AMD Radeon PRO V620 vs NVIDIA RTX 5000 Ada Generation Comparison

NVIDIA’s RTX 5000 Ada Generation and AMD’s Radeon PRO V620 are both 32 GB workstation cards, but they target different eras and design philosophies. The data shows a clear performance hierarchy, with the NVIDIA card leading in every available benchmark, though the AMD card holds its own in its respective peer group. This analysis breaks down the measurable differences and helps determine which card fits specific workloads.

Head-to-Head Benchmarks

The benchmark results are unambiguous: the NVIDIA RTX 5000 Ada Generation wins both recorded tests by a substantial margin. In Geekbench OpenCL, the NVIDIA card scores 175286 against AMD’s 128580, a delta of 36.3%. That is not a marginal victory; it represents a massive compute advantage in raw parallel processing. For Vulkan, the gap narrows slightly but remains decisive: NVIDIA scores 194041 versus AMD’s 144364, a 34.4% difference. Across both tests, NVIDIA secures 2 wins and AMD secures 0.

Contextualizing these scores against their respective rivals makes the picture clearer. The RTX 5000 Ada’s average benchmark score is 184664, placing it 0.5% above the NVIDIA A100 SXM4 80 GB (183725) and 1.4% above the newer RTX PRO 5000 Blackwell (182109). It also beats the GeForce RTX 4090 D (178050) by 3.7%. This places the RTX 5000 Ada in the top tier of all GPUs, sitting at the 98th percentile. Its only close rival in the data is the A100 SXM4 40 GB, which scores 187147 and leads by 1.3% — a negligible difference for most workloads.

The Radeon PRO V620, by contrast, averages 136472, which puts it at the 96th percentile. Its nearest rivals are all within 0.9%: the Radeon Pro W6800X Duo (135774), the Radeon PRO W6800 (135396), the NVIDIA A10M (135230), and the NVIDIA RTX 4000 Ada Generation (135218). This clustering indicates that the V620 is a solid mid-tier workstation card, but it operates in a different performance class than the RTX 5000 Ada. The 36.3% OpenCL and 34.4% Vulkan deficits are not minor tweaks; they are structural differences in compute throughput.

Architecture Differences

The two cards are built on fundamentally different architectures. The NVIDIA RTX 5000 Ada uses the AD102 chip on the Ada Lovelace architecture, fabricated on a 5 nm process at TSMC. It packs 76,300 million transistors into a 609 mm² die, yielding a transistor density of 125.3 million per mm². This is a modern, high-density design optimized for both compute and graphics workloads.

The AMD Radeon PRO V620 uses the Navi 21 chip on RDNA 2.0, built on a 7 nm process, also at TSMC. It contains 26,800 million transistors on a 520 mm² die, with a density of 51.5 million per mm². The older node and lower density explain much of the performance gap. The V620’s architecture is more about efficiency per watt than raw density, but its compute resources are far more limited.

The shading unit count tells the story: NVIDIA has 12,800 shading units, 400 TMUs, and 176 ROPs, while AMD has 4,608 shading units, 288 TMUs, and 128 ROPs. NVIDIA’s 100 RT cores and 400 tensor cores (AMD has no tensor core equivalent listed) give it a massive advantage in ray tracing and AI-accelerated tasks. The FP32 throughput is 65.28 TFLOPS for NVIDIA versus 20.28 TFLOPS for AMD. Interestingly, AMD’s FP16 rate is 40.55 TFLOPS (2:1 ratio), which is higher than its FP32, while NVIDIA’s FP16 is 65.28 TFLOPS (1:1) — meaning NVIDIA does not halve its rate for half-precision work.

FAQ

Q: Which card has higher memory bandwidth?

A: The NVIDIA RTX 5000 Ada leads with 576.0 GB/s, compared to the AMD Radeon PRO V620’s 512.0 GB/s. Both use 32 GB of GDDR6 on a 256-bit bus, but NVIDIA’s memory runs at 18 Gbps effective versus AMD’s 16 Gbps.

Q: Are these cards suitable for AI workloads?

A: The NVIDIA card has 400 tensor cores, making it explicitly suited for AI acceleration. The AMD card has no tensor cores listed, so its AI capability is limited to general compute. Benchmark data shows NVIDIA has a 36.3% lead in OpenCL, which often correlates with compute-heavy AI tasks.

Q: Do both cards support the same graphics APIs?

A: Yes, both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. There is no difference in API support, despite the architectural differences.

Q: What is the power draw difference?

A: The NVIDIA card has a TDP of 250 W and requires a 600 W PSU, while the AMD card has a TDP of 300 W and requires a 700 W PSU. NVIDIA achieves higher performance with lower power consumption.

Q: Which card has more display outputs?

A: The NVIDIA RTX 5000 Ada has 4x DisplayPort 1.4a outputs. The AMD Radeon PRO V620 has no display outputs, meaning it is designed for compute-only or server environments where video output is handled separately.

Q: Is one card newer than the other?

A: Yes. The NVIDIA card was released on 2023-08-08, while the AMD card came out earlier on 2021-11-03. The NVIDIA card is still marked as Active in production, while the AMD card is End-of-life.

Specification Differences

The core specifications diverge sharply. The process node is 5 nm for NVIDIA versus 7 nm for AMD. Transistor count is 76,300 million versus 26,800 million. Die size is 609 mm² versus 520 mm². The NVIDIA card has 12,800 shading units, 400 TMUs, and 176 ROPs, while AMD has 4,608 shading units, 288 TMUs, and 128 ROPs. NVIDIA’s RT cores number 100 versus AMD’s 72. NVIDIA has 400 tensor cores; AMD has none listed.

Clock speeds differ: NVIDIA runs at 1155 MHz base and 2550 MHz boost, while AMD runs at 1825 MHz base and 2200 MHz boost. Despite AMD’s higher base clock, NVIDIA’s higher boost and vastly larger shader count dominate. Memory speed is 2250 MHz (18 Gbps effective) for NVIDIA versus 2000 MHz (16 Gbps effective) for AMD. Bandwidth is 576.0 GB/s versus 512.0 GB/s.

Pixel rate is 448.8 GPixel/s for NVIDIA versus 281.6 GPixel/s for AMD. Texture rate is 1,020.0 GTexel/s versus 633.6 GTexel/s. FP32 is 65.28 TFLOPS versus 20.28 TFLOPS. TDP is 250 W versus 300 W, with NVIDIA using a single 16-pin connector and AMD using two 8-pin connectors. The suggested PSU is 600 W versus 700 W. Physical dimensions are similar in length (267 mm) and height (112 mm vs 120 mm), but AMD is thicker at 50 mm width versus NVIDIA’s unspecified width. Both are dual-slot and use PCIe 4.0 x16.

Where Each One Wins

The NVIDIA RTX 5000 Ada Generation wins in every measured category. It is faster in both OpenCL and Vulkan by over 34%. It has more shading units, more ROPs, more TMUs, more RT cores, and the only tensor cores in this comparison. It delivers higher pixel fill rate, higher texture rate, and more than three times the FP32 throughput. It does all this while consuming 50 W less power and fitting in a smaller height profile. For workloads that rely on raw compute — rendering, simulation, AI inference, or high-resolution video processing — the NVIDIA card is the clear choice.

The AMD Radeon PRO V620 wins in one specific scenario: when a card with no display outputs is required for a headless server or compute farm. Its lack of video outputs means it is purpose-built for environments where every watt and slot goes to computation, not display. It also has a higher base clock (1825 MHz vs 1155 MHz), which could benefit short, lightly-threaded tasks that do not scale to the full shader array. However, in the benchmark data, this clock advantage does not translate into any actual win. The AMD card is also end-of-life, so it may be available at a different price point in the used market, though that is outside the scope of this analysis.

The Verdict

The data is unequivocal: the NVIDIA RTX 5000 Ada Generation is the superior card. It wins both benchmarks by 34-36%, has a higher average score (184664 vs 136472), and sits at the 98th percentile versus the 96th for AMD. It is more power-efficient, has tensor cores for AI work, and supports display output. The only reason to choose the AMD Radeon PRO V620 is if you specifically need a card with no display outputs for a headless compute node, or if you are constrained to a system that requires the AMD card’s specific physical profile (it is 8 mm taller and 50 mm wide). For any workstation task that involves rendering, compute, or graphics output, the NVIDIA card is the correct pick based on the benchmark evidence. The AMD card is not a bad performer — it sits within 0.9% of the RTX 4000 Ada Generation — but it is in a different league than the RTX 5000 Ada. Choose NVIDIA if you need top-tier performance; choose AMD only if your rack demands a display-less compute card.