AMD Radeon RX 7900M vs NVIDIA CMP 90HX Comparison

FAQ

Q: How do the two GPUs compare in raw compute performance?

A: The AMD Radeon RX 7900M delivers 38.52 TFLOPS of FP32 compute, while the NVIDIA CMP 90HX delivers 21.89 TFLOPS. In the recorded Geekbench OpenCL test, the AMD card scores 129,499 against 69,000 for the NVIDIA card, a difference of 87.7% in favor of AMD.

Q: What is the memory configuration difference between these two cards?

A: The AMD Radeon RX 7900M has 16 GB of GDDR6 memory on a 256-bit bus with 576.0 GB/s bandwidth. The NVIDIA CMP 90HX has 10 GB of GDDR6X memory on a 320-bit bus with 760.3 GB/s bandwidth, giving NVIDIA a 184.3 GB/s bandwidth advantage despite lower capacity.

Q: Which card is more power-efficient based on the data?

A: The AMD Radeon RX 7900M has a TDP of 180 W, while the NVIDIA CMP 90HX has a TDP of 320 W. The AMD card achieves higher benchmark scores at nearly half the power draw, indicating substantially better efficiency per watt in the recorded tests.

Q: What is the architectural generation difference?

A: The AMD Radeon RX 7900M uses the RDNA 3.0 architecture on a 5 nm TSMC process with 57,700 million transistors. The NVIDIA CMP 90HX uses the Ampere architecture on an 8 nm Samsung process with 28,300 million transistors. The AMD chip is built on a more advanced process node.

Q: Are these cards suitable for gaming or display output?

A: The AMD Radeon RX 7900M is an integrated GPU (IGP) with display outputs described as "Portable Device Dependent" and supports DirectX 12 Ultimate. The NVIDIA CMP 90HX has no display outputs at all, making it unsuitable for any visual output without a secondary GPU.

Q: How do the two cards rank among all GPUs in the database?

A: The AMD Radeon RX 7900M sits in the 94th percentile of all GPUs, while the NVIDIA CMP 90HX sits in the 90th percentile. The AMD card's average benchmark score is 97,487 versus 69,000 for the NVIDIA card.

Where Each One Wins

The AMD Radeon RX 7900M wins the only directly comparable benchmark in the database. In Geekbench OpenCL, it scores 129,499 versus 69,000 for the NVIDIA CMP 90HX, a margin of 87.7%. Beyond that single test, the AMD card also has additional benchmark entries in 3DMark Steel Nomad DX12 (4,201) and Geekbench Vulkan (158,760), while the NVIDIA card has no recorded results in either of those tests. The AMD card also achieves a higher average benchmark score across all recorded tests: 97,487 versus 69,000 for the NVIDIA card.

The NVIDIA CMP 90HX does hold advantages in specific hardware characteristics that do not translate into a benchmark win here. Its memory bandwidth is 760.3 GB/s versus 576.0 GB/s for the AMD card, and it has a wider 320-bit memory bus compared to the AMD card's 256-bit bus. Its 10 GB of GDDR6X memory operates at 19 Gbps effective versus 18 Gbps effective for the AMD card's GDDR6 memory. However, these advantages do not show up in the recorded OpenCL test, where the AMD card dominates.

The AMD card also wins on power efficiency. At 180 W TDP, it draws 140 W less than the NVIDIA card's 320 W TDP, yet it produces higher benchmark scores. This makes the AMD card the clear choice for scenarios where power draw matters, such as mobile or compact builds. The NVIDIA card, with its dual-slot design and 2x 8-pin power connectors, requires a 700 W suggested PSU, which is a substantial system requirement.

Architecture Differences

The AMD Radeon RX 7900M is built on the RDNA 3.0 architecture using a 5 nm process at TSMC. It packs 57,700 million transistors onto a 529 mm² die, resulting in a transistor density of 109.1M per mm². The chip is codenamed Plum Bonito and belongs to the Navi Mobile (RX 7000M) generation. It features 4,608 shading units, 288 texture mapping units, 192 ROPs, and 72 ray tracing cores. It has no dedicated tensor cores.

The NVIDIA CMP 90HX uses the Ampere architecture on an 8 nm process at Samsung. It contains 28,300 million transistors on a larger 628 mm² die, giving a much lower transistor density of 45.1M per mm². The chip is the GA102 and belongs to the Mining GPUs generation. It features 6,400 shading units, 200 texture mapping units, 80 ROPs, 50 ray tracing cores, and 200 tensor cores. The NVIDIA card has more shading units and tensor cores, but the AMD card has more TMUs and ROPs.

The process node difference is significant. The AMD card's 5 nm process allows nearly 2.4 times the transistor density of the NVIDIA card's 8 nm process, despite the NVIDIA die being larger. This density advantage likely contributes to the AMD card's higher compute throughput per watt. The AMD card also supports FP16 at 77.05 TFLOPS (2:1 ratio), while the NVIDIA card's FP16 performance equals its FP32 at 21.89 TFLOPS (1:1 ratio).

Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The bus interfaces differ: the AMD card uses PCIe 4.0 x16, while the NVIDIA card uses PCIe 1.0 x4, a substantial bandwidth limitation for the NVIDIA card.

Specification Differences

The AMD Radeon RX 7900M has a base clock of 1825 MHz and a boost clock of 2090 MHz. The NVIDIA CMP 90HX has a base clock of 1500 MHz and a boost clock of 1710 MHz. The AMD card's memory runs at 2250 MHz (18 Gbps effective), while the NVIDIA card's memory runs at 1188 MHz (19 Gbps effective).

Memory capacity differs: 16 GB GDDR6 for AMD versus 10 GB GDDR6X for NVIDIA. The bus widths are 256-bit for AMD and 320-bit for NVIDIA. Bandwidth is 576.0 GB/s for AMD and 760.3 GB/s for NVIDIA, a 32% advantage for the NVIDIA card.

Pixel and texture rates favor AMD. The AMD card produces 401.3 GPixel/s and 601.9 GTexel/s, while the NVIDIA card produces 136.8 GPixel/s and 342.0 GTexel/s. The AMD card's pixel rate is nearly three times higher.

Power and physical specifications differ greatly. The AMD card is an IGP with no power connectors and a 180 W TDP. The NVIDIA card is dual-slot, requires 2x 8-pin power connectors, has a 320 W TDP, and suggests a 700 W PSU. The NVIDIA card measures 285 mm in length and 112 mm in height; the AMD card has no recorded dimensions.

Release dates differ by over two years: the AMD card launched on 2023-10-18, while the NVIDIA card launched on 2021-07-27. The AMD card is still active in production, while the NVIDIA card is end-of-life. The AMD card's predecessor is Polaris Mobile; the NVIDIA card has no recorded predecessor or successor.

Head-to-Head Benchmarks

The only head-to-head benchmark in the database is Geekbench OpenCL. The AMD Radeon RX 7900M scores 129,499, while the NVIDIA CMP 90HX scores 69,000. The delta is 87.7% in favor of AMD. This is a decisive win, nearly doubling the NVIDIA card's score.

Contextualizing this score, the AMD card's nearest rivals include the AMD Radeon Pro VII at 97,131 (0.4% delta), the NVIDIA Quadro RTX 6000 at 101,872 (negative 4.3% delta), the AMD Radeon Instinct MI60 at 92,466 (5.4% delta), and the NVIDIA RTX A4500 at 91,671 (6.3% delta). The AMD card's average score of 97,487 places it slightly above the Radeon Pro VII and above the Instinct MI60 and RTX A4500, while slightly below the Quadro RTX 6000.

The NVIDIA CMP 90HX's nearest rivals include the Intel Arc A770 at 68,809 (0.3% delta), the AMD Radeon Instinct MI25 at 68,562 (0.6% delta), the AMD Radeon Pro WX 8200 at 69,870 (negative 1.2% delta), and the NVIDIA Quadro P6000 at 69,986 (negative 1.4% delta). The NVIDIA card's score of 69,000 sits right in the middle of this group, essentially tied with the Arc A770 and MI25, while slightly below the WX 8200 and P6000.

The AMD card also has two additional benchmark entries that the NVIDIA card lacks: 3DMark Steel Nomad DX12 at 4,201 and Geekbench Vulkan at 158,760. These extra data points further separate the two cards, as the NVIDIA card has no recorded performance in those tests.

The Verdict

The data clearly favors the AMD Radeon RX 7900M in every recorded benchmark metric. It wins the only head-to-head test by 87.7%, has additional benchmark results in two other tests, and ranks in the 94th percentile of all GPUs versus the 90th percentile for the NVIDIA CMP 90HX.

Users should pick the AMD Radeon RX 7900M if they need a GPU with high compute performance, lower power draw, display output capability, and modern architecture features. Its 180 W TDP, 16 GB memory capacity, and active production status make it a practical choice for systems where efficiency and current availability matter. The card's RDNA 3.0 architecture and 5 nm process represent a newer generation of technology.

Users should pick the NVIDIA CMP 90HX only if they have a specific need for its characteristics: higher memory bandwidth (760.3 GB/s), larger memory bus (320-bit), or tensor cores (200). The card is end-of-life, consumes 320 W, requires a 700 W PSU, has no display outputs, and uses a legacy PCIe 1.0 x4 interface. Its 10 GB memory capacity is lower than the AMD card's 16 GB, and its compute performance is less than a third of the AMD card's in the recorded OpenCL test. The NVIDIA card's only advantages are memory bandwidth and tensor core count, neither of which translates into a benchmark win in the database records.

For any general-purpose compute workload, the AMD Radeon RX 7900M is the superior choice based on the recorded data. The NVIDIA CMP 90HX remains a niche product for mining-specific scenarios where its memory bandwidth and tensor cores might be relevant, but the benchmark evidence does not support any performance advantage over the AMD card.