AMD Radeon RX 5600 OEM vs NVIDIA CMP 90HX Comparison

FAQ

Q: What is the performance difference between the NVIDIA CMP 90HX and the AMD Radeon RX 5600 OEM?

A: In the only shared benchmark, Geekbench OpenCL, the NVIDIA CMP 90HX scores 69,000 points versus 62,427 points for the AMD Radeon RX 5600 OEM. That represents a 10.5% advantage for the NVIDIA card.

Q: Which card has the higher overall benchmark percentile?

A: The NVIDIA CMP 90HX sits at the 90th percentile among all GPUs, while the AMD Radeon RX 5600 OEM is at the 87th percentile. The NVIDIA card also has a higher average benchmark score of 69,000 compared to 58,085 for the AMD card.

Q: How do the memory configurations differ between these two cards?

A: The NVIDIA CMP 90HX has 10 GB of GDDR6X memory on a 320-bit bus with 760.3 GB/s bandwidth. The AMD Radeon RX 5600 OEM has 6 GB of GDDR6 memory on a 192-bit bus with 288.0 GB/s bandwidth. The NVIDIA card offers more capacity, a faster memory type, and more than double the bandwidth.

Q: What are the power requirements for each card?

A: The NVIDIA CMP 90HX has a TDP of 320 W and requires a 700 W power supply with two 8-pin connectors. The AMD Radeon RX 5600 OEM has a TDP of 125 W, needs only a 300 W power supply, and uses a single 8-pin connector.

Q: Which card supports ray tracing or tensor operations?

A: Only the NVIDIA CMP 90HX includes dedicated hardware for these features, with 50 RT cores and 200 tensor cores. The AMD Radeon RX 5600 OEM has no RT cores and no tensor cores listed in the database.

Q: What is the release timeline for these products?

A: The AMD Radeon RX 5600 OEM was released on January 20, 2020, while the NVIDIA CMP 90HX came later on July 27, 2021. Both are now listed as end-of-life products.

Architecture Differences

The NVIDIA CMP 90HX and AMD Radeon RX 5600 OEM represent fundamentally different design philosophies. The NVIDIA card uses the GA102 chip built on Samsung's 8 nm process, while the AMD card uses the Navi 10 chip fabricated by TSMC on a 7 nm node. This process difference is significant: AMD's smaller node allows for a more compact die, measuring 251 mm² versus NVIDIA's 628 mm².

Transistor counts tell a dramatic story. The GA102 packs 28,300 million transistors, nearly three times the 10,300 million found in the Navi 10. However, transistor density is surprisingly similar, with NVIDIA at 45.1 million transistors per square millimeter and AMD at 41.0 million. The NVIDIA architecture is simply much larger overall.

The compute resources differ enormously. NVIDIA's CMP 90HX has 6,400 shading units, 200 texture mapping units, and 80 render output units. The AMD card has 2,048 shading units, 128 TMUs, and 64 ROPs. NVIDIA also integrates 50 RT cores and 200 tensor cores, features entirely absent from the AMD part. This means the CMP 90HX can handle hardware-accelerated ray tracing and AI workloads, while the RX 5600 OEM relies purely on traditional rasterization hardware.

The memory subsystems are equally divergent. The CMP 90HX uses GDDR6X memory with a 320-bit bus, while the RX 5600 OEM uses GDDR6 on a 192-bit bus. Clock speeds differ as well: the NVIDIA card has a base clock of 1500 MHz and boost of 1710 MHz, while the AMD card runs at 1130 MHz base, 1375 MHz game clock, and 1560 MHz boost. The NVIDIA memory runs at 19 Gbps effective, substantially faster than AMD's 12 Gbps effective.

API support shows another split. The NVIDIA card supports DirectX 12 Ultimate (12_2), while the AMD card only supports DirectX 12 (12_1). Both cards support OpenGL 4.6 and Vulkan 1.4. The CMP 90HX also has no display outputs, a clear indicator of its mining-focused design, while the RX 5600 OEM includes 1x HDMI 2.0b and 3x DisplayPort 1.4a outputs.

Head-to-Head Benchmarks

The database contains one direct comparison between these two cards: Geekbench OpenCL. In this test, the NVIDIA CMP 90HX scores 69,000 points against 62,427 points for the AMD Radeon RX 5600 OEM. The 10.5% delta in favor of NVIDIA is substantial, though not overwhelming given the architectural differences. This means the CMP 90HX wins the only head-to-head benchmark recorded, with one win for NVIDIA and zero for AMD.

Looking at how each card stacks against its own nearest rivals provides more context. The CMP 90HX's 69,000 average score places it just 0.3% ahead of the Intel Arc A770 (68,809) and 0.6% ahead of the AMD Radeon Instinct MI25 (68,562). It trails the AMD Radeon Pro WX 8200 by 1.2% (69,870) and the NVIDIA Quadro P6000 by 1.4% (69,986). This cluster of scores within roughly 1.5% suggests the CMP 90HX performs near the top of its immediate performance tier.

The RX 5600 OEM's average benchmark score of 58,085 is more modest. It sits 0.3% behind the Intel Arc A570M (58,239) and 0.5% behind the AMD Radeon RX 6950 XT (58,392). It leads the Intel Arc A580 by 0.6% (57,756) and trails the NVIDIA P102-100 by 0.8% (58,528). Notably, the RX 6950 XT in this database appears with an unexpectedly low average score, likely reflecting a specific OpenCL workload rather than gaming performance.

The performance gap between the two cards in the head-to-head test aligns with their raw specifications. The CMP 90HX delivers 21.89 TFLOPS of FP32 compute, while the RX 5600 OEM manages 6.390 TFLOPS. That is more than a threefold difference in theoretical floating-point throughput, yet the real-world OpenCL delta is only 10.5%. This suggests the AMD architecture extracts efficiency from its smaller memory footprint and lower power draw.

Specification Differences

The two cards diverge across nearly every specification category. Starting with process technology, the NVIDIA CMP 90HX uses an 8 nm Samsung process, while the AMD Radeon RX 5600 OEM uses a 7 nm TSMC process. This gives AMD a density advantage per square millimeter, though NVIDIA compensates with a much larger die.

Memory specifications differ completely. The CMP 90HX offers 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. The RX 5600 OEM has 6 GB of GDDR6 on a 192-bit bus with 288.0 GB/s bandwidth. Memory clock speeds also differ, with NVIDIA at 1188 MHz (19 Gbps effective) and AMD at 1500 MHz (12 Gbps effective). The NVIDIA card's bandwidth advantage is roughly 2.6 times that of the AMD card.

Compute resources show massive differences. NVIDIA has 6,400 shading units, 200 TMUs, and 80 ROPs, producing a pixel rate of 136.8 GPixel/s and a texture rate of 342.0 GTexel/s. AMD has 2,048 shading units, 128 TMUs, and 64 ROPs, with pixel rate of 99.84 GPixel/s and texture rate of 199.7 GTexel/s. The FP32 throughput is 21.89 TFLOPS for NVIDIA versus 6.390 TFLOPS for AMD. For FP16, NVIDIA maintains a 1:1 ratio at 21.89 TFLOPS, while AMD achieves 12.78 TFLOPS at a 2:1 ratio.

Power and connectivity also differ. The CMP 90HX draws 320 W TDP with a 700 W suggested PSU and dual 8-pin connectors. The RX 5600 OEM draws 125 W TDP with a 300 W suggested PSU and a single 8-pin connector. The bus interface differs markedly: NVIDIA uses PCIe 1.0 x4, while AMD uses PCIe 4.0 x16. The CMP 90HX has no display outputs, while the RX 5600 OEM provides standard video outputs.

Clock speeds show NVIDIA's advantage in boost clocks (1710 MHz versus 1560 MHz) but AMD counters with a higher memory clock in raw MHz (1500 MHz versus 1188 MHz), though the effective data rates favor NVIDIA due to GDDR6X technology.

Where Each One Wins

The NVIDIA CMP 90HX wins decisively in raw compute performance. Its 21.89 TFLOPS FP32 throughput, 50 RT cores, and 200 tensor cores make it suitable for compute-heavy workloads that can leverage these features. The 10 GB GDDR6X memory with 760.3 GB/s bandwidth provides ample capacity and speed for large datasets or memory-intensive calculations. The CMP 90HX also leads in the only recorded benchmark, beating the RX 5600 OEM by 10.5% in Geekbench OpenCL. Its higher 90th percentile ranking versus 87th for AMD reinforces this performance tier.

The AMD Radeon RX 5600 OEM wins in efficiency and practical usability. Its 125 W TDP is less than half of NVIDIA's 320 W, and it requires only a 300 W power supply compared to 700 W. The single 8-pin connector simplifies installation. The card also has display outputs, making it usable in standard desktop systems, whereas the CMP 90HX has no outputs at all. The PCIe 4.0 x16 interface provides modern system compatibility, while the CMP 90HX's PCIe 1.0 x4 interface is severely limited for data transfer.

For gaming workloads, the RX 5600 OEM is the only realistic option of the two, as the CMP 90HX cannot output video. The AMD card supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4, making it functional for current games, though it lacks hardware ray tracing. The CMP 90HX supports DirectX 12 Ultimate with ray tracing capability, but this is irrelevant without display outputs.

For compute or mining tasks where display output is unnecessary, the CMP 90HX offers substantially more processing power. Its 6400 shading units and 200 tensor cores provide a large execution footprint for parallel workloads. The RX 5600 OEM, with 2048 shading units, is better suited for light compute tasks or as a secondary card in systems where power constraints matter.

The production status of both cards is end-of-life, but their release dates explain their positioning. The RX 5600 OEM launched in January 2020 for the RDNA 1 generation. The CMP 90HX arrived in July 2021 as part of NVIDIA's mining-specific lineup, explaining its lack of display outputs and its unusual PCIe 1.0 x4 interface. The AMD card's predecessor is listed as Vega with successor Navi II, while the CMP 90HX has no listed predecessor or successor, reflecting its niche status.