AMD Radeon PRO V620 vs NVIDIA GeForce RTX 5090 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon PRO V620 has a higher average benchmark score at 136,472, while the NVIDIA GeForce RTX 5090 sits at 79,842. The AMD card also holds a higher percentile rank at 96 versus 92 for the NVIDIA card.

Q: How do the two cards compare in the Geekbench OpenCL test?

A: The NVIDIA GeForce RTX 5090 wins decisively, scoring 334,370 against the AMD Radeon PRO V620's 128,580. This represents a 61.5% lead for the NVIDIA card in that specific workload.

Q: What about the Vulkan benchmark results?

A: The NVIDIA GeForce RTX 5090 also wins the Geekbench Vulkan test with a score of 376,728 versus 144,364 for the AMD Radeon PRO V620, a 61.7% advantage.

Q: Which GPU has more memory and what type is it?

A: Both cards feature 32 GB of memory. The AMD Radeon PRO V620 uses GDDR6 on a 256-bit bus with 512.0 GB/s of bandwidth, while the NVIDIA GeForce RTX 5090 uses GDDR7 on a 512-bit bus with 1.79 TB/s of bandwidth.

Q: What are the production statuses of these two GPUs?

A: The AMD Radeon PRO V620 is end-of-life, having been released on November 3, 2021, while the NVIDIA GeForce RTX 5090 is active, with a release date of January 29, 2025.

Q: Who are the nearest rivals for each card according to the database?

A: The AMD Radeon PRO V620's closest rivals include the AMD Radeon Pro W6800X Duo (0.5% gap), AMD Radeon PRO W6800 (0.8% gap), NVIDIA A10M (0.9% gap), and NVIDIA RTX 4000 Ada Generation (0.9% gap). The NVIDIA GeForce RTX 5090's nearest rivals are the NVIDIA Tesla P100 PCIe 16 GB (0.3% gap), NVIDIA Tesla P100 PCIe 12 GB (0.6% gap), AMD Radeon RX 6850M XT (1.1% gap), and AMD Radeon Pro Vega 64X (-1.4% gap).

The Verdict

The data splits this comparison into two very different use cases. For raw compute performance in the two recorded benchmark tests, the NVIDIA GeForce RTX 5090 is the clear winner, beating the AMD Radeon PRO V620 by more than 60% in both Geekbench OpenCL and Vulkan. Anyone prioritizing raw throughput in those workloads should choose the NVIDIA card.

However, the AMD Radeon PRO V620 shows a much higher average benchmark score across the full database (136,472 versus 79,842) and a better percentile ranking (96th versus 92nd percentile). Its nearest rivals are all professional workstation cards with scores clustered tightly around 135,000, suggesting it sits in a consistent performance tier. The RTX 5090's average is pulled down by its mixed benchmark results, which include low DirectX scores (for example, 185 in Passmark DirectX 12 and 226 in Passmark DirectX 10) alongside very high compute scores.

The RTX 5090 is a dual-slot card with a 575 W TDP, requires a 950 W suggested PSU, and carries a 1,999 USD launch MSRP. The AMD card is also dual-slot but draws 300 W and suggests a 700 W PSU. The AMD card has no display outputs, making it a compute-only accelerator, while the RTX 5090 includes 1x HDMI 2.1b and 3x DisplayPort 2.1b outputs.

Head-to-Head Benchmarks

The only two benchmark tests directly run on both cards are Geekbench OpenCL and Geekbench Vulkan, and the NVIDIA GeForce RTX 5090 wins both.

In Geekbench OpenCL, the RTX 5090 scores 334,370 against 128,580 for the AMD Radeon PRO V620. That is a 61.5% delta in favor of NVIDIA. The gap is nearly identical in Geekbench Vulkan, where the RTX 5090 posts 376,728 versus 144,364, a 61.7% lead.

These are substantial margins. The NVIDIA card more than doubles the AMD card's OpenCL score and comes close to tripling its Vulkan score. Looking at the RTX 5090's broader benchmark set, its compute-oriented results are consistently strong: 39,650 in Passmark G3D, 26,756 in Passmark GPU Compute, and 1,413 in Passmark G2D. Its DirectX-specific Passmark scores are much lower (395 in DirectX 9, 341 in DirectX 11, 226 in DirectX 10, 185 in DirectX 12), which explains why the average sits at 79,842 despite the impressive compute numbers.

The AMD card's two recorded benchmarks are both above 128,000, and its average of 136,472 reflects that consistency. The head-to-head data shows a clear winner for each individual test, but the aggregate picture is more nuanced because the database includes a different mix of benchmarks for each card.

Specification Differences

The two GPUs differ across nearly every major specification category.

Process and die: The AMD Radeon PRO V620 uses a 7 nm process with 26,800 million transistors on a 520 mm² die. The NVIDIA GeForce RTX 5090 uses a 5 nm process with 92,200 million transistors on a 750 mm² die. Transistor density is 51.5M per mm² for AMD versus 122.9M per mm² for NVIDIA.

Clock speeds: The AMD card has a base clock of 1825 MHz and a boost clock of 2200 MHz. The NVIDIA card runs at 2017 MHz base and 2407 MHz boost. Memory clocks are 2000 MHz (16 Gbps effective) for AMD and 1750 MHz (28 Gbps effective) for NVIDIA.

Memory subsystem: Both have 32 GB, but AMD uses GDDR6 with a 256-bit bus and 512.0 GB/s bandwidth. NVIDIA uses GDDR7 with a 512-bit bus and 1.79 TB/s bandwidth.

Compute units: The AMD card has 4,608 shading units, 288 TMUs, 128 ROPs, and 72 ray tracing cores. The NVIDIA card has 21,760 shading units, 680 TMUs, 176 ROPs, 170 ray tracing cores, and 680 tensor cores.

Rates and throughput: Pixel rate is 281.6 GPixel/s for AMD versus 423.6 GPixel/s for NVIDIA. Texture rate is 633.6 GTexel/s versus 1,636.8 GTexel/s. FP32 is 20.28 TFLOPS versus 104.8 TFLOPS. FP16 is 40.55 TFLOPS (2:1 ratio) for AMD versus 104.8 TFLOPS (1:1 ratio) for NVIDIA.

Power and physical: TDP is 300 W for AMD and 575 W for NVIDIA. AMD uses 2x 8-pin power connectors; NVIDIA uses 1x 16-pin. Suggested PSU is 700 W versus 950 W. The AMD card is 267 mm long, 120 mm tall, and 50 mm wide. The NVIDIA card is 304 mm long, 137 mm tall, and 40 mm wide.

Interface and outputs: AMD uses PCIe 4.0 x16 and has no display outputs. NVIDIA uses PCIe 5.0 x16 and provides 1x HDMI 2.1b plus 3x DisplayPort 2.1b.

Status and release: AMD is end-of-life with a November 3, 2021 release date. NVIDIA is active with a January 29, 2025 release date. NVIDIA lists a 1,999 USD launch MSRP.

Architecture Differences

The AMD Radeon PRO V620 is built on RDNA 2.0 architecture using the Navi 21 chip, part of the Radeon Pro Navi (Navi II Series) generation. The NVIDIA GeForce RTX 5090 uses Blackwell 2.0 architecture on the GB202 chip, part of the GeForce 50-series.

The foundry is the same, TSMC, but the process nodes differ: 7 nm for AMD and 5 nm for NVIDIA. This node difference, combined with the larger die, gives NVIDIA a much higher transistor count (92,200 million versus 26,800 million) and more than double the transistor density (122.9M per mm² versus 51.5M per mm²).

The compute architecture differs substantially. AMD's RDNA 2.0 design includes 4,608 shading units and 72 ray tracing cores but no tensor cores. NVIDIA's Blackwell 2.0 design packs 21,760 shading units, 170 ray tracing cores, and 680 tensor cores. The FP16 ratio also differs: AMD runs at 2:1 (half the FP32 rate), while NVIDIA runs at 1:1 (equal to FP32).

Memory architecture is another major split. AMD pairs its 32 GB GDDR6 with a 256-bit bus, while NVIDIA uses 32 GB GDDR7 on a 512-bit bus, yielding 1.79 TB/s versus 512.0 GB/s of bandwidth. The bus interface also advances from PCIe 4.0 x16 to PCIe 5.0 x16.

Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so the API feature set is equivalent even though the underlying hardware generations are three years apart.

Where Each One Wins

The AMD Radeon PRO V620 wins on consistency and aggregate standing. Its average benchmark score of 136,472 places it in the 96th percentile of all GPUs, and its nearest rivals (AMD Radeon Pro W6800X Duo, AMD Radeon PRO W6800, NVIDIA A10M, NVIDIA RTX 4000 Ada Generation) are all within roughly 1% of its score. That cluster indicates the card delivers stable, predictable performance across the database's benchmark suite. It also wins on power efficiency: 300 W TDP versus 575 W, with a 700 W suggested PSU versus 950 W, and it is shorter at 267 mm versus 304 mm.

The NVIDIA GeForce RTX 5090 wins on raw compute performance in the head-to-head tests, with 61.5% and 61.7% margins in OpenCL and Vulkan respectively. It also offers display outputs, a larger memory bandwidth (1.79 TB/s versus 512.0 GB/s), a newer PCIe 5.0 interface, and tensor cores for workloads that leverage them. Its 104.8 TFLOPS FP32 and FP16 throughput dwarf the AMD card's 20.28 TFLOPS FP32 and 40.55 TFLOPS FP16.

The practical split is clear. For compute-heavy tasks measured by OpenCL and Vulkan, the RTX 5090 is the superior choice. For a workstation accelerator that sits in a tightly clustered performance tier with high aggregate scores and lower power demands, the Radeon PRO V620 holds its own despite being an end-of-life product. The RTX 5090's inconsistent benchmark profile, with high compute scores but low DirectX scores, means its average is dragged down, but the head-to-head results leave no doubt about which card delivers more raw performance in the tests where both were measured.