AMD Radeon Pro W6800X Duo vs NVIDIA CMP 90HX Comparison

The Verdict

The AMD Radeon Pro W6800X Duo and the NVIDIA CMP 90HX occupy completely different corners of the GPU market, and the benchmark data reflects that divergence. The AMD card is the clear performance leader, posting an average benchmark score of 135,774 against the NVIDIA card's 69,000. That places the W6800X Duo in the 96th percentile of all GPUs, while the CMP 90HX sits in the 90th percentile.

For anyone whose workload involves Metal, OpenCL, or Vulkan compute tasks, the AMD Radeon Pro W6800X Duo is the only sensible choice. Its Geekbench Metal score of 157,365 and Vulkan score of 125,622 are simply in another league. The NVIDIA CMP 90HX, by contrast, is a mining-oriented card with no display outputs, no Vulkan benchmark recorded, and only a single OpenCL score of 69,000 to its name. The data shows an 80.2% advantage for the AMD card in the only head-to-head test available, Geekbench OpenCL.

The CMP 90HX does have one advantage: its architecture supports tensor cores, 200 of them, which the AMD card lacks entirely. But with no benchmark data to quantify that advantage, and given that the card cannot even output a display signal, its practical appeal is extremely narrow. The verdict is straightforward: pick the W6800X Duo for any general-purpose compute, workstation, or content creation role. Pick the CMP 90HX only if the specific task is mining and the lack of display outputs is not a limitation.

Architecture Differences

The two GPUs come from different manufacturers, different foundries, and different design philosophies. The AMD Radeon Pro W6800X Duo uses the Navi 21 chip built on RDNA 2.0 architecture, manufactured by TSMC on a 7 nm process. It packs 26,800 million transistors onto a 520 mm² die, resulting in a transistor density of 51.5M per mm². The NVIDIA CMP 90HX uses the GA102 chip built on Ampere architecture, manufactured by Samsung on an 8 nm process. It contains 28,300 million transistors across a larger 628 mm² die, but its transistor density is lower at 45.1M per mm².

The AMD card offers 3,840 shading units, 240 texture mapping units, and 96 render output units. It also includes 60 ray tracing cores, but no tensor cores. The NVIDIA card counters with 6,400 shading units, 200 TMUs, and 80 ROPs, along with 50 ray tracing cores and 200 tensor cores. The shading unit advantage for NVIDIA is substantial, nearly 67% more, but the AMD card compensates with higher clock speeds and a different compute layout.

Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so API compatibility is identical. The production status for both is end-of-life, and their release dates are nearly the same, the AMD card on August 2, 2021, and the NVIDIA card on July 27, 2021. The AMD card belongs to the Radeon Pro Mac generation, while the NVIDIA card belongs to the Mining GPUs generation, which explains their divergent feature sets.

Head-to-Head Benchmarks

The only direct benchmark comparison in the database is Geekbench OpenCL, and it is a decisive victory for the AMD Radeon Pro W6800X Duo. The AMD card scores 124,335, while the NVIDIA CMP 90HX scores 69,000. That is an 80.2% delta, meaning the AMD card delivers nearly double the OpenCL compute performance of the NVIDIA card. This is the single most important number in the entire comparison, as it establishes the performance hierarchy with no ambiguity.

Beyond the head-to-head, the AMD card's other benchmark scores reinforce its dominance. Its Geekbench Metal score of 157,365 is the highest recorded metric for either card, and its Vulkan score of 125,622 is nearly identical to its OpenCL result. The NVIDIA card has no Metal or Vulkan scores recorded, so the AMD card wins those categories by default. In the broader database context, the AMD card's average score of 135,774 puts it 0.3% ahead of the AMD Radeon PRO W6800, 0.4% ahead of the NVIDIA A10M, and 0.4% ahead of the NVIDIA RTX 4000 Ada Generation. It trails the AMD Radeon PRO V620 by 0.5%, but that gap is trivial.

The NVIDIA CMP 90HX, with its average score of 69,000, sits just 0.3% ahead of the Intel Arc A770 and 0.6% ahead of the AMD Radeon Instinct MI25. It trails the AMD Radeon Pro WX 8200 by 1.2% and the NVIDIA Quadro P6000 by 1.4%. In other words, the CMP 90HX performs at roughly the level of a mid-range card from several generations ago, while the W6800X Duo competes with modern workstation flagships.

Specification Differences

The specification sheet tells the story of two cards designed for entirely different purposes. The AMD Radeon Pro W6800X Duo uses 32 GB of GDDR6 memory on a 256-bit bus, delivering 512.0 GB/s of bandwidth. The NVIDIA CMP 90HX uses 10 GB of GDDR6X memory on a 320-bit bus, delivering 760.3 GB/s of bandwidth. The NVIDIA card has the bandwidth advantage, 48.6% more, but the AMD card has more than three times the memory capacity.

Clock speeds favor the AMD card. Its base clock is 1,800 MHz with a boost of 1,967 MHz, compared to the NVIDIA card's 1,500 MHz base and 1,710 MHz boost. Memory clocks also differ, with the AMD card running at 2,000 MHz with 16 Gbps effective speed, while the NVIDIA card runs at 1,188 MHz with 19 Gbps effective speed. The higher memory transfer rate for NVIDIA compensates for its lower raw clock.

Compute throughput numbers are mixed. The NVIDIA card has the higher FP32 rating at 21.89 TFLOPS, compared to the AMD card's 15.11 TFLOPS. However, the AMD card achieves 30.21 TFLOPS in FP16 with a 2:1 ratio, while the NVIDIA card is locked at 21.89 TFLOPS in FP16 with a 1:1 ratio. Pixel fill rate favors AMD at 188.8 GPixel/s versus 136.8 GPixel/s, and texture fill rate also favors AMD at 472.1 GTexel/s versus 342.0 GTexel/s.

Power and physical specifications differ substantially. The AMD card has a TDP of 400 W and requires an 800 W power supply, while the NVIDIA card has a TDP of 320 W and requires a 700 W PSU. The AMD card is a quad-slot design measuring 267 mm in length and 120 mm in height, while the NVIDIA card is a dual-slot design measuring 285 mm in length and 112 mm in height. The NVIDIA card uses two 8-pin power connectors, while the AMD card's connector type is not specified in the database.

The bus interfaces could not be more different. The AMD card uses Apple MPX, which explains its "Radeon Pro Mac" generation label, and it offers display outputs including one HDMI 2.1 and four Thunderbolt ports. The NVIDIA card uses a PCIe 1.0 x4 interface, a severely limited bus for a modern GPU, and it has no display outputs at all. This is the defining specification difference: one card is a professional workstation GPU with full display capabilities, the other is a mining card with zero display outputs.

FAQ

Q: Which card has better overall compute performance?

A: The AMD Radeon Pro W6800X Duo. Its average benchmark score is 135,774, which is 96.8% higher than the NVIDIA CMP 90HX's 69,000. In the direct OpenCL comparison, the AMD card leads by 80.2%.

Q: Does the NVIDIA CMP 90HX support display output?

A: No. The NVIDIA CMP 90HX has no display outputs whatsoever. The AMD Radeon Pro W6800X Duo, by contrast, provides one HDMI 2.1 port and four Thunderbolt ports.

Q: How do the two cards compare in memory capacity and bandwidth?

A: The AMD card has 32 GB of GDDR6 memory with 512.0 GB/s bandwidth on a 256-bit bus. The NVIDIA card has 10 GB of GDDR6X memory with 760.3 GB/s bandwidth on a 320-bit bus. The NVIDIA card has higher bandwidth, but the AMD card has 220% more memory capacity.

Q: Which card has the higher FP32 compute throughput?

A: The NVIDIA CMP 90HX. It delivers 21.89 TFLOPS of FP32 performance, compared to 15.11 TFLOPS for the AMD Radeon Pro W6800X Duo. However, the AMD card achieves 30.21 TFLOPS in FP16, while the NVIDIA card remains at 21.89 TFLOPS.

Q: What are the power requirements for each card?

A: The AMD Radeon Pro W6800X Duo has a 400 W TDP and requires an 800 W power supply. The NVIDIA CMP 90HX has a 320 W TDP and requires a 700 W power supply. The NVIDIA card is also physically smaller, taking up two slots instead of four.

Q: Which card is better for gaming or general workstation use?

A: The AMD Radeon Pro W6800X Duo is the only viable option for general use. It has display outputs, supports DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4, and delivers far higher benchmark scores. The NVIDIA CMP 90HX cannot output video at all, making it unsuitable for any interactive workload.