AMD Radeon PRO V620 vs NVIDIA RTX A3000 Mobile Comparison

FAQ

Q: How do the two GPUs compare in overall benchmark performance?

A: The AMD Radeon PRO V620 records an average benchmark score of 136,472, while the NVIDIA RTX A3000 Mobile scores 70,140. This places the AMD part in the 96th percentile of all GPUs in the database, versus the 91st percentile for the NVIDIA part.

Q: Which GPU wins in the OpenCL benchmark and by how much?

A: The AMD Radeon PRO V620 wins the Geekbench OpenCL test with a score of 128,580 against 79,091 for the RTX A3000 Mobile. That is a 62.6% advantage for the AMD card.

Q: Is the Vulkan result closer or further apart than OpenCL?

A: The gap widens in Vulkan. The AMD Radeon PRO V620 scores 144,364, while the NVIDIA RTX A3000 Mobile scores 61,189, a 135.9% lead for the AMD part.

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

A: The AMD Radeon PRO V620 has 32 GB of GDDR6 on a 256-bit bus, delivering 512.0 GB/s of bandwidth. The NVIDIA RTX A3000 Mobile has 6 GB of GDDR6 on a 192-bit bus, delivering 264.0 GB/s.

Q: How do the closest rivals compare to each GPU in the database?

A: The AMD part sits within 0.9% of the AMD Radeon PRO W6800, the AMD Radeon Pro W6800X Duo, the NVIDIA A10M, and the NVIDIA RTX 4000 Ada Generation. The NVIDIA mobile part is within 1.7% of the NVIDIA Quadro P6000, the AMD Radeon Pro WX 8200, the AMD Radeon RX 6600 LE, and the NVIDIA CMP 90HX.

Q: What is the production status and release timing for each?

A: Both are end-of-life products. The NVIDIA RTX A3000 Mobile was released in April 2021, and the AMD Radeon PRO V620 followed in November 2021.

Architecture Differences

The AMD Radeon PRO V620 is built on the RDNA 2.0 architecture using the Navi 21 chip, fabricated on a 7 nm process at TSMC. The die contains 26,800 million transistors across a 520 mm² area, giving a transistor density of 51.5M per mm². The NVIDIA RTX A3000 Mobile uses the Ampere architecture with the GA104 chip, manufactured on an 8 nm process at Samsung. That die holds 17,400 million transistors over 392 mm², for a density of 44.4M per mm². The AMD chip is both larger and denser, and it uses a more advanced process node.

Compute resources differ significantly. The AMD GPU has 4,608 shading units, 288 texture mapping units, and 128 ROPs. The NVIDIA GPU has 4,096 shading units, 128 TMUs, and 64 ROPs. Ray tracing hardware also diverges: the AMD card carries 72 ray tracing cores, while the NVIDIA card has 32. The NVIDIA part includes 128 tensor cores, a feature the AMD part does not list at all.

The clock behavior is distinctly different. The AMD Radeon PRO V620 runs at a base clock of 1825 MHz and boosts to 2200 MHz, with memory clocked at 2000 MHz (16 Gbps effective). The NVIDIA RTX A3000 Mobile has a much lower base clock of 600 MHz and a boost of 1230 MHz, with memory at 1375 MHz (11 Gbps effective). The AMD part is a desktop-style card with a dual-slot cooler, 267 mm length, 120 mm height, and 50 mm width, requiring two 8-pin power connectors. The NVIDIA part is a mobile GPU with no listed dimensions and no dedicated power connectors, as it depends on the host laptop for power delivery.

The AMD card lists no display outputs, meaning it is intended for compute or virtualized workloads. The NVIDIA card lists "Portable Device Dependent" outputs, as it drives displays through the host system. Both support PCIe 4.0 x16, DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The NVIDIA card has a stated TDP of 70 W, while the AMD card is rated at 300 W with a suggested power supply of 700 W.

Head-to-Head Benchmarks

The database records two direct comparisons between these GPUs, and the AMD Radeon PRO V620 wins both. In Geekbench OpenCL, the AMD card posts 128,580 against 79,091 for the NVIDIA RTX A3000 Mobile, a 62.6% margin. That is a substantial lead, but the Vulkan result is even more lopsided. The AMD card scores 144,364, while the NVIDIA card manages 61,189, which works out to a 135.9% advantage.

The Vulkan gap is worth interpreting carefully. The AMD part nearly doubles the NVIDIA mobile GPU in that API, which points to a large difference in raw throughput and memory bandwidth. The AMD card's 512.0 GB/s of bandwidth versus 264.0 GB/s is a direct contributor, as is its 20.28 TFLOPS of FP32 compute against 10.08 TFLOPS for the NVIDIA part. The AMD card also has a 2:1 FP16 ratio, yielding 40.55 TFLOPS, while the NVIDIA card computes FP16 at a 1:1 ratio, so its FP16 figure is the same 10.08 TFLOPS.

Texture and pixel throughput tell the same story. The AMD card outputs 633.6 GTexel/s and 281.6 GPixel/s. The NVIDIA card outputs 157.4 GTexel/s and 78.72 GPixel/s. The AMD card is roughly four times faster in pixel fill and texture fill, which aligns with the very large Vulkan delta. The NVIDIA card does have tensor cores, which the AMD card lacks, and those can accelerate certain AI workloads, but in the two recorded general compute benchmarks, the AMD part dominates.

The average benchmark score of 136,472 for the AMD card versus 70,140 for the NVIDIA card reflects those two results. The AMD card's nearest rivals in the database, the AMD Radeon Pro W6800X Duo at 135,774 and the AMD Radeon PRO W6800 at 135,396, sit just 0.5% and 0.8% behind. The NVIDIA RTX A3000 Mobile, by contrast, is clustered with the NVIDIA Quadro P6000 at 69,986 and the AMD Radeon Pro WX 8200 at 69,870, with deltas of 0.2% and 0.4%. The RTX A3000 Mobile actually trails the AMD Radeon RX 6600 LE by 1% (70,829 versus 70,140) and beats the NVIDIA CMP 90HX by 1.7%.

Specification Differences

The two GPUs differ across almost every major specification field. The process node is 7 nm for AMD and 8 nm for NVIDIA. Transistor count is 26,800 million versus 17,400 million, and die size is 520 mm² versus 392 mm². Transistor density is 51.5M per mm² versus 44.4M per mm².

Memory capacity is a major differentiator: 32 GB versus 6 GB. Both use GDDR6, but the bus width is 256-bit for AMD and 192-bit for NVIDIA. Memory bandwidth is 512.0 GB/s versus 264.0 GB/s. Memory clocks are 2000 MHz (16 Gbps effective) versus 1375 MHz (11 Gbps effective).

Compute units differ across every category. Shading units are 4,608 versus 4,096. TMUs are 288 versus 128. ROPs are 128 versus 64. Ray tracing cores are 72 versus 32. Tensor cores exist only on the NVIDIA card, with 128 of them. The AMD card has no Tensor core equivalent listed.

Clock speeds are far apart. Base clock is 1825 MHz versus 600 MHz, and boost clock is 2200 MHz versus 1230 MHz. Pixel rate is 281.6 GPixel/s versus 78.72 GPixel/s. Texture rate is 633.6 GTexel/s versus 157.4 GTexel/s. FP32 compute is 20.28 TFLOPS versus 10.08 TFLOPS. FP16 compute is 40.55 TFLOPS (2:1) versus 10.08 TFLOPS (1:1).

Power characteristics differ just as sharply. The AMD card has a 300 W TDP, a dual-slot cooler, two 8-pin power connectors, and a suggested 700 W power supply. The NVIDIA mobile part has a 70 W TDP, no listed slot width, no power connectors, and no suggested power supply. Physical dimensions are listed only for the AMD card: 267 mm long, 120 mm high, and 50 mm wide. The NVIDIA card has no listed dimensions.

Outputs also differ. The AMD card has no display outputs, while the NVIDIA card lists "Portable Device Dependent" outputs. Both use PCIe 4.0 x16 and share the same API support for DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The AMD card's predecessor is the Radeon Pro Vega, while the NVIDIA card's predecessor is the Quadro Turing-M and its successor is Ada-MW.

The Verdict

The data supports a clear split. The AMD Radeon PRO V620 is the far faster GPU in both recorded benchmarks, with a 62.6% OpenCL lead and a 135.9% Vulkan lead. It also carries 32 GB of memory versus 6 GB, doubles the FP32 throughput at 20.28 TFLOPS against 10.08 TFLOPS, and offers more than double the memory bandwidth. For compute-heavy workloads that rely on raw throughput, the AMD card is the obvious choice, provided the system can accommodate a 300 W dual-slot card with two 8-pin connectors and a 700 W power supply.

The NVIDIA RTX A3000 Mobile is the only option here for a laptop form factor. Its 70 W TDP, lack of dedicated power connectors, and portable-device-dependent outputs make it suitable for mobile workstations. It also includes 128 tensor cores, which the AMD card does not have, so AI inference and tensor-accelerated tasks may favor the NVIDIA part even though its general compute scores are lower. The NVIDIA card's 91st percentile ranking and its position within 1.7% of several desktop-class rivals show it is not a weak performer; it is simply outclassed by the AMD card's desktop-class specifications.

Users who need maximum compute performance in a fixed chassis should pick the AMD Radeon PRO V620. Users who require a mobile GPU with tensor acceleration and lower power draw should pick the NVIDIA RTX A3000 Mobile. The benchmark database records two wins for the AMD card and zero for the NVIDIA card, and the specification sheet reinforces that verdict.