AMD Radeon PRO V620 vs NVIDIA GeForce RTX 3090 Ti Comparison

AMD Radeon PRO V620 vs NVIDIA GeForce RTX 3090 Ti pits a professional workstation card against a flagship consumer GPU, and the benchmark data shows a clear performance hierarchy. The GeForce RTX 3090 Ti wins both head-to-head tests decisively, while the Radeon PRO V620 counters with a substantial memory capacity advantage and a lower power envelope. These are two end-of-life cards with very different design philosophies, and the data reflects that split in everything from silicon size to compute throughput.

Head-to-Head Benchmarks

The GeForce RTX 3090 Ti dominates the available benchmark comparisons, taking both recorded tests. In Geekbench OpenCL, the NVIDIA card scores 174,441 against the AMD card’s 128,580, a delta of -26.3% from the AMD card’s perspective. That is a substantial margin — the RTX 3090 Ti is more than a quarter faster in raw compute throughput for this workload. The Vulkan result is even more lopsided: the RTX 3090 Ti posts 215,633 while the Radeon PRO V620 manages 144,364, a -33.1% gap. In Vulkan, the NVIDIA card is roughly half again as fast as the AMD card.

These are not marginal wins. The RTX 3090 Ti’s advantage in both APIs suggests a fundamental throughput difference rather than a workload-specific quirk. The average benchmark score tells a slightly different story, however. The Radeon PRO V620 averages 136,472 across its benchmark suite, while the RTX 3090 Ti averages 131,938. That puts the AMD card ahead by about 3.4% on average, even though it loses both individual head-to-head tests. The reason is that the Radeon PRO V620’s benchmark pool includes only Geekbench OpenCL and Vulkan scores, while the RTX 3090 Ti’s pool also includes a 3DMark Steel Nomad DX12 result of 5,741 — a low score that drags its average down. The data shows the RTX 3090 Ti wins every test it shares with the AMD card, but the AMD card’s average is boosted by having fewer, higher-scoring tests in its pool.

Percentile rankings place both cards near the top of the database. The Radeon PRO V620 sits at the 96th percentile of all GPUs, while the RTX 3090 Ti sits at the 95th percentile. That is a negligible difference for practical purposes — both are elite performers. The nearest rival data reinforces this: the Radeon PRO V620’s closest competitor is the AMD Radeon Pro W6800X Duo at 135,774 (0.5% delta), while the RTX 3090 Ti’s nearest rival is the NVIDIA L4 at 131,072 (0.7% delta). Both cards are clustered with other high-end workstation parts, indicating they occupy similar performance territory in aggregate, even if the head-to-head results favor NVIDIA.

Architecture Differences

The silicon behind these two cards could hardly be more different. The AMD Radeon PRO V620 uses the Navi 21 chip built on RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC. The die measures 520 mm² and contains 26,800 million transistors, yielding a density of 51.5 million transistors per square millimeter. The NVIDIA GeForce RTX 3090 Ti uses the GA102 chip on Ampere architecture, fabricated on an 8 nm process at Samsung. Its die is substantially larger at 628 mm² with 28,300 million transistors, but the density is lower at 45.1 million per square millimeter. The NVIDIA chip packs more transistors overall, but the AMD chip achieves higher density on a smaller die.

The compute configurations diverge sharply. The Radeon PRO V620 has 4,608 shading units, 288 texture mapping units, 128 ROPs, and 72 ray tracing cores. The RTX 3090 Ti has 10,752 shading units, 336 TMUs, 112 ROPs, 84 RT cores, and adds 336 tensor cores that the AMD card lacks entirely. The shading unit count is more than double on the NVIDIA side, which explains much of the raw throughput advantage. In terms of peak compute, the RTX 3090 Ti delivers 40.00 TFLOPS FP32 and 40.00 TFLOPS FP16 (1:1 ratio), while the Radeon PRO V620 delivers 20.28 TFLOPS FP32 and 40.55 TFLOPS FP16 (2:1 ratio). The NVIDIA card has exactly double the FP32 throughput, while the AMD card slightly exceeds it in FP16 thanks to its packed math.

Clock speeds tell a complementary story. The Radeon PRO V620 runs at a base clock of 1825 MHz and boost of 2200 MHz, while the RTX 3090 Ti runs at 1560 MHz base and 1860 MHz boost. The AMD card’s higher clocks help it close the gap in some workloads, but the NVIDIA card’s massive shader count overwhelms that advantage in the benchmark data. Pixel rate favors AMD at 281.6 GPixel/s versus 208.3 GPixel/s, but texture rate is nearly identical: 633.6 GTexel/s for AMD versus 625.0 GTexel/s for NVIDIA. The Radeon card’s higher ROP count and clock speed give it the edge in fill-rate-bound tasks.

Memory subsystems are a major differentiator. The Radeon PRO V620 has 32 GB of GDDR6 on a 256-bit bus with 512.0 GB/s bandwidth. The RTX 3090 Ti has 24 GB of GDDR6X on a 384-bit bus with 1.01 TB/s bandwidth. The NVIDIA card has roughly double the bandwidth but two-thirds the capacity. The AMD card’s memory runs at 2000 MHz (16 Gbps effective), while the NVIDIA card’s runs at 1313 MHz (21 Gbps effective). The higher data rate per pin on the GDDR6X modules compensates for the lower clock speed. Both cards use PCIe 4.0 x16 interfaces, and both support DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4.

Where Each One Wins

The RTX 3090 Ti wins on raw compute performance. Both head-to-head benchmarks — Geekbench OpenCL and Geekbench Vulkan — go to NVIDIA with margins of 26.3% and 33.1% respectively. The data shows the NVIDIA card is the clear choice for any workload that stresses general compute throughput, particularly in Vulkan where the gap is widest. Its 40.00 TFLOPS FP32 rate versus 20.28 TFLOPS for the AMD card is the underlying driver. For tasks like rendering, simulation, or any GPU-compute application that scales with shading unit count, the RTX 3090 Ti is the stronger performer.

The Radeon PRO V620 wins on memory capacity and efficiency. Its 32 GB frame buffer is 8 GB larger than the RTX 3090 Ti’s 24 GB, which matters for workloads that exceed the NVIDIA card’s memory ceiling. The AMD card also draws significantly less power: 300 W TDP versus 450 W TDP, with a suggested PSU of 700 W versus 850 W. The AMD card is a dual-slot design that is 267 mm long, 120 mm tall, and 50 mm wide, while the RTX 3090 Ti is a triple-slot card measuring 336 mm by 140 mm by 61 mm. The Radeon card is shorter, narrower, and thinner, making it easier to fit into constrained chassis. It also has no display outputs, which is a notable difference from the RTX 3090 Ti’s 1x HDMI 2.1 and 3x DisplayPort 1.4a.

FAQ

Q: Which card has better raw compute performance?

A: The RTX 3090 Ti wins both head-to-head benchmarks. It scores 174,441 in Geekbench OpenCL versus 128,580 for the Radeon PRO V620, and 215,633 in Geekbench Vulkan versus 144,364.

Q: How do their average benchmark scores compare?

A: The Radeon PRO V620 has a higher average benchmark score of 136,472, while the RTX 3090 Ti averages 131,938. This is because the NVIDIA card’s average includes a 3DMark Steel Nomad DX12 score of 5,741, which lowers its mean.

Q: Which card has more memory?

A: The Radeon PRO V620 has 32 GB of GDDR6, while the RTX 3090 Ti has 24 GB of GDDR6X. The NVIDIA card has higher bandwidth at 1.01 TB/s versus 512.0 GB/s.

Q: What are the power requirements?

A: The Radeon PRO V620 has a 300 W TDP and suggests a 700 W PSU. The RTX 3090 Ti has a 450 W TDP and suggests an 850 W PSU.

Q: Do both cards support the same APIs?

A: Yes, both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Q: Which card has more shading units?

A: The RTX 3090 Ti has 10,752 shading units, while the Radeon PRO V620 has 4,608. The NVIDIA card also has 336 tensor cores, which the AMD card lacks.

The Verdict

The data supports a straightforward conclusion: the NVIDIA GeForce RTX 3090 Ti is the faster card in every shared benchmark, with wins of 26.3% in OpenCL and 33.1% in Vulkan. Its doubling of FP32 throughput (40.00 TFLOPS versus 20.28 TFLOPS) and higher memory bandwidth (1.01 TB/s versus 512.0 GB/s) make it the superior choice for compute-heavy workloads that fit within its 24 GB memory capacity. The RTX 3090 Ti also offers display outputs and tensor cores, expanding its utility beyond pure compute.

The AMD Radeon PRO V620 is the better choice for workloads that need more memory capacity. Its 32 GB frame buffer exceeds the RTX 3090 Ti’s 24 GB, and it does so at 300 W versus 450 W with a smaller physical footprint. The AMD card’s higher clock speeds (2200 MHz boost versus 1860 MHz) and higher pixel rate (281.6 GPixel/s versus 208.3 GPixel/s) give it advantages in certain fill-rate-bound tasks. However, the benchmark data shows it trails the NVIDIA card significantly in the two tests recorded. For users who prioritize raw speed, the RTX 3090 Ti wins. For users who need maximum memory per watt or have physical space constraints, the Radeon PRO V620 offers a compelling profile despite its lower compute scores.

Specification Differences

| Specification | AMD Radeon PRO V620 | NVIDIA GeForce RTX 3090 Ti |

|---|---|---|

| Architecture | RDNA 2.0 | Ampere |

| Process Node | 7 nm (TSMC) | 8 nm (Samsung) |

| Die Size | 520 mm² | 628 mm² |

| Transistors | 26,800 million | 28,300 million |

| Transistor Density | 51.5M / mm² | 45.1M / mm² |

| Base Clock | 1825 MHz | 1560 MHz |

| Boost Clock | 2200 MHz | 1860 MHz |

| Memory Size | 32 GB GDDR6 | 24 GB GDDR6X |

| Memory Bus | 256 bit | 384 bit |

| Memory Bandwidth | 512.0 GB/s | 1.01 TB/s |

| Memory Clock | 2000 MHz (16 Gbps effective) | 1313 MHz (21 Gbps effective) |

| Shading Units | 4,608 | 10,752 |

| TMUs | 288 | 336 |

| ROPs | 128 | 112 |

| RT Cores | 72 | 84 |

| Tensor Cores | None | 336 |

| Pixel Rate | 281.6 GPixel/s | 208.3 GPixel/s |

| Texture Rate | 633.6 GTexel/s | 625.0 GTexel/s |

| FP32 Performance | 20.28 TFLOPS | 40.00 TFLOPS |

| FP16 Performance | 40.55 TFLOPS (2:1) | 40.00 TFLOPS (1:1) |

| TDP | 300 W | 450 W |

| Slot Width | Dual-slot | Triple-slot |

| Power Connectors | 2x 8-pin | 1x 16-pin |

| Suggested PSU | 700 W | 850 W |

| Display Outputs | No outputs | 1x HDMI 2.1, 3x DisplayPort 1.4a |

| Length | 267 mm (10.5 inches) | 336 mm (13.2 inches) |

| Height | 120 mm (4.7 inches) | 140 mm (5.5 inches) |

| Width | 50 mm (2 inches) | 61 mm (2.4 inches) |

| Release Date | 2021-11-03 | 2022-01-26 |

| Launch MSRP | Not available | 1,999 USD |

| Production Status | End-of-life | End-of-life |