AMD Radeon RX 6850M XT vs NVIDIA CMP 90HX Comparison

The AMD Radeon RX 6850M XT and NVIDIA CMP 90HX are both end-of-life GPUs from the previous generation, but they target completely different use cases. The RX 6850M XT is a mobile gaming processor, while the CMP 90HX is a mining-specific card with no display outputs. The benchmark data shows a clear overall winner, but the specifications reveal why each product exists. The RX 6850M XT leads in the only shared benchmark, while the CMP 90HX counters with massive compute resources and memory bandwidth that its mining workload demands.

Head-to-Head Benchmarks

The single overlapping benchmark is Geekbench OpenCL, and the AMD Radeon RX 6850M XT wins decisively. The RX 6850M XT scores 85,040, while the NVIDIA CMP 90HX scores 69,000. This represents a 23.2% advantage for the AMD part. In this test, the RX 6850M XT outperforms the CMP 90HX by 16,040 points, a substantial margin that reflects the AMD chip’s higher per-clock efficiency and its 2,463 MHz game clock versus the NVIDIA card’s 1,710 MHz boost clock.

The RX 6850M XT holds a 92nd percentile ranking among all GPUs, while the CMP 90HX sits at the 90th percentile. The delta between these two cards is significant in context: the RX 6850M XT’s average benchmark score is 78,940, while the CMP 90HX averages 69,000. The AMD part also has a richer benchmark history, with scores in 3DMark Steel Nomad DX12 (2,257), Geekbench Metal (128,981), Geekbench OpenCL (85,040), and Geekbench Vulkan (99,483). The CMP 90HX has only the single OpenCL result, making its performance profile far less complete.

When placed against their nearest rivals, the two cards occupy different performance tiers. The RX 6850M XT’s closest competitor is the NVIDIA Tesla P100 PCIe 12 GB, which scores 79,396, a mere 0.6% difference. The CMP 90HX’s nearest rival is the Intel Arc A770 at 68,809, just 0.3% away. This means the RX 6850M XT sits in a higher absolute performance bracket, roughly 14% above the CMP 90HX in average score, even though both are in the 90th percentile.

Architecture Differences

The two GPUs come from fundamentally different architectures. The AMD Radeon RX 6850M XT uses the Navi 22 chip built on RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC. The NVIDIA CMP 90HX uses the GA102 chip based on Ampere architecture, fabricated on an 8 nm process at Samsung. These process differences are stark: the AMD chip packs 17,200 million transistors into a 335 mm² die, yielding a transistor density of 51.3 million per mm². The NVIDIA chip has 28,300 million transistors on a 628 mm² die, with a lower density of 45.1 million per mm². The AMD part is smaller and denser, while the NVIDIA part is nearly twice the die size.

The CMP 90HX compensates with raw resources. It has 6,400 shading units, 200 texture mapping units, 80 ROPs, 50 ray tracing cores, and 200 tensor cores. The RX 6850M XT has 2,560 shading units, 160 TMUs, 64 ROPs, and 40 ray tracing cores, with no tensor cores. This is a 2.5x difference in shading units, yet the AMD part still wins the OpenCL benchmark. The NVIDIA card’s FP32 throughput is 21.89 TFLOPS, while the AMD card delivers 13.21 TFLOPS. However, the FP16 numbers reveal a key difference: the RX 6850M XT achieves 26.43 TFLOPS FP16 via a 2:1 ratio, while the CMP 90HX achieves only 21.89 TFLOPS at a 1:1 ratio.

Memory architecture diverges significantly. The RX 6850M XT has 12 GB of GDDR6 on a 192-bit bus, delivering 432.0 GB/s bandwidth at 18 Gbps effective. The CMP 90HX has 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s at 19 Gbps. The NVIDIA card has 76% more memory bandwidth, which is critical for mining workloads but less relevant for the compute tasks measured in the shared benchmark. The RX 6850M XT also has a higher pixel rate at 165.2 GPixel/s versus 136.8 GPixel/s, but a lower texture rate at 413.0 GTexel/s versus 342.0 GTexel/s.

Power and physical characteristics diverge sharply. The RX 6850M XT has a 165 W TDP and is an IGP (integrated graphics processor) with no power connectors, designed for laptops. The CMP 90HX has a 320 W TDP, is dual-slot, requires two 8-pin power connectors, and needs a 700 W suggested PSU. The CMP 90HX measures 285 mm in length and 112 mm in height. The RX 6850M XT has no display outputs (portable device dependent), while the CMP 90HX has no outputs at all, as it is mining-only.

Where Each One Wins

The RX 6850M XT wins the only directly comparable performance metric, the Geekbench OpenCL test, by 23.2%. This indicates its architecture is more efficient at general compute workloads despite having fewer shading units and lower raw TFLOPS. The AMD card also has a broader benchmark portfolio, including a 3DMark Steel Nomad DX12 score of 2,257, which the CMP 90HX lacks entirely, suggesting it cannot meaningfully run modern gaming workloads. The RX 6850M XT’s 12 GB memory capacity is also larger, and its support for DirectX 12 Ultimate, OpenGL 4.6, and Vulkan 1.4 mirrors the CMP 90HX’s API support, but the AMD card adds portable device dependent display outputs, making it usable in a laptop.

The CMP 90HX wins in raw compute resources and memory bandwidth. Its 21.89 TFLOPS FP32 and FP16 throughput is 66% higher than the RX 6850M XT’s FP32 figure, and its 760.3 GB/s bandwidth is 76% higher. The 200 tensor cores and 50 ray tracing cores are absent from the AMD part. The CMP 90HX also has a higher shading unit count (6,400 vs 2,560), TMU count (200 vs 160), and ROP count (80 vs 64). For mining algorithms that rely on memory bandwidth and raw arithmetic, the CMP 90HX is purpose-built, but the benchmark data does not include any mining-specific tests to validate this advantage.

The RX 6850M XT also wins on power efficiency. It delivers a higher average benchmark score (78,940) at 165 W, while the CMP 90HX delivers 69,000 at 320 W. This means the AMD card produces 14.4% more performance per unit of power draw. The RX 6850M XT’s 7 nm process and lower transistor count contribute to this efficiency, while the CMP 90HX’s 8 nm process and 28,300 million transistors demand far more energy.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon RX 6850M XT has an average benchmark score of 78,940, while the NVIDIA CMP 90HX scores 69,000. This places the RX 6850M XT in the 92nd percentile and the CMP 90HX in the 90th percentile.

Q: How much faster is the RX 6850M XT in the shared OpenCL test?

A: The RX 6850M XT scores 85,040 in Geekbench OpenCL, which is 23.2% higher than the CMP 90HX’s 69,000. The absolute difference is 16,040 points.

Q: Does the CMP 90HX have any benchmark wins over the RX 6850M XT?

A: No. The head-to-head data shows the RX 6850M XT wins 1 benchmark and the CMP 90HX wins 0. The CMP 90HX has only one recorded benchmark score, while the RX 6850M XT has four.

Q: What memory configurations do these GPUs use?

A: The RX 6850M XT uses 12 GB of GDDR6 on a 192-bit bus with 432.0 GB/s bandwidth. The CMP 90HX uses 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth.

Q: Are these GPUs suitable for gaming?

A: The RX 6850M XT has a 3DMark Steel Nomad DX12 score of 2,257 and supports DirectX 12 Ultimate, indicating gaming capability. The CMP 90HX has no display outputs and no gaming benchmarks listed, making it unsuitable for gaming.

Q: What are the TDP requirements for each card?

A: The RX 6850M XT has a 165 W TDP and requires no power connectors. The CMP 90HX has a 320 W TDP, needs two 8-pin power connectors, and has a suggested PSU of 700 W.

Specification Differences

The two GPUs differ in nearly every specification category. The RX 6850M XT uses a 7 nm process from TSMC, while the CMP 90HX uses an 8 nm process from Samsung. The AMD chip has 17,200 million transistors on a 335 mm² die, while the NVIDIA chip has 28,300 million on a 628 mm² die. Transistor density is 51.3M/mm² for AMD versus 45.1M/mm² for NVIDIA.

Clock speeds differ substantially. The RX 6850M XT has a base clock of 2,321 MHz, a boost clock of 2,581 MHz, and a game clock of 2,463 MHz, with memory at 2,250 MHz (18 Gbps effective). The CMP 90HX has a base clock of 1,500 MHz and a boost clock of 1,710 MHz, with memory at 1,188 MHz (19 Gbps effective). The AMD card runs over 800 MHz higher at boost.

Memory capacity and type differ: 12 GB GDDR6 versus 10 GB GDDR6X. Bus width is 192-bit versus 320-bit, and bandwidth is 432.0 GB/s versus 760.3 GB/s. Compute units differ: 2,560 shading units, 160 TMUs, 64 ROPs, and 40 RT cores for AMD; 6,400 shading units, 200 TMUs, 80 ROPs, 50 RT cores, and 200 tensor cores for NVIDIA. The AMD card achieves 13.21 TFLOPS FP32 and 26.43 TFLOPS FP16 (2:1), while the NVIDIA card achieves 21.89 TFLOPS for both FP32 and FP16 (1:1).

Physical differences include TDP (165 W vs 320 W), slot width (IGP vs dual-slot), power connectors (none vs 2x 8-pin), and dimensions (the CMP 90HX is 285 mm long and 112 mm tall). The RX 6850M XT has portable device dependent display outputs, while the CMP 90HX has no outputs. The bus interface also differs: PCIe 4.0 x16 for AMD versus PCIe 1.0 x4 for NVIDIA. Release dates differ as well, with the RX 6850M XT launching on January 3, 2022, and the CMP 90HX launching on July 27, 2021.

The Verdict

The data points to a clear winner for general compute: the AMD Radeon RX 6850M XT. It wins the only head-to-head benchmark by 23.2%, has a higher average score (78,940 vs 69,000), and holds a higher percentile ranking (92 vs 90). Its 165 W TDP is nearly half the CMP 90HX’s 320 W, yet it still outperforms the NVIDIA card in OpenCL. The RX 6850M XT also has a broader benchmark footprint, including a 3DMark Steel Nomad score, which the CMP 90HX lacks entirely.

The NVIDIA CMP 90HX is not without merit, but its strengths lie outside the measured benchmarks. It has 2.5x more shading units, 66% higher FP32 throughput, and 76% more memory bandwidth. Its 200 tensor cores and 50 RT cores are unique among the two. These specifications are designed for mining workloads, which are not represented in the benchmark suite. The CMP 90HX also has a much larger die (628 mm²) and more transistors (28,300 million), indicating higher raw capability that the OpenCL test does not capture.

For any user seeking a general-purpose GPU, the RX 6850M XT is the better choice based on every available benchmark. It performs better, consumes less power, and has display output capability. For a mining-specific deployment, the CMP 90HX’s memory bandwidth and compute resources may be more appropriate, but the data provided does not include mining benchmarks to confirm this. The verdict from the benchmark data is unambiguous: the RX 6850M XT is the superior product in measured performance, while the CMP 90HX remains a niche tool with unverified advantages in its intended workload.