Diefolio
English

AMD · GPU

AMD Radeon PRO V620

AMD Radeon PRO V620 — Specifications and benchmarks. .

These specifications are supported by the linked sources. Core counts, clocks and memory capacities do not predict workload performance. Check benchmark versions, environments and sample counts.

Evidence review in progress · Search indexing deferred

Compare this product →

Power

ParameterValue and conditions
Board power
300 W
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Total Board Power (TBP)",
  "raw_value": "300W",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Total Board Power (TBP)",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
300 W
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Total Board Power (TBP)",
  "raw_value": "300W",
  "qualifiers": {
    "source_field": "Total Board Power (TBP)",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}

Compute

ParameterValue and conditions
Published process node
TSMC 7nm FinFET
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Lithography",
  "raw_value": "TSMC 7nm FinFET",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Lithography",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
TSMC 7nm FinFET
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Lithography",
  "raw_value": "TSMC 7nm FinFET",
  "qualifiers": {
    "source_field": "Lithography",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Architecture
AMD RDNA 2
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "GPU Architecture",
  "raw_value": "AMD RDNA 2",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "GPU Architecture",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
AMD RDNA 2
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "GPU Architecture",
  "raw_value": "AMD RDNA 2",
  "qualifiers": {
    "source_field": "GPU Architecture",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Compute units
72
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Compute Units",
  "raw_value": "72",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Compute Units",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
72
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Compute Units",
  "raw_value": "72",
  "qualifiers": {
    "source_field": "Compute Units",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
AMD ray accelerators
72
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Ray Accelerators",
  "raw_value": "72",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Ray Accelerators",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
72
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Ray Accelerators",
  "raw_value": "72",
  "qualifiers": {
    "source_field": "Ray Accelerators",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Stream processors
4,608
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Stream Processors",
  "raw_value": "4608",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Stream Processors",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
4,608
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Stream Processors",
  "raw_value": "4608",
  "qualifiers": {
    "source_field": "Stream Processors",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}

Memory

ParameterValue and conditions
Infinity Cache (documented MB)
128 MB
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "AMD Infinity Cache Technology",
  "raw_value": "128 MB",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "AMD Infinity Cache Technology",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
128 MB
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "AMD Infinity Cache Technology",
  "raw_value": "128 MB",
  "qualifiers": {
    "source_field": "AMD Infinity Cache Technology",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Memory bandwidth
512 GB/s
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Peak Memory Bandwidth",
  "raw_value": "512",
  "qualifiers": {
    "baseline_generation": "95d4e3b61a136b693f43",
    "evidence_scope": "exact listed manufacturer SKU; not a measured benchmark or a partner-board setting",
    "evidence_type": "manufacturer-specification",
    "field_evidence": {
      "Peak Memory Bandwidth": {
        "corroborating_sources": [],
        "location": {
          "reviewed_product_index_zero_based": 0
        },
        "quantity_scope": "manufacturer peak/maximum",
        "source_label": "Peak Memory Bandwidth",
        "source_value": "512 GB/s",
        "transformation": "explicit peak GB/s value; no bus-width or clock calculation"
      }
    },
    "normalization_reviewed_at": "2026-10-11",
    "source_field": "Peak Memory Bandwidth",
    "source_row": 1,
    "source_sha256": "4a11419a8903f732cddf2ab33f9cf70e9ee392d3733526052871e4ede37fbd00"
  },
  "document_id": "gap-fill-2026-10-11-6e37eb607ec29720-4a11419a8903f732"
}
Memory bus
256 bit
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Memory Interface",
  "raw_value": "256-bit",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Memory Interface",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
256 bit
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Memory Interface",
  "raw_value": "256-bit",
  "qualifiers": {
    "source_field": "Memory Interface",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Graphics memory type
GDDR6
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Dedicated Memory Type",
  "raw_value": "GDDR6",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Dedicated Memory Type",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
GDDR6
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Dedicated Memory Type",
  "raw_value": "GDDR6",
  "qualifiers": {
    "source_field": "Dedicated Memory Type",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
Dedicated VRAM
32 GB
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Dedicated Memory Size",
  "raw_value": "32 GB",
  "qualifiers": {
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Dedicated Memory Size",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
32 GB
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Dedicated Memory Size",
  "raw_value": "32 GB",
  "qualifiers": {
    "source_field": "Dedicated Memory Size",
    "source_row": 1
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}

Connectivity and features

ParameterValue and conditions
GPU FP16 Vector dense
40.55 TFLOPS
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Peak Half Precision (FP16 Vector) Performance",
  "raw_value": "40.55 TFLOPs",
  "qualifiers": {
    "engine": "vector",
    "measurement_type": "manufacturer_peak",
    "precision": "FP16",
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Peak Half Precision (FP16 Vector) Performance",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40,
    "sparsity": "dense"
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
40.55 TFLOPS
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Peak Half Precision (FP16 Vector) Performance",
  "raw_value": "40.55 TFLOPs",
  "qualifiers": {
    "engine": "vector",
    "measurement_type": "manufacturer_peak",
    "precision": "FP16",
    "source_field": "Peak Half Precision (FP16 Vector) Performance",
    "source_row": 1,
    "sparsity": "dense"
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}
GPU FP32 Vector dense
20.28 TFLOPS
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Peak Single Precision (FP32 Vector) Performance",
  "raw_value": "20.28 TFLOPs",
  "qualifiers": {
    "engine": "vector",
    "measurement_type": "manufacturer_peak",
    "precision": "FP32",
    "release_evidence": "dated-manufacturer-listing",
    "source_field": "Peak Single Precision (FP32 Vector) Performance",
    "source_product_url": "https://www.amd.com/en/products/accelerators/radeon-pro/amd-radeon-pro-v620.html",
    "source_row": 40,
    "sparsity": "dense"
  },
  "document_id": "amd-pro-graphics-table-1787b597aed7af7f"
}
20.28 TFLOPS
Source and conditions

View source ↗

{
  "state": "known",
  "raw_label": "Peak Single Precision (FP32 Vector) Performance",
  "raw_value": "20.28 TFLOPs",
  "qualifiers": {
    "engine": "vector",
    "measurement_type": "manufacturer_peak",
    "precision": "FP32",
    "source_field": "Peak Single Precision (FP32 Vector) Performance",
    "source_row": 1,
    "sparsity": "dense"
  },
  "document_id": "detail-e07f0b4810b604d2-4a11419a8903f732"
}

Public benchmarks

Compare results with matching versions, scenes, settings and statistics. A run count of 0 identifies an aggregate or reported value, not a newly measured run.

Blender (4)

Blender scene rendering

Too few submissions · No representative score

Run count: 2

Too few submissions · Individual results onlyn = 2

5.1.1 · classroom · HIP · Windows · 1P · 32 cores · 64 threads

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
744.54 / 745.54
Measurement dates
2026-06-21 – 2026-06-22
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

View source ↗

{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "individual_results": [
      {
        "measured_at": "2026-06-21T12:10:12.720639+00:00",
        "score": 744.5414955537336,
        "source_record_id": "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3"
      },
      {
        "measured_at": "2026-06-22T03:12:07.488321+00:00",
        "score": 745.5350919979007,
        "source_record_id": "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
      }
    ],
    "median": null,
    "observed_max": 745.5350919979007,
    "observed_min": 744.5414955537336,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 2,
    "sample_status": "individual",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "HIP",
      "benchmark_script": "3.1.2",
      "device_name": "AMD Radeon Pro V620",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 32,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 64,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "classroom",
    "scene_checksum": "936f0a528f03eba291c67ea87743a0ebde58938fb87471b262721b89b46b875d",
    "version": "5.1.1"
  },
  "condition_id": "b0134d43e37954a249b8",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3",
    "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
  ],
  "measured_from": "2026-06-21T12:10:12.720639+00:00",
  "measured_to": "2026-06-22T03:12:07.488321+00:00",
  "median": null,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "radeon-pro-v620",
  "run_count": 2,
  "sample_status": "individual"
}

Blender scene rendering

Too few submissions · No representative score

Run count: 2

Too few submissions · Individual results onlyn = 2

5.1.1 · junkshop · HIP · Windows · 1P · 32 cores · 64 threads

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
918.78 / 919.83
Measurement dates
2026-06-21 – 2026-06-22
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

View source ↗

{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "individual_results": [
      {
        "measured_at": "2026-06-21T12:09:26.854642+00:00",
        "score": 918.7786536257012,
        "source_record_id": "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3"
      },
      {
        "measured_at": "2026-06-22T03:11:21.497790+00:00",
        "score": 919.8330529904524,
        "source_record_id": "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
      }
    ],
    "median": null,
    "observed_max": 919.8330529904524,
    "observed_min": 918.7786536257012,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 2,
    "sample_status": "individual",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "HIP",
      "benchmark_script": "3.1.2",
      "device_name": "AMD Radeon Pro V620",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 32,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 64,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "junkshop",
    "scene_checksum": "f5515a21a1337b908212d8d76733530df83014464a24c585ab7b96000da0cdce",
    "version": "5.1.1"
  },
  "condition_id": "9954bf304d61713115b0",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3",
    "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
  ],
  "measured_from": "2026-06-21T12:09:26.854642+00:00",
  "measured_to": "2026-06-22T03:11:21.497790+00:00",
  "median": null,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "radeon-pro-v620",
  "run_count": 2,
  "sample_status": "individual"
}

Blender scene rendering

Too few submissions · No representative score

Run count: 2

Too few submissions · Individual results onlyn = 2

5.1.1 · monster · HIP · Windows · 1P · 32 cores · 64 threads

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
974.7 / 976.69
Measurement dates
2026-06-21 – 2026-06-22
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

View source ↗

{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "individual_results": [
      {
        "measured_at": "2026-06-21T12:08:31.375767+00:00",
        "score": 974.6979742527102,
        "source_record_id": "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3"
      },
      {
        "measured_at": "2026-06-22T03:10:26.613289+00:00",
        "score": 976.6862020830388,
        "source_record_id": "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
      }
    ],
    "median": null,
    "observed_max": 976.6862020830388,
    "observed_min": 974.6979742527102,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 2,
    "sample_status": "individual",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "HIP",
      "benchmark_script": "3.1.2",
      "device_name": "AMD Radeon Pro V620",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 32,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 64,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster",
    "scene_checksum": "8ff812a02f720433de1403dab447be9541eecca71cc919224a7790d017c1b2ce",
    "version": "5.1.1"
  },
  "condition_id": "ac7939326d2ad10e0620",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3",
    "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
  ],
  "measured_from": "2026-06-21T12:08:31.375767+00:00",
  "measured_to": "2026-06-22T03:10:26.613289+00:00",
  "median": null,
  "parameter_id": "benchmark.blender.samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "radeon-pro-v620",
  "run_count": 2,
  "sample_status": "individual"
}

Blender complete benchmark

Too few submissions · No representative score

Run count: 2

Too few submissions · Individual results onlyn = 2

5.1.1 · monster + junkshop + classroom · HIP · Windows · 1P · 32 cores · 64 threads

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
2,638.02 / 2,642.05
Measurement dates
2026-06-21 – 2026-06-22
Duplicate submissions removed
0
Quantile calculation
linear interpolation · (n−1) × p

View source ↗

{
  "analysis": {
    "duplicates_removed": 0,
    "independent_system_count": null,
    "individual_results": [
      {
        "measured_at": "2026-06-21T12:10:12.720639+00:00",
        "score": 2638.018123432145,
        "source_record_id": "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3"
      },
      {
        "measured_at": "2026-06-22T03:12:07.488321+00:00",
        "score": 2642.054347071392,
        "source_record_id": "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
      }
    ],
    "median": null,
    "observed_max": 2642.054347071392,
    "observed_min": 2638.018123432145,
    "policy": "submitted-runs-v1",
    "quantile_method": "linear interpolation at (n-1)*p",
    "sample_count": 2,
    "sample_status": "individual",
    "settings_status": "unknown"
  },
  "comparison_readiness": "observational-only",
  "condition": {
    "benchmark": "Blender",
    "environment": {
      "backend": "HIP",
      "benchmark_script": "3.1.2",
      "device_name": "AMD Radeon Pro V620",
      "dist_name": "",
      "dist_version": "",
      "driver": null,
      "num_cpu_cores": 32,
      "num_cpu_sockets": 1,
      "num_cpu_threads": 64,
      "power_limit": null,
      "ram_bytes": null,
      "settings_status": "unknown",
      "system": "Windows"
    },
    "scene": "monster + junkshop + classroom",
    "scene_checksum": "c3453da9769193d968fb1cc675fd7f92a3424abe9ac730218f2e00b953389356",
    "version": "5.1.1"
  },
  "condition_id": "07a53bb4e838193c7ae6",
  "document_id": "blender-open-data-3b13b06e2e84b715",
  "example_record_ids": [
    "d2738c86-fdc1-47e9-aeb4-aaf64cfbb6b3",
    "79bbb4b0-85e0-455b-a3c8-3668940b9b61"
  ],
  "measured_from": "2026-06-21T12:10:12.720639+00:00",
  "measured_to": "2026-06-22T03:12:07.488321+00:00",
  "median": null,
  "parameter_id": "benchmark.blender.total_samples_per_minute",
  "population": "public submitted runs; not independent systems",
  "product_id": "radeon-pro-v620",
  "run_count": 2,
  "sample_status": "individual"
}

Sources

Source terms and retrieval dates are listed below. Published values and conditions may change.