066016,
        "to": 3123.214421604015
      },
      {
        "count": 1,
        "from": 3123.214421604015,
        "to": 3201.250135601428
      },
      {
        "count": 2,
        "from": 3201.250135601428,
        "to": 3279.2858495988417
      },
      {
        "count": 4,
        "from": 3279.2858495988417,
        "to": 3357.3215635962547
      },
      {
        "count": 2,
        "from": 3357.3215635962547,
        "to": 3435.357277593668
      },
      {
        "count": 3,
        "from": 3435.357277593668,
        "to": 3513.392991591081
      },
      {
        "count": 8,
        "from": 3513.392991591081,
        "to": 3591.4287055884943
      },
      {
        "count": 5,
        "from": 3591.4287055884943,
        "to": 3669.4644195859078
      },
      {
        "count": 12,
        "from": 3669.4644195859078,
        "to": 3747.5001335833213
      },
      {
        "count": 16,
        "from": 3747.5001335833213,
        "to": 3825.5358475807343
      },
      {
        "count": 23,
        "from": 3825.5358475807343,
        "to": 3903.5715615781473
      },
      {
        "count": 89,
        "from": 3903.5715615781473,
        "to": 3981.607275575561
      },
      {
        "count": 26,
        "from": 3981.607275575561,
        "to": 4059.642989572974
      },
      {
        "count": 11,
        "from": 4059.642989572974,
        "to": 4137.678703570387
      },
      {
        "count": 4,
        "from": 4137.678703570387,
        "to": 4215.7144175678
      },
      {
        "count": 1,
        "from": 4215.7144175678,
        "to": 4293.750131565213
      },
      {
        "count": 3,
        "from": 4293.750131565213,
        "to": 4371.7858455626265
      },
      {
        "count": 1,
        "from": 4371.7858455626265,
        "to": 4449.82155956004
      },
      {
        "count": 4,
        "from": 4449.82155956004,
        "to": 4527.8572735574535
      }
    ],
    "independent_system_count": null,
    "median": 3933.8884699257205,
    "observed_max": 4527.8572735574535,
    "observed_min": 2967.1429936091886,
    "percentiles": {
      "p25": 3825.8142533884206,
      "p5": 3487.1708353863487,
      "p50": 3933.8884699257205,
      "p75": 3978.3782914840585,
      "p95": 4185.393793445304
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 216,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "4.5.0"
  },
  "condition_id": "7d5694bf0148efc4218d",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "6ca31088-6948-4bfe-ac9b-5885414c36ec",
    "69f02af8-3106-46a9-b4f8-085291b7619e",
    "3ed9f735-885c-448c-925d-64b3672d3cba"
  ],
  "measured_from": "2025-07-17T12:51:15.450186+00:00",
  "measured_to": "2026-03-10T16:59:24.396096+00:00",
  "median": 3933.8884699257205,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 3825.8142533884206,
    "p5": 3487.1708353863487,
    "p50": 3933.8884699257205,
    "p75": 3978.3782914840585,
    "p95": 4185.393793445304
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 216,
  "sample_status": "distribution"
}
No comparable data
One-sided evidenceBlender · 4.5.0 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 5090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 8,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 16,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "4.5.0"
}

Blender scene rendering

7,287.069613286671 samples/min

Submitted benchmark distributionn = 216

4.5.0 · monster · OPTIX · Windows · 1P · 8 cores · 16 threads

5,053.12–5,212.33: 1 Submissions5,212.33–5,371.55: 0 Submissions5,371.55–5,530.77: 0 Submissions5,530.77–5,689.98: 0 Submissions5,689.98–5,849.2: 1 Submissions5,849.2–6,008.42: 3 Submissions6,008.42–6,167.63: 0 Submissions6,167.63–6,326.85: 4 Submissions6,326.85–6,486.07: 4 Submissions6,486.07–6,645.28: 8 Submissions6,645.28–6,804.5: 4 Submissions6,804.5–6,963.72: 11 Submissions6,963.72–7,122.93: 20 Submissions7,122.93–7,282.15: 48 Submissions7,282.15–7,441.36: 70 Submissions7,441.36–7,600.58: 27 Submissions7,600.58–7,759.8: 6 Submissions7,759.8–7,919.01: 2 Submissions7,919.01–8,078.23: 3 Submissions8,078.23–8,237.45: 4 Submissionsn = 705,053.128,237.45

Box: middle 50% (P25–P75)Vertical line: median (P50)Horizontal line: P5–P95

P5
6,358.98
P25
7,110.84
P50 · Median
7,287.07
P75
7,400.44
P95
7,653.06

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
5,053.12 / 8,237.45
Measurement dates
2025-07-17 – 2026-03-10
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 216

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "histogram": [
      {
        "count": 1,
        "from": 5053.117917791829,
        "to": 5212.334368981057
      },
      {
        "count": 0,
        "from": 5212.334368981057,
        "to": 5371.550820170284
      },
      {
        "count": 0,
        "from": 5371.550820170284,
        "to": 5530.767271359511
      },
      {
        "count": 0,
        "from": 5530.767271359511,
        "to": 5689.983722548739
      },
      {
        "count": 1,
        "from": 5689.983722548739,
        "to": 5849.200173737967
      },
      {
        "count": 3,
        "from": 5849.200173737967,
        "to": 6008.416624927194
      },
      {
        "count": 0,
        "from": 6008.416624927194,
        "to": 6167.633076116421
      },
      {
        "count": 4,
        "from": 6167.633076116421,
        "to": 6326.849527305649
      },
      {
        "count": 4,
        "from": 6326.849527305649,
        "to": 6486.065978494877
      },
      {
        "count": 8,
        "from": 6486.065978494877,
        "to": 6645.282429684104
      },
      {
        "count": 4,
        "from": 6645.282429684104,
        "to": 6804.498880873331
      },
      {
        "count": 11,
        "from": 6804.498880873331,
        "to": 6963.715332062558
      },
      {
        "count": 20,
        "from": 6963.715332062558,
        "to": 7122.931783251786
      },
      {
        "count": 48,
        "from": 7122.931783251786,
        "to": 7282.148234441013
      },
      {
        "count": 70,
        "from": 7282.148234441013,
        "to": 7441.36468563024
      },
      {
        "count": 27,
        "from": 7441.36468563024,
        "to": 7600.581136819468
      },
      {
        "count": 6,
        "from": 7600.581136819468,
        "to": 7759.797588008696
      },
      {
        "count": 2,
        "from": 7759.797588008696,
        "to": 7919.014039197923
      },
      {
        "count": 3,
        "from": 7919.014039197923,
        "to": 8078.23049038715
      },
      {
        "count": 4,
        "from": 8078.23049038715,
        "to": 8237.446941576378
      }
    ],
    "independent_system_count": null,
    "median": 7287.069613286671,
    "observed_max": 8237.446941576378,
    "observed_min": 5053.117917791829,
    "percentiles": {
      "p25": 7110.838509008761,
      "p5": 6358.980862937951,
      "p50": 7287.069613286671,
      "p75": 7400.444535035924,
      "p95": 7653.061950124811
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 216,
    "sample_status": "distribution",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 5090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 8,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 16,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "4.5.0"
  },
  "condition_id": "b0c4d1d631def2df8e6e",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "6ca31088-6948-4bfe-ac9b-5885414c36ec",
    "69f02af8-3106-46a9-b4f8-085291b7619e",
    "3ed9f735-885c-448c-925d-64b3672d3cba"
  ],
  "measured_from": "2025-07-17T12:50:30.403736+00:00",
  "measured_to": "2026-03-10T16:58:39.046964+00:00",
  "median": 7287.069613286671,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 7110.838509008761,
    "p5": 6358.980862937951,
    "p50": 7287.069613286671,
    "p75": 7400.444535035924,
    "p95": 7653.061950124811
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-5090",
  "run_count": 216,
  "sample_status": "distribution"
}
No comparable data
One-sided evidenceBlender · 4.5.0 · classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 24,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "classroom",
  "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
  "version": "4.5.0"
}
No comparable data

Blender scene rendering

2,852.399815642552 samples/min

Median and quartilesn = 33

4.5.0 · classroom · OPTIX · Windows · 1P · 16 cores · 24 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
2,777.99
P50 · Median
2,852.4
P75
2,895.81

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,532.05 / 2,998.49
Measurement dates
2025-07-25 – 2026-03-08
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 33

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2852.399815642552,
    "observed_max": 2998.4926575459644,
    "observed_min": 2532.053487801554,
    "percentiles": {
      "p25": 2777.986215440814,
      "p50": 2852.399815642552,
      "p75": 2895.8143644199313
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 33,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 24,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "4.5.0"
  },
  "condition_id": "b89faf128c2f780b6fd3",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "4b4267b4-9d00-4077-85fe-2aabc82d78c4",
    "5af570aa-7dd2-41f5-a568-07a219c1bd0a",
    "980bde14-9645-45ac-ade4-6017bbf95e94"
  ],
  "measured_from": "2025-07-25T14:06:49.080736+00:00",
  "measured_to": "2026-03-08T20:31:15.306844+00:00",
  "median": 2852.399815642552,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 2777.986215440814,
    "p50": 2852.399815642552,
    "p75": 2895.8143644199313
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 33,
  "sample_status": "quartiles"
}
One-sided evidenceBlender · 4.5.0 · monster + junkshop + classroom · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 24,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster + junkshop + classroom",
  "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
  "version": "4.5.0"
}
No comparable data

Blender complete benchmark

11,305.605054123425 samples/min

Median and quartilesn = 33

4.5.0 · monster + junkshop + classroom · OPTIX · Windows · 1P · 16 cores · 24 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
11,014.97
P50 · Median
11,305.61
P75
11,454.87

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
10,197.74 / 12,089.52
Measurement dates
2025-07-25 – 2026-03-08
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 33

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 11305.605054123425,
    "observed_max": 12089.524914889054,
    "observed_min": 10197.740346325285,
    "percentiles": {
      "p25": 11014.971555796257,
      "p50": 11305.605054123425,
      "p75": 11454.865719579007
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 33,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 24,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "4.5.0"
  },
  "condition_id": "01245e0ba9c067ef6e62",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "4b4267b4-9d00-4077-85fe-2aabc82d78c4",
    "5af570aa-7dd2-41f5-a568-07a219c1bd0a",
    "980bde14-9645-45ac-ade4-6017bbf95e94"
  ],
  "measured_from": "2025-07-25T14:06:49.080736+00:00",
  "measured_to": "2026-03-08T20:31:15.306844+00:00",
  "median": 11305.605054123425,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "percentiles": {
    "p25": 11014.971555796257,
    "p50": 11305.605054123425,
    "p75": 11454.865719579007
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 33,
  "sample_status": "quartiles"
}
One-sided evidenceBlender · 4.5.0 · junkshop · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 24,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "junkshop",
  "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
  "version": "4.5.0"
}
No comparable data

Blender scene rendering

2,803.046244396757 samples/min

Median and quartilesn = 33

4.5.0 · junkshop · OPTIX · Windows · 1P · 16 cores · 24 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
2,756.88
P50 · Median
2,803.05
P75
2,836.17

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
2,486.31 / 3,062.98
Measurement dates
2025-07-25 – 2026-03-08
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 33

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 2803.046244396757,
    "observed_max": 3062.9812313184552,
    "observed_min": 2486.313629540434,
    "percentiles": {
      "p25": 2756.882430533558,
      "p50": 2803.046244396757,
      "p75": 2836.168461271585
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 33,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 24,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "4.5.0"
  },
  "condition_id": "d9be52c9f99d0056c924",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "4b4267b4-9d00-4077-85fe-2aabc82d78c4",
    "5af570aa-7dd2-41f5-a568-07a219c1bd0a",
    "980bde14-9645-45ac-ade4-6017bbf95e94"
  ],
  "measured_from": "2025-07-25T14:06:07.646109+00:00",
  "measured_to": "2026-03-08T20:30:34.524011+00:00",
  "median": 2803.046244396757,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 2756.882430533558,
    "p50": 2803.046244396757,
    "p75": 2836.168461271585
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 33,
  "sample_status": "quartiles"
}
One-sided evidenceBlender · 4.5.0 · monster · OPTIX
Full test conditions
{
  "benchmark": "Blender",
  "environment": {
    "backend": "OPTIX",
    "benchmark_script": "3.1.1",
    "device_name": "NVIDIA GeForce RTX 4090",
    "dist_name": "",
    "dist_version": "",
    "driver": null,
    "num_cpu_cores": 16,
    "num_cpu_sockets": 1,
    "num_cpu_threads": 24,
    "power_limit": null,
    "ram_bytes": null,
    "settings_status": "unknown",
    "system": "Windows"
  },
  "scene": "monster",
  "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
  "version": "4.5.0"
}
No comparable data

Blender scene rendering

5,673.980956805273 samples/min

Median and quartilesn = 33

4.5.0 · monster · OPTIX · Windows · 1P · 16 cores · 24 threads

Box: middle 50% (P25–P75)Vertical line: median (P50)

P25
5,506.59
P50 · Median
5,673.98
P75
5,723.08

Counts represent submissions; the number of independent systems is unknown. Stock or overclocked status, RAM and power settings are also unknown.

Measurement scope and interpretation

Sample thresholds are an initial display policy, not a guarantee of statistical confidence. P95 is not a theoretical limit. OS submission shares do not represent user market share.

Observed minimum / maximum
5,179.37 / 6,043.83
Measurement dates
2025-07-25 – 2026-03-08
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

Run count: 33

View source ↗

Result and source conditions
{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "median": 5673.980956805273,
    "observed_max": 6043.8253054552515,
    "observed_min": 5179.373228983298,
    "percentiles": {
      "p25": 5506.588359105658,
      "p50": 5673.980956805273,
      "p75": 5723.080384230446
    },
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 33,
    "sample_status": "quartiles",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "OPTIX",
      "benchmark_script": "3.1.1",
      "device_name": "NVIDIA GeForce RTX 4090",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 16,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 24,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "4.5.0"
  },
  "condition_id": "472eb3be12ca5e6d0f74",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "4b4267b4-9d00-4077-85fe-2aabc82d78c4",
    "5af570aa-7dd2-41f5-a568-07a219c1bd0a",
    "980bde14-9645-45ac-ade4-6017bbf95e94"
  ],
  "measured_from": "2025-07-25T14:05:16.079484+00:00",
  "measured_to": "2026-03-08T20:29:41.349239+00:00",
  "median": 5673.980956805273,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "percentiles": {
    "p25": 5506.588359105658,
    "p50": 5673.980956805273,
    "p75": 5723.080384230446
  },
  "population": "public submitted runs; not independent systems",
  "product_id": "geforce-rtx-4090",
  "run_count": 33,
  "sample_status": "quartiles"
}

DIEFOLIO / PRICE CONTEXT

Price & shopping

A dollar reference. Your currency.

GeForce RTX 5090

Launch starting price

USD 1,999.00

US dollar reference
Price sourceNVIDIA ↗

Price reference date: 2025-01-06
Source checked: 2026-10-11

International shop · check delivery country

AmazonNewegg

GeForce RTX 4090

Launch starting price

USD 1,599.00

US dollar reference
Price sourceNVIDIA ↗

Price reference date: 2022-09-20
Source checked: 2026-10-11

International shop · check delivery country

AmazonNewegg

Conversion applies to the US reference price. It excludes local retail prices, taxes, shipping and board-model differences.

Check current selling prices at the linked shops. Verify the exact model, stock and delivery terms.

Key specifications

Parameter and conditionsGeForce RTX 5090GeForce RTX 4090
Dedicated VRAM
32 GB
Source and conditions

View source ↗

{
  "document_id": "nvidia-geforce-0d2a5639216392f8",
  "parameter_id": "gpu.vram",
  "product_id": "geforce-rtx-5090",
  "qualifiers": {},
  "raw_label": "VRAM",
  "raw_value": "32",
  "state": "known",
  "value": 32
}
24 GB
Source and conditions

View source ↗

{
  "document_id": "nvidia-geforce-0d2a5639216392f8",
  "parameter_id": "gpu.vram",
  "product_id": "geforce-rtx-4090",
  "qualifiers": {},
  "raw_label": "VRAM",
  "raw_value": "24",
  "state": "known",
  "value": 24
}
CUDA cores
21,760
Source and conditions

View source ↗

{
  "document_id": "nvidia-geforce-0d2a5639216392f8",
  "parameter_id": "gpu.cuda_cores",
  "product_id": "geforce-rtx-5090",
  "qualifiers": {},
  "raw_label": "NVIDIA CUDA Cores",
  "raw_value": "21760",
  "state": "known",
  "value": 21760
}
16,384
Source and conditions

View source ↗

{
  "document_id": "nvidia-geforce-0d2a5639216392f8",
  "parameter_id": "gpu.cuda_cores",
  "product_id": "geforce-rtx-4090",
  "qualifiers": {},
  "raw_label": "NVIDIA CUDA Cores",
  "raw_value": "16384",
  "state": "known",
  "value": 16384
}
Memory bandwidth
1,792 GB/s
Source and conditions

View source ↗

{
  "document_id": "nvidia-geforce-0d2a5639216392f8",
  "parameter_id": "gpu.memory_bandwidth",
  "product_id": "geforce-rtx-5090",
  "qualifiers": {},
  "raw_label": "Memory Bandwidth",
  "raw_value": "1792 GB/sec",
  "state": "known",
  "value": 1792
}
Unavailable

Core counts and clock speeds are not workload performance rankings. TDP, base power and maximum turbo power are different metrics, not measured power consumption.

Product context and sources

GeForce RTX 5090

VRAM, bandwidth and compute configuration are separate specifications. AI TOPS is not converted into game FPS or model generation speed. Check measurements with the software, driver and precision required by the workload.

Full specifications and sources →

GeForce RTX 4090

24GB of VRAM does not guarantee compatibility with every AI model. Check the model, quantization, context length and runtime; CUDA core counts across generations are not measured performance ratios.

Full specifications and sources →