AMD Radeon Pro W6600X vs NVIDIA CMP 90HX Comparison

The AMD Radeon Pro W6600X and the NVIDIA CMP 90HX are two end-of-life graphics cards with fundamentally different design goals. The database records a single benchmark for each: the W6600X posts a Geekbench Metal score of 107342, while the CMP 90HX records a Geekbench OpenCL score of 69000. These results come from different API tests, so direct comparison requires careful interpretation, but the surrounding specifications and rival data provide a clear picture of where each card stands in its respective context.

FAQ

Q: What is the recorded benchmark score for the AMD Radeon Pro W6600X?

A: The AMD Radeon Pro W6600X records a Geekbench Metal score of 107342, placing it in the 94th percentile of all GPUs in the database.

Q: What is the recorded benchmark score for the NVIDIA CMP 90HX?

A: The NVIDIA CMP 90HX records a Geekbench OpenCL score of 69000, placing it in the 90th percentile of all GPUs in the database.

Q: How does the W6600X compare to its nearest rivals?

A: The W6600X is 0.6% ahead of the AMD Radeon Pro Vega II Duo, 5.4% ahead of the NVIDIA Quadro RTX 6000, and 2.1% behind the AMD Radeon Pro Vega II, with a 3.1% deficit versus the AMD Radeon PRO W7900.

Q: How does the CMP 90HX compare to its nearest rivals?

A: The CMP 90HX is 0.3% ahead of the Intel Arc A770, 0.6% ahead of the AMD Radeon Instinct MI25, and 1.2% behind the AMD Radeon Pro WX 8200, with a 1.4% deficit versus the NVIDIA Quadro P6000.

Q: Which card has a higher transistor count?

A: The NVIDIA CMP 90HX has 28,300 million transistors on a 628 mm² die, while the AMD Radeon Pro W6600X has 11,060 million transistors on a 237 mm² die.

Q: Do both cards have display outputs?

A: No, both the AMD Radeon Pro W6600X and the NVIDIA CMP 90HX have no display outputs, making them unsuitable for direct display connection.

Where Each One Wins

The AMD Radeon Pro W6600X wins in the recorded Metal benchmark, scoring 107342, which is significantly higher than the CMP 90HX’s OpenCL score of 69000. However, this is not an apples-to-apples comparison because the tests use different APIs. The W6600X’s score places it in the 94th percentile, while the CMP 90HX sits in the 90th percentile, so the W6600X ranks higher in overall GPU performance distribution.

In terms of compute throughput, the CMP 90HX has a clear advantage in raw FP32 performance, delivering 21.89 TFLOPS versus the W6600X’s 10.15 TFLOPS. The CMP 90HX also leads in texture fill rate (342.0 GTexel/s versus 317.3 GTexel/s), memory bandwidth (760.3 GB/s versus 256.0 GB/s), and memory capacity (10 GB versus 8 GB). These figures suggest the CMP 90HX is better suited for workloads that stress raw parallel compute and memory throughput, such as OpenCL-based tasks.

The W6600X, meanwhile, wins in pixel rate (158.7 GPixel/s versus 136.8 GPixel/s) and has a higher boost clock (2479 MHz versus 1710 MHz). It also has a much lower power draw, with a 120 W TDP compared to the CMP 90HX’s 320 W, and a lower suggested PSU requirement (300 W versus 700 W). For users prioritizing efficiency or systems with limited power delivery, the W6600X is the more practical choice.

Architecture Differences

The AMD Radeon Pro W6600X is built on the Navi 23 chip using RDNA 2.0 architecture, manufactured on a 7 nm process at TSMC. It integrates 11,060 million transistors on a 237 mm² die, resulting in a transistor density of 46.7M per mm². The architecture includes 2048 shading units, 128 texture mapping units, 64 render output units, and 32 ray tracing cores. It does not list tensor cores. The card supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The NVIDIA CMP 90HX uses the GA102 chip with Ampere architecture, built on an 8 nm process at Samsung. It packs 28,300 million transistors on a 628 mm² die, with a transistor density of 45.1M per mm². The card features 6400 shading units, 200 texture mapping units, 80 render output units, 50 ray tracing cores, and 200 tensor cores. It supports the same API versions: DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

The key architectural distinction is the presence of tensor cores in the CMP 90HX, which are absent in the W6600X. Additionally, the CMP 90HX has over three times the shading units and nearly double the ray tracing cores. The process nodes differ, with the W6600X using a smaller 7 nm node versus the CMP 90HX’s 8 nm node, although the CMP 90HX compensates with a much larger die and higher transistor count.

Specification Differences

The two cards differ across nearly every major specification. The W6600X has a base clock of 2068 MHz and a boost clock of 2479 MHz, while the CMP 90HX operates at 1500 MHz base and 1710 MHz boost. Memory configurations are starkly different: the W6600X uses 8 GB of GDDR6 on a 128-bit bus with 256.0 GB/s bandwidth, whereas the CMP 90HX uses 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth.

Compute resources show the CMP 90HX’s scale: 6400 shading units versus 2048, 200 TMUs versus 128, and 80 ROPs versus 64. The CMP 90HX also has 50 RT cores and 200 tensor cores, while the W6600X has 32 RT cores and no tensor cores. FP32 throughput is 21.89 TFLOPS for the CMP 90HX versus 10.15 TFLOPS for the W6600X, and FP16 performance is 21.89 TFLOPS for the CMP 90HX versus 20.31 TFLOPS for the W6600X.

Power and physical specifications also diverge. The W6600X has a 120 W TDP and a suggested PSU of 300 W, while the CMP 90HX has a 320 W TDP and a suggested PSU of 700 W. The W6600X uses an Apple MPX bus interface, whereas the CMP 90HX uses PCIe 1.0 x4. The CMP 90HX has defined dimensions of 285 mm length and 112 mm height, while the W6600X lists no dimensions. The W6600X has a launch MSRP of 699 USD, while the CMP 90HX has no recorded launch MSRP.

Head-to-Head Benchmarks

The database contains no direct head-to-head benchmark entries between these two cards, and each has a single recorded score from different tests. The W6600X’s Geekbench Metal score of 107342 is substantially higher than the CMP 90HX’s Geekbench OpenCL score of 69000, but this comparison is limited by the different APIs. Within their own benchmark categories, the W6600X ranks 4 percentage points higher in percentile terms (94th versus 90th).

The W6600X’s nearest rival comparison shows it outperforms the AMD Radeon Pro Vega II Duo by 0.6% and the NVIDIA Quadro RTX 6000 by 5.4%, while trailing the AMD Radeon Pro Vega II by 2.1% and the AMD Radeon PRO W7900 by 3.1%. The CMP 90HX’s rival data shows it beats the Intel Arc A770 by 0.3% and the AMD Radeon Instinct MI25 by 0.6%, but falls behind the AMD Radeon Pro WX 8200 by 1.2% and the NVIDIA Quadro P6000 by 1.4%.

In raw compute terms, the CMP 90HX delivers more than double the FP32 throughput of the W6600X (21.89 TFLOPS versus 10.15 TFLOPS), and its memory bandwidth is nearly three times higher (760.3 GB/s versus 256.0 GB/s). The W6600X counters with a higher pixel rate (158.7 GPixel/s versus 136.8 GPixel/s) and a higher boost clock (2479 MHz versus 1710 MHz). These numbers indicate that the CMP 90HX is the stronger card for compute-heavy, memory-intensive tasks, while the W6600X excels in rasterization-focused metrics and efficiency.

The Verdict

The data points to a clear split: the AMD Radeon Pro W6600X and NVIDIA CMP 90HX serve different purposes, despite both lacking display outputs. The W6600X is positioned for users who need a high-performing card in Metal-based environments, as evidenced by its 94th percentile ranking and its strong showing against rivals like the NVIDIA Quadro RTX 6000. Its lower TDP of 120 W and 300 W suggested PSU make it suitable for systems with modest power budgets.

The NVIDIA CMP 90HX, with its 90th percentile ranking and OpenCL score of 69000, is geared toward compute workloads that leverage its 6400 shading units, 200 tensor cores, and 760.3 GB/s memory bandwidth. Its 10 GB GDDR6X memory and 320-bit bus give it a significant advantage in data-heavy tasks, and its FP32 throughput of 21.89 TFLOPS is more than double the W6600X’s 10.15 TFLOPS.

For users running Metal-based applications, the W6600X is the better choice, based on its higher recorded score and percentile. For users prioritizing raw compute performance and memory throughput, the CMP 90HX is the stronger candidate. The W6600X also offers a launch MSRP of 699 USD, which is the only pricing data available. Neither card supports display output, so both are strictly for compute or rendering workloads where external display connectivity is not required. Ultimately, the selection depends on the software environment and workload type, with the W6600X leading in Metal and the CMP 90HX leading in OpenCL-centric compute tasks.