NVIDIA CMP 90HX vs NVIDIA RTX 4000 Ada Generation Comparison

Head-to-Head Benchmarks

The recorded database contains a single direct benchmark comparison between these two GPUs, and the result is decisive. In the Geekbench OpenCL test, the NVIDIA RTX 4000 Ada Generation scores 146,593, while the NVIDIA CMP 90HX scores 69,000. That is a delta of 112.5% in favor of the RTX 4000 Ada Generation, meaning it more than doubles the CMP 90HX’s raw compute output in this workload. This is not a marginal victory; it is a generational gap expressed in a single metric.

Looking at the broader benchmark averages, the RTX 4000 Ada Generation holds an average score of 135,218 across all recorded tests, placing it in the 95th percentile of all GPUs in the database. The CMP 90HX, by contrast, averages 69,000 from its single OpenCL result, sitting at the 90th percentile. While both cards rank high against the entire field, the absolute separation between them is substantial: the RTX 4000 Ada Generation’s average is roughly 96% higher than the CMP 90HX’s average, a gap that mirrors the head-to-head OpenCL delta.

The RTX 4000 Ada Generation’s nearest rivals in the database include the NVIDIA A10M at 135,230 (0% delta), the AMD Radeon PRO W6800 at 135,396 (-0.1%), the AMD Radeon Pro W6800X Duo at 135,774 (-0.4%), and the AMD Radeon PRO V620 at 136,472 (-0.9%). These are all within a percentage point of the RTX 4000 Ada Generation’s average, which suggests it sits in a tightly competitive cluster among professional workstation cards. The CMP 90HX, however, is grouped with very different company: the Intel Arc A770 at 68,809 (0.3% delta), the AMD Radeon Instinct MI25 at 68,562 (0.6%), the AMD Radeon Pro WX 8200 at 69,870 (-1.2%), and the NVIDIA Quadro P6000 at 69,986 (-1.4%). The CMP 90HX is effectively at parity with those cards, all of which are far below the RTX 4000 Ada Generation’s performance tier.

The win count is 1 for the RTX 4000 Ada Generation and 0 for the CMP 90HX, and given the magnitude of the OpenCL delta, there is no ambiguity about which card dominates in compute performance. The data shows that the RTX 4000 Ada Generation is not merely faster; it is in a different performance class entirely.

FAQ

Q: How much faster is the NVIDIA RTX 4000 Ada Generation than the NVIDIA CMP 90HX in OpenCL?

A: The RTX 4000 Ada Generation scores 146,593 versus 69,000 for the CMP 90HX, a 112.5% advantage. This means it delivers more than double the OpenCL performance.

Q: Which GPU has a higher average benchmark score?

A: The RTX 4000 Ada Generation averages 135,218 across all recorded tests, while the CMP 90HX averages 69,000. The RTX 4000 Ada Generation’s average is about 96% higher.

Q: Where does each card rank relative to all GPUs in the database?

A: The RTX 4000 Ada Generation sits at the 95th percentile, while the CMP 90HX sits at the 90th percentile. Both are above average, but the RTX 4000 Ada Generation is clearly higher in the distribution.

Q: What kind of rivals does each card face in the database?

A: The RTX 4000 Ada Generation’s nearest rivals are professional workstation cards like the NVIDIA A10M, AMD Radeon PRO W6800, and AMD Radeon PRO V620, all within 1% of its average score. The CMP 90HX’s nearest rivals include the Intel Arc A770, AMD Radeon Instinct MI25, and NVIDIA Quadro P6000, all within 1.5% of its score.

Q: Does the CMP 90HX have any benchmark where it wins?

A: No. The database records one head-to-head benchmark, and the RTX 4000 Ada Generation wins that test. The CMP 90HX has no recorded wins in this comparison.

Q: Is the CMP 90HX still in production?

A: No. The database lists the CMP 90HX as end-of-life, while the RTX 4000 Ada Generation is listed as active.

Architecture Differences

The two GPUs are built on fundamentally different architectures and process technologies. The RTX 4000 Ada Generation uses the AD104 chip built on Ada Lovelace architecture, fabricated by TSMC on a 5 nm process. It packs 35,800 million transistors into a die size of 294 mm², yielding a transistor density of 121.8 million per mm². The CMP 90HX, by contrast, uses the GA102 chip built on Ampere architecture, fabricated by Samsung on an 8 nm process. It contains 28,300 million transistors spread across a much larger die of 628 mm², giving a transistor density of just 45.1 million per mm². The density difference reflects the newer manufacturing node: the Ada chip crams over 2.7 times more transistors per square millimeter into less than half the die area.

Memory configurations diverge sharply as well. The RTX 4000 Ada Generation has 20 GB of GDDR6 on a 160-bit bus, delivering 360.0 GB/s of bandwidth. The CMP 90HX has 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s. The CMP 90HX has more than double the memory bandwidth, but less than half the capacity. The memory clock rates also differ: the RTX 4000 Ada Generation runs at 2250 MHz (18 Gbps effective), while the CMP 90HX runs at 1188 MHz (19 Gbps effective). The CMP 90HX’s wider bus compensates for its lower clock, but the RTX 4000 Ada Generation’s smaller bus is paired with a newer memory controller.

The compute resources tell a mixed story. The RTX 4000 Ada Generation has 6,144 shading units, 192 TMUs, 64 ROPs, 48 RT cores, and 192 tensor cores. The CMP 90HX has 6,400 shading units, 200 TMUs, 80 ROPs, 50 RT cores, and 200 tensor cores. In raw unit counts, the CMP 90HX actually has slightly more of every compute component. However, the clock speeds flip the performance equation: the RTX 4000 Ada Generation boosts to 2175 MHz, while the CMP 90HX boosts to 1710 MHz. The RTX 4000 Ada Generation’s higher clocks drive its FP32 throughput to 26.73 TFLOPS, whereas the CMP 90HX manages 21.89 TFLOPS despite having more units. Pixel rate is nearly identical (139.2 GPixel/s versus 136.8 GPixel/s), but the texture rate favors the RTX 4000 Ada Generation at 417.6 GTexel/s versus 342.0 GTexel/s.

Power and physical design are also starkly different. The RTX 4000 Ada Generation has a 130 W TDP, is single-slot, uses one 16-pin connector, and requires a 300 W suggested PSU. The CMP 90HX has a 320 W TDP, is dual-slot, uses two 8-pin connectors, and requires a 700 W suggested PSU. The RTX 4000 Ada Generation is 245 mm long and 112 mm tall, while the CMP 90HX is 285 mm long and 112 mm tall. The RTX 4000 Ada Generation has four DisplayPort 1.4a outputs; the CMP 90HX has no display outputs at all, reflecting its mining-oriented design.

The Verdict

The data points to a clear conclusion: the NVIDIA RTX 4000 Ada Generation is the superior GPU for any compute-focused workload. Its OpenCL score of 146,593 versus 69,000 is a 112.5% lead, and its average benchmark score of 135,218 versus 69,000 places it nearly 96% ahead. The RTX 4000 Ada Generation also ranks at the 95th percentile of all GPUs, versus the 90th percentile for the CMP 90HX. For users who need raw compute performance, the RTX 4000 Ada Generation is the obvious pick.

The CMP 90HX does have advantages in the database, but they are not performance advantages. It offers more memory bandwidth (760.3 GB/s versus 360.0 GB/s) and a wider 320-bit bus, which could matter for certain memory-bound tasks. It also has slightly more shading units, TMUs, ROPs, RT cores, and tensor cores. However, those hardware advantages are undone by its lower boost clock (1710 MHz versus 2175 MHz) and older architecture, resulting in lower measured performance. The CMP 90HX’s nearest rivals in the database, such as the Intel Arc A770 and the NVIDIA Quadro P6000, are all within 1.4% of its average score, indicating it is competitive with older or mid-range cards, not with the current workstation generation.

The RTX 4000 Ada Generation’s nearest rivals are all professional cards within a percentage point of its average: the NVIDIA A10M, AMD Radeon PRO W6800, AMD Radeon Pro W6800X Duo, and AMD Radeon PRO V620. This shows it fits into a specific performance tier among modern workstations. The CMP 90HX, on the other hand, is grouped with older or different-purpose hardware. The RTX 4000 Ada Generation is also the only one of the two that is still in active production, and it offers display outputs, making it usable for general workstation tasks. The CMP 90HX is end-of-life and has no display outputs, limiting it to headless mining or compute setups.

For anyone choosing between these two, the RTX 4000 Ada Generation is the better investment in almost every measurable way. It is faster, more efficient (130 W versus 320 W TDP), smaller, and built on a newer process node. The only scenarios where the CMP 90HX might be preferable are those that specifically require its higher memory bandwidth and wider bus, and even then, the RTX 4000 Ada Generation’s massive compute lead makes it hard to justify. The benchmark data is unambiguous: the RTX 4000 Ada Generation wins the head-to-head, wins the average score comparison, and holds a higher percentile rank.

Specification Differences

The two GPUs differ across nearly every specification field in the database. The RTX 4000 Ada Generation uses a 5 nm TSMC process, while the CMP 90HX uses an 8 nm Samsung process. The RTX 4000 Ada Generation has 35,800 million transistors on a 294 mm² die with a density of 121.8M / mm²; the CMP 90HX has 28,300 million transistors on a 628 mm² die with a density of 45.1M / mm². Base clocks are identical at 1500 MHz, but boost clocks differ: 2175 MHz for the RTX 4000 Ada Generation versus 1710 MHz for the CMP 90HX. Memory clocks are 2250 MHz (18 Gbps effective) versus 1188 MHz (19 Gbps effective).

Memory capacity and type differ: the RTX 4000 Ada Generation has 20 GB of GDDR6 on a 160-bit bus with 360.0 GB/s bandwidth; the CMP 90HX has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. Compute unit counts are higher on the CMP 90HX: 6,400 shading units, 200 TMUs, 80 ROPs, 50 RT cores, and 200 tensor cores, versus 6,144 shading units, 192 TMUs, 64 ROPs, 48 RT cores, and 192 tensor cores on the RTX 4000 Ada Generation. However, pixel rate is nearly equal (139.2 GPixel/s versus 136.8 GPixel/s), and the RTX 4000 Ada Generation leads in texture rate (417.6 GTexel/s versus 342.0 GTexel/s) and FP32 throughput (26.73 TFLOPS versus 21.89 TFLOPS).

Power requirements are very different: 130 W TDP with a 300 W suggested PSU for the RTX 4000 Ada Generation, versus 320 W TDP with a 700 W suggested PSU for the CMP 90HX. The RTX 4000 Ada Generation is single-slot with one 16-pin connector; the CMP 90HX is dual-slot with two 8-pin connectors. Physical dimensions: 245 mm length and 112 mm height for the RTX 4000 Ada Generation, versus 285 mm length and 112 mm height for the CMP 90HX. The RTX 4000 Ada Generation has four DisplayPort 1.4a outputs and a PCIe 4.0 x16 interface; the CMP 90HX has no outputs and a PCIe 1.0 x4 interface. Production status: active versus end-of-life. Release dates: August 2023 for the RTX 4000 Ada Generation, July 2021 for the CMP 90HX. The RTX 4000 Ada Generation’s predecessor is Workstation Ampere and its successor is Blackwell PRO W; the CMP 90HX has no listed predecessor or successor. Both support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, and both have a 1:1 FP16/FP32 ratio.