GPU Comparison

The NVIDIA CMP 90HX and AMD Radeon RX 6950 XT occupy opposite ends of the GPU spectrum: one is a mining-only accelerator with no display outputs, the other a fully featured flagship gaming card. The benchmark data available shows a single direct comparison, and the results are decisively one-sided. In the Geekbench OpenCL test, the AMD Radeon RX 6950 XT scores 205,998, while the NVIDIA CMP 90HX manages 69,000. That translates to a 66.5% advantage for the AMD card, making it the clear winner in raw compute performance as measured by this specific workload. The CMP 90HX holds a percentile ranking of 90 among all GPUs, while the RX 6950 XT sits at 88, but that ranking is based on an average benchmark score of 58,392 for the AMD card versus 69,000 for the NVIDIA card, a discrepancy that reflects the different benchmark suites each card was tested under.

Head-to-Head Benchmarks

The only shared benchmark in the data is Geekbench OpenCL, and it is not close. The AMD Radeon RX 6950 XT delivers 205,998 points, which is 136,998 points higher than the NVIDIA CMP 90HX’s 69,000. The delta percentage of -66.5% confirms that the NVIDIA card is performing at roughly one-third the level of the AMD card in this test. For context, the CMP 90HX’s nearest rivals in the overall database include the Intel Arc A770 at 68,809 (a 0.3% difference) and the AMD Radeon Pro WX 8200 at 69,870 (a -1.2% difference). The RX 6950 XT, by contrast, sits near the AMD Radeon PRO V710 at 58,657 (-0.5%) and the NVIDIA P102-100 at 58,528 (-0.2%). These rival comparisons reinforce the gulf between the two cards: the CMP 90HX is clustered with mid-range workstation and consumer GPUs, while the RX 6950 XT’s OpenCL score places it in a different performance tier entirely, even though its average score across all benchmarks is lower due to the inclusion of lighter DirectX 9 and DirectX 10 tests.

Beyond the single head-to-head result, the RX 6950 XT shows a broader benchmark profile. It scores 222,653 in Geekbench Metal, 165,212 in Geekbench Vulkan, and 28,070 in Passmark G3D. Its Passmark GPU Compute score is 14,199, and it handles DirectX 11 with a score of 300 and DirectX 9 with 303. The CMP 90HX has no other benchmark entries in the data, so its capabilities outside of OpenCL cannot be assessed from this pack. The practical takeaway is that for any compute workload represented by OpenCL, the RX 6950 XT is the overwhelming choice. The NVIDIA card’s 90th percentile ranking does not translate to real-world dominance here; it merely reflects a limited test set.

Architecture Differences

The two cards are built on fundamentally different architectures and process nodes. The NVIDIA CMP 90HX uses the GA102 chip on the Ampere architecture, fabricated by Samsung on an 8 nm process. It packs 28,300 million transistors into a 628 mm² die, resulting in a transistor density of 45.1 million per square millimeter. The AMD Radeon RX 6950 XT uses the Navi 21 chip on the RDNA 2.0 architecture, built by TSMC on a 7 nm process. It contains 26,800 million transistors on a smaller 520 mm² die, achieving a higher density of 51.5 million per square millimeter. This means the AMD card packs more transistors per area despite having fewer total transistors, a direct consequence of the more advanced 7 nm node.

Core configurations differ substantially. The CMP 90HX has 6,400 shading units, 200 texture mapping units, and 80 ROPs, along with 50 ray tracing cores and 200 tensor cores. The RX 6950 XT has fewer shading units at 5,120, but more TMUs at 320 and more ROPs at 128. It also has 80 ray tracing cores and no tensor cores at all, reflecting its gaming-first design. Clock speeds tell the rest of the story: the NVIDIA card runs at a 1500 MHz base and 1710 MHz boost, while the AMD card boosts to 2310 MHz with a 2100 MHz game clock and 1860 MHz base. The higher clocks on the AMD card contribute to its superior fill rates, 295.7 GPixel/s versus 136.8 GPixel/s, and texture rates of 739.2 GTexel/s versus 342.0 GTexel/s.

Memory architecture is another clear divergence. The CMP 90HX features 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s of bandwidth. The RX 6950 XT offers 16 GB of GDDR6 on a narrower 256-bit bus, yielding 576.0 GB/s. Despite the bandwidth deficit, the AMD card’s larger frame buffer is more suited to modern gaming and content creation. The NVIDIA card’s FP32 throughput is 21.89 TFLOPS, while the AMD card reaches 23.65 TFLOPS. FP16 performance is where they split: the CMP 90HX does 21.89 TFLOPS at a 1:1 ratio, but the RX 6950 XT hits 47.31 TFLOPS at a 2:1 ratio, doubling its FP16 output.

FAQ

Q: Which card has a higher boost clock?

A: The AMD Radeon RX 6950 XT boosts to 2310 MHz, significantly higher than the NVIDIA CMP 90HX’s 1710 MHz boost clock.

Q: How much memory does each card have, and what type?

A: The NVIDIA CMP 90HX has 10 GB of GDDR6X, while the AMD Radeon RX 6950 XT has 16 GB of GDDR6.

Q: Does the CMP 90HX support display outputs?

A: No. The NVIDIA CMP 90HX has no display outputs, making it unsuitable for gaming or any visual output tasks.

Q: What is the transistor count on each chip?

A: The CMP 90HX’s GA102 chip contains 28,300 million transistors, while the RX 6950 XT’s Navi 21 chip has 26,800 million transistors.

Q: Which card has a higher FP16 throughput?

A: The AMD Radeon RX 6950 XT achieves 47.31 TFLOPS FP16, compared to 21.89 TFLOPS on the NVIDIA CMP 90HX.

Q: Are both cards end-of-life products?

A: Yes, both the NVIDIA CMP 90HX and the AMD Radeon RX 6950 XT are listed as end-of-life in the production status.

Specification Differences

The two cards differ in nearly every measurable specification. The process node is 8 nm for the NVIDIA card versus 7 nm for the AMD card. Transistor density favors AMD at 51.5M / mm² versus 45.1M / mm². Die size is 628 mm² for NVIDIA and 520 mm² for AMD. The CMP 90HX has a base clock of 1500 MHz and boost of 1710 MHz; the RX 6950 XT has a base of 1860 MHz, a game clock of 2100 MHz, and a boost of 2310 MHz. Memory speed is 1188 MHz (19 Gbps effective) on the NVIDIA card versus 2250 MHz (18 Gbps effective) on the AMD card.

Shading units: 6,400 versus 5,120. TMUs: 200 versus 320. ROPs: 80 versus 128. Ray tracing cores: 50 versus 80. Tensor cores: 200 versus none. Pixel rate: 136.8 GPixel/s versus 295.7 GPixel/s. Texture rate: 342.0 GTexel/s versus 739.2 GTexel/s. FP32: 21.89 TFLOPS versus 23.65 TFLOPS. FP16: 21.89 TFLOPS (1:1) versus 47.31 TFLOPS (2:1). TDP: 320 W versus 335 W. Slot width: dual-slot versus triple-slot. Bus interface: PCIe 1.0 x4 versus PCIe 4.0 x16. Display outputs: none versus 1x HDMI 2.1 and 2x DisplayPort 1.4a. Dimensions: 285 mm long and 112 mm tall versus 267 mm long, 120 mm tall, and 50 mm wide. The RX 6950 XT has a launch MSRP of 1,099 USD; the CMP 90HX has no launch MSRP listed. Both share the same DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4 API support, and both require a 700 W suggested PSU with 2x 8-pin power connectors.

Where Each One Wins

The AMD Radeon RX 6950 XT wins the only direct benchmark comparison, taking the Geekbench OpenCL test by a 66.5% margin. It also wins on every performance-related metric: higher boost clock, higher pixel and texture rates, higher FP32 and FP16 throughput, and more ROPs. Its 16 GB memory capacity and PCIe 4.0 x16 interface make it suitable for modern gaming and general compute tasks, and its display outputs allow it to function as a standard graphics card. The data shows it is the superior choice for any workload that relies on OpenCL, Metal, Vulkan, or DirectX performance, as evidenced by its strong scores across multiple benchmark suites.

The NVIDIA CMP 90HX wins on transistor count, with 28,300 million versus 26,800 million, and on memory bandwidth, at 760.3 GB/s versus 576.0 GB/s. It also has more shading units (6,400 versus 5,120) and tensor cores (200 versus none), which could theoretically aid in AI workloads, though no benchmark data in the pack confirms this. Its 90th percentile ranking versus the AMD card’s 88th percentile suggests it performs well in its limited test set, but the absence of any other benchmark results makes it impossible to verify broader strengths. The card’s lack of display outputs and PCIe 1.0 x4 interface severely limit its practical applications to dedicated mining or compute setups.

The Verdict

The data is unambiguous: the AMD Radeon RX 6950 XT is the better GPU by a wide margin in the one test where they are directly compared. Its Geekbench OpenCL score of 205,998 dwarfs the NVIDIA CMP 90HX’s 69,000, and it leads in almost every architectural specification that matters for real-world performance, higher clocks, more ROPs, higher fill rates, and superior FP16 throughput. The RX 6950 XT also brings 16 GB of memory, display outputs, and a modern PCIe 4.0 x16 interface, making it a functional graphics card for gaming and productivity. The CMP 90HX, with its 10 GB memory, no outputs, and legacy PCIe 1.0 x4 bus, is a niche product designed for a single purpose. Its higher memory bandwidth and extra shading units do not compensate for the massive performance gap. For anyone choosing between these two, the RX 6950 XT is the rational pick based on benchmark results and feature set. The CMP 90HX’s only advantages are its lower TDP of 320 W versus 335 W and its higher transistor count, but neither translates into a competitive performance advantage in the available data.