NVIDIA CMP 90HX vs NVIDIA RTX A4500 Mobile Comparison
NVIDIA CMP 90HX
RTX A4500 Mobile
PERFORMANCE BENCHMARKS
Analysis: NVIDIA CMP 90HX vs NVIDIA RTX A4500 Mobile
The NVIDIA RTX A4500 Mobile and NVIDIA CMP 90HX are both Ampere-generation parts, but they are engineered for entirely different purposes. The A4500 Mobile is a professional laptop GPU designed for compute and rendering workloads, while the CMP 90HX is a desktop mining card with no display outputs. The benchmark data shows a clear performance gap between them, with the mobile part holding a significant lead in the only shared test.
Head-to-Head Benchmarks
The sole head-to-head benchmark available is Geekbench OpenCL, and the results are decisively in favor of the NVIDIA RTX A4500 Mobile. The A4500 Mobile scores 105,307 points, while the CMP 90HX trails at 69,000 points. This represents a delta of 52.6%, meaning the A4500 Mobile is more than half as fast again as the CMP 90HX in this compute-oriented test. This is a remarkable outcome given that the CMP 90HX is a desktop card with a higher thermal design power and a larger, more powerful chip.
The magnitude of this win suggests that the A4500 Mobile’s architecture and driver optimizations are far better suited to the OpenCL workload. Even though the CMP 90HX has more shading units, texture mapping units, and ray tracing cores, it cannot translate that raw hardware into a competitive score. The A4500 Mobile’s 52.6% advantage is not a marginal victory; it is a substantial gap that places the two products in different performance tiers for general-purpose compute.
Looking at broader context, the A4500 Mobile’s average benchmark score of 91,134 places it in the 93rd percentile of all GPUs. Its nearest rival is the desktop NVIDIA RTX A4500, which scores 91,671, a negligible 0.6% difference. This indicates that the mobile part is essentially on par with its desktop namesake in average performance. The CMP 90HX, by contrast, has an average score of 69,000 and sits in the 90th percentile. Its nearest rival is the Intel Arc A770 at 68,809 points, a 0.3% difference, showing that the mining card is clustered with mid-range desktop parts rather than high-end compute accelerators.
Architecture Differences
The two GPUs are built on the same 8 nm Samsung process node, but they use different chips. The RTX A4500 Mobile is built on the GA104 chip, which contains 17,400 million transistors on a 392 mm² die. The CMP 90HX uses the larger GA102 chip, which packs 28,300 million transistors onto a 628 mm² die. The transistor density is similar — 44.4 million per mm² for the A4500 Mobile and 45.1 million per mm² for the CMP 90HX — but the CMP 90HX’s larger die gives it a clear hardware resource advantage.
This advantage is visible in the core counts. The CMP 90HX has 6,400 shading units, 200 texture mapping units, 50 ray tracing cores, and 200 tensor cores. The A4500 Mobile has 5,888 shading units, 184 texture mapping units, 46 ray tracing cores, and 184 tensor cores. The CMP 90HX also has higher clock speeds, with a base of 1,500 MHz and a boost of 1,710 MHz, compared to the A4500 Mobile’s 930 MHz base and 1,500 MHz boost.
However, the A4500 Mobile fights back in other areas. It has 96 raster output units versus 80 on the CMP 90HX, and its memory configuration is different. The A4500 Mobile uses 16 GB of GDDR6 on a 256-bit bus, delivering a bandwidth of 512.0 GB/s. The CMP 90HX uses 10 GB of GDDR6X on a 320-bit bus, which yields a higher bandwidth of 760.3 GB/s. The memory clock speeds also differ, with the A4500 Mobile running at 2,000 MHz (16 Gbps effective) and the CMP 90HX at 1,188 MHz (19 Gbps effective).
The bus interface is another major differentiator. The A4500 Mobile uses PCIe 4.0 x16, while the CMP 90HX is limited to PCIe 1.0 x4. This is a severe bottleneck for the mining card in any workload that requires host communication. The A4500 Mobile also has display outputs, described as "Portable Device Dependent," while the CMP 90HX has no outputs at all, reflecting its dedicated mining purpose.
FAQ
Q: Which GPU has a higher maximum boost clock?
A: The CMP 90HX has a boost clock of 1,710 MHz, which is 210 MHz higher than the RTX A4500 Mobile’s boost of 1,500 MHz.
Q: How do the memory bandwidths compare?
A: The CMP 90HX offers 760.3 GB/s of bandwidth from its 320-bit GDDR6X interface, while the A4500 Mobile delivers 512.0 GB/s over a 256-bit GDDR6 bus.
Q: What is the difference in shading unit count?
A: The CMP 90HX has 6,400 shading units, while the A4500 Mobile has 5,888. The CMP 90HX also has more texture mapping units, ray tracing cores, and tensor cores.
Q: Which GPU has a higher theoretical FP32 performance?
A: The CMP 90HX has a higher FP32 rating at 21.89 TFLOPS, compared to the A4500 Mobile’s 17.66 TFLOPS.
Q: Can the CMP 90HX be used for display output?
A: No, the CMP 90HX has no display outputs, whereas the A4500 Mobile has outputs that are dependent on the portable device it is installed in.
Q: Which GPU scores higher in Geekbench OpenCL?
A: The RTX A4500 Mobile scores 105,307, which is 52.6% higher than the CMP 90HX’s score of 69,000.
Specification Differences
| Specification | NVIDIA RTX A4500 Mobile | NVIDIA CMP 90HX |
|---|---|---|
| Chip | GA104 | GA102 |
| Generation | Ampere-MW (Ax000) | Mining GPUs |
| Transistors | 17,400 million | 28,300 million |
| Die Size | 392 mm² | 628 mm² |
| Transistor Density | 44.4M / mm² | 45.1M / mm² |
| Base Clock | 930 MHz | 1500 MHz |
| Boost Clock | 1500 MHz | 1710 MHz |
| Memory Clock | 2000 MHz (16 Gbps effective) | 1188 MHz (19 Gbps effective) |
| Memory Size | 16 GB | 10 GB |
| Memory Type | GDDR6 | GDDR6X |
| Memory Bus Width | 256 bit | 320 bit |
| Memory Bandwidth | 512.0 GB/s | 760.3 GB/s |
| Shading Units | 5888 | 6400 |
| TMUs | 184 | 200 |
| ROPs | 96 | 80 |
| RT Cores | 46 | 50 |
| Tensor Cores | 184 | 200 |
| Pixel Rate | 144.0 GPixel/s | 136.8 GPixel/s |
| Texture Rate | 276.0 GTexel/s | 342.0 GTexel/s |
| FP32 Performance | 17.66 TFLOPS | 21.89 TFLOPS |
| FP16 Performance | 17.66 TFLOPS (1:1) | 21.89 TFLOPS (1:1) |
| TDP | 140 W | 320 W |
| Slot Width | Not specified | Dual-slot |
| Power Connectors | None | 2x 8-pin |
| Suggested PSU | Not specified | 700 W |
| Bus Interface | PCIe 4.0 x16 | PCIe 1.0 x4 |
| Display Outputs | Portable Device Dependent | No outputs |
| Release Date | 2022-03-21 | 2021-07-27 |
| Dimensions (Length) | Not specified | 285 mm (11.2 inches) |
| Dimensions (Height) | Not specified | 112 mm (4.4 inches) |
Where Each One Wins
The NVIDIA RTX A4500 Mobile is the clear winner in the compute benchmark that matters most. Its Geekbench OpenCL score of 105,307 is 52.6% higher than the CMP 90HX’s 69,000, and its average benchmark score of 91,134 places it in the 93rd percentile of all GPUs. This makes it the superior choice for any professional workload that relies on OpenCL compute, such as rendering, simulation, or data processing. The A4500 Mobile also has the advantage of a modern PCIe 4.0 x16 interface, which is essential for moving data to and from the GPU efficiently. Its 16 GB of memory is also more generous than the CMP 90HX’s 10 GB, and it has more raster output units, which can help with fill-rate-limited tasks.
The CMP 90HX, on the other hand, has raw hardware superiority in several areas. It has more shading units, texture mapping units, ray tracing cores, and tensor cores. Its FP32 performance of 21.89 TFLOPS is 24% higher than the A4500 Mobile’s 17.66 TFLOPS. It also has significantly higher memory bandwidth at 760.3 GB/s versus 512.0 GB/s. However, these advantages do not translate into a higher Geekbench OpenCL score. The CMP 90HX’s PCIe 1.0 x4 interface is a severe limitation, and its lack of display outputs makes it unsuitable for any interactive or display-based task.
In practical terms, the A4500 Mobile is a mobile workstation GPU that can handle professional graphics and compute tasks while being integrated into a laptop, with display output capability and a much lower power draw of 140 W. The CMP 90HX is a mining-specific card with a 320 W TDP, dual-slot cooling, and 2x 8-pin power connectors, designed to be installed in a mining rig and run continuously. The data shows that for general compute performance, the A4500 Mobile is the better part, despite having a smaller die and fewer cores. The CMP 90HX’s theoretical advantages in clock speed and memory bandwidth do not overcome its architectural limitations in the tested workload.