AMD Radeon PRO V620 vs AMD Radeon Pro W6800X Duo Comparison

The Verdict

The data presents a clear but narrow hierarchy. The AMD Radeon PRO V620 leads the AMD Radeon Pro W6800X Duo in both shared benchmarks, with an average benchmark score of 136,472 against 135,774. That is a 0.5% overall advantage, which places the V620 ahead of the W6800X Duo in the nearestRivals list. However, the W6800X Duo counters with a unique Geekbench Metal score of 157,365, a metric the V620 does not appear in. For general compute workloads measured by OpenCL and Vulkan, the V620 is the stronger pick. For Apple-centric Metal workloads, the W6800X Duo is the only option with data. The V620 is end-of-life, as is the W6800X Duo. The W6800X Duo carries a launch MSRP of 4,999 USD. The V620 has no launch MSRP listed.

Architecture Differences

Both cards are built on the same foundational design. The chip is Navi 21 for both, using the RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC. Transistor count is identical at 26,800 million, and die size is the same at 520 mm², yielding a transistor density of 51.5M / mm² for each. The memory subsystem is also identical on paper: 32 GB of GDDR6 on a 256-bit bus, with 2000 MHz memory clock and 16 Gbps effective speed, producing 512.0 GB/s of bandwidth.

The divergence begins with the execution units. The V620 has 4,608 shading units, 288 TMUs, 128 ROPs, and 72 ray tracing cores. The W6800X Duo has fewer of each: 3,840 shading units, 240 TMUs, 96 ROPs, and 60 ray tracing cores. This is a 768-shading-unit gap, a 48-TMU gap, a 32-ROP gap, and a 12-RT-core gap. Clock speeds also differ. The V620 runs a base clock of 1825 MHz and a boost of 2200 MHz. The W6800X Duo runs a base clock of 1800 MHz and a boost of 1967 MHz. The V620 is faster in both clock domains.

The board designs are completely different. The V620 is a dual-slot card with 2x 8-pin power connectors and a suggested PSU of 700 W. The W6800X Duo is a quad-slot card with no listed power connectors and a suggested PSU of 800 W. The bus interface differs fundamentally: the V620 uses PCIe 4.0 x16, while the W6800X Duo uses Apple MPX. The V620 has no display outputs. The W6800X Duo has 1x HDMI 2.1 and 4x Thunderbolt. The V620 is 267 mm long, 120 mm tall, and 50 mm wide. The W6800X Duo shares the 267 mm length and 120 mm height, but width is not listed. The generation labels differ: the V620 is in the Radeon Pro Navi (Navi II Series), while the W6800X Duo is in the Radeon Pro Mac (Navi II Series). The V620 lists a predecessor of Radeon Pro Vega; the W6800X Duo has no predecessor listed.

Where Each One Wins

The V620 wins decisively in the two shared benchmark categories. In Geekbench OpenCL, it scores 128,580 against 124,335 for the W6800X Duo, a 3.4% advantage. In Geekbench Vulkan, the V620 scores 144,364 against 125,622, a 14.9% lead. These are the only two head-to-head benchmarks listed, and the V620 wins both. The winsA count is 2, and winsB is 0.

The W6800X Duo has one exclusive benchmark result: Geekbench Metal, where it scores 157,365. This is its highest single score across all tests, and it is higher than either of the V620’s scores. The V620 has no Metal result in the FACT PACK. This suggests a specialization: the W6800X Duo appears optimized for Metal-based workloads, which aligns with its Apple MPX bus interface and its generation label of Radeon Pro Mac. The V620, with no display outputs and a standard PCIe interface, appears designed for server or rack-mounted compute tasks where display output is irrelevant.

The percentile ranking is identical: both sit at the 96th percentile versus all GPUs. The average benchmark scores are close, with the V620 at 136,472 and the W6800X Duo at 135,774. The deltaPct between them is 0.5% in favor of the V620 when viewed from the V620’s perspective, and -0.5% when viewed from the W6800X Duo’s perspective. In the nearestRivals list for the V620, the W6800X Duo is the closest competitor. For the W6800X Duo, the closest competitor is the AMD Radeon PRO W6800, with a deltaPct of 0.3%. The NVIDIA A10M and NVIDIA RTX 4000 Ada Generation are also within 0.9% of both cards. The data shows a tight cluster at the top of this performance tier.

FAQ

Q: Which card is faster on average?

A: The AMD Radeon PRO V620 has a higher average benchmark score of 136,472, compared to 135,774 for the AMD Radeon Pro W6800X Duo. The deltaPct is 0.5% in favor of the V620.

Q: Does the W6800X Duo win any benchmark?

A: The W6800X Duo has a Geekbench Metal score of 157,365, which is its only listed result in that test. The V620 has no Metal benchmark listed. In the two shared tests, the W6800X Duo does not win.

Q: What is the largest performance gap between the two cards?

A: The largest gap is in Geekbench Vulkan, where the V620 scores 144,364 versus 125,622 for the W6800X Duo, a 14.9% advantage. The OpenCL gap is smaller, at 3.4%.

Q: Are these cards the same architecture?

A: Yes. Both use the Navi 21 chip, RDNA 2.0 architecture, a 7 nm process from TSMC, and have identical transistor counts of 26,800 million and die sizes of 520 mm².

Q: Do the cards have the same memory configuration?

A: Yes. Both have 32 GB of GDDR6 on a 256-bit bus, with 2000 MHz memory clock and 16 Gbps effective speed, yielding 512.0 GB/s of bandwidth.

Q: Which card has more compute units?

A: The V620 has more of every execution unit type. It has 4,608 shading units, 288 TMUs, 128 ROPs, and 72 RT cores. The W6800X Duo has 3,840 shading units, 240 TMUs, 96 ROPs, and 60 RT cores.

Q: What are the power requirements?

A: The V620 has a TDP of 300 W and a suggested PSU of 700 W, with 2x 8-pin connectors. The W6800X Duo has a TDP of 400 W and a suggested PSU of 800 W, with no listed power connectors.

Q: Is the W6800X Duo a Mac-specific card?

A: The data indicates so. Its bus interface is Apple MPX, its generation is Radeon Pro Mac, and it has 1x HDMI 2.1 and 4x Thunderbolt outputs. The V620 has no display outputs and uses PCIe 4.0 x16.

Head-to-Head Benchmarks

The two shared benchmarks tell a consistent story, but the magnitude differs. In Geekbench OpenCL, the V620 posts 128,580 against the W6800X Duo’s 124,335. The deltaPct is 3.4%. This is a solid but not overwhelming lead. The V620’s higher shading unit count and boost clock likely contribute to this edge. The W6800X Duo’s lower boost clock of 1967 MHz versus 2200 MHz puts it at a disadvantage in raw compute throughput. The FP32 ratings reflect this: the V620 delivers 20.28 TFLOPS, while the W6800X Duo delivers 15.11 TFLOPS. The FP16 numbers are similarly lopsided, with the V620 at 40.55 TFLOPS (2:1) and the W6800X Duo at 30.21 TFLOPS (2:1).

In Geekbench Vulkan, the gap widens considerably. The V620 scores 144,364, while the W6800X Duo scores 125,622. That is a 14.9% delta. This is the largest single difference between the two cards. The Vulkan result suggests that the V620’s architecture is better suited to this API, or that the W6800X Duo’s drivers are less optimized for Vulkan. The pixel rate and texture rate numbers support the V620’s advantage: it has a pixel rate of 281.6 GPixel/s and a texture rate of 633.6 GTexel/s, versus 188.8 GPixel/s and 472.1 GTexel/s for the W6800X Duo. These are substantial differences in raw throughput capabilities.

The W6800X Duo’s sole bright spot is the Geekbench Metal score of 157,365. This is not directly comparable to either of the V620’s scores, as the V620 has no Metal result. However, it is worth remembering the W6800X Duo’s Metal score is higher than its own OpenCL score (124,335) by 26.6% and its Vulkan score (125,622) by 25.3%. This indicates that the W6800X Duo performs significantly better under Metal than under other APIs. The V620’s best score is 144,364 in Vulkan, which is below the W6800X Duo’s Metal score. If a workload is Metal-specific, the W6800X Duo would be the choice based on the available data.

Specification Differences

The two cards differ in several key fields. The shading unit count is 4,608 for the V620 and 3,840 for the W6800X Duo. TMUs are 288 versus 240. ROPs are 128 versus 96. RT cores are 72 versus 60. Base clock is 1825 MHz versus 1800 MHz. Boost clock is 2200 MHz versus 1967 MHz. Pixel rate is 281.6 GPixel/s versus 188.8 GPixel/s. Texture rate is 633.6 GTexel/s versus 472.1 GTexel/s. FP32 is 20.28 TFLOPS versus 15.11 TFLOPS. FP16 is 40.55 TFLOPS (2:1) versus 30.21 TFLOPS (2:1). TDP is 300 W versus 400 W. Slot width is dual-slot versus quad-slot. Power connectors are 2x 8-pin versus none listed. Suggested PSU is 700 W versus 800 W. Bus interface is PCIe 4.0 x16 versus Apple MPX. Display outputs are none versus 1x HDMI 2.1 and 4x Thunderbolt. Width is 50 mm versus not listed. Generation is Radeon Pro Navi (Navi II Series) versus Radeon Pro Mac (Navi II Series). Release date is 2021-11-03 versus 2021-08-02. The W6800X Duo has a launch MSRP of 4,999 USD; the V620 has none listed.

The memory configuration is identical, as are the chip, architecture, process node, foundry, transistors, die size, transistor density, and memory clock. The APIS are the same: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Both cards are end-of-life. The W6800X Duo has no predecessor listed, while the V620 lists Radeon Pro Vega as its predecessor. Neither has a successor listed.