Intel Arc Pro A60 vs NVIDIA CMP 90HX Comparison

The NVIDIA CMP 90HX and Intel Arc Pro A60 are two fundamentally different graphics cards, one built for a niche that no longer exists and the other designed for professional workstations. The data shows the CMP 90HX, despite its mining-only pedigree, holds a clear performance lead in the single shared benchmark, but the Arc Pro A60 counters with a more modern feature set, higher memory capacity, and drastically lower power requirements. This analysis breaks down what each card offers based strictly on the available benchmark data and specifications.

FAQ

Q: Which card is faster in the shared OpenCL benchmark?

A: The NVIDIA CMP 90HX scores 69,000 in Geekbench OpenCL, which is 8.7% ahead of the Intel Arc Pro A60's 63,485. The CMP 90HX also outperforms its nearest rival, the Intel Arc A770, by 0.3%, while the Arc Pro A60 matches the NVIDIA GeForce RTX 4090's average score of 60,347 with a 0% delta.

Q: How do their overall performance percentiles compare?

A: The NVIDIA CMP 90HX sits in the 90th percentile of all GPUs, while the Intel Arc Pro A60 is in the 88th percentile. Despite this 2-point difference, the CMP 90HX's average benchmark score of 69,000 is significantly higher than the Arc Pro A60's average of 60,326, which is pulled down by its lower Vulkan score of 57,166.

Q: Which card has more memory and what type?

A: The Intel Arc Pro A60 has 12 GB of GDDR6 memory on a 192-bit bus, providing 384.0 GB/s of bandwidth. In contrast, the NVIDIA CMP 90HX has 10 GB of GDDR6X memory on a 320-bit bus, delivering a much higher 760.3 GB/s of bandwidth.

Q: What are the power consumption differences?

A: The NVIDIA CMP 90HX has a TDP of 320 W and requires a 700 W power supply, while the Intel Arc Pro A60 has a TDP of just 130 W and a suggested 300 W PSU. The CMP 90HX also needs two 8-pin power connectors, whereas the Arc Pro A60's power connector configuration is not listed.

Q: Can these cards output video to displays?

A: No. The NVIDIA CMP 90HX has no display outputs whatsoever, as it was designed solely for mining. The Intel Arc Pro A60, however, supports four DisplayPort 2.0 outputs, making it a functional workstation card for multi-monitor setups.

Q: Which card has the more advanced manufacturing process?

A: The Intel Arc Pro A60 uses a 6 nm process from TSMC, while the NVIDIA CMP 90HX uses an 8 nm process from Samsung. The Arc Pro A60 also has a higher boost clock of 2050 MHz compared to the CMP 90HX's 1710 MHz, despite having a much lower base clock of 900 MHz versus 1500 MHz.

Architecture Differences

The two cards come from entirely different architectural lineages. The NVIDIA CMP 90HX is built on the Ampere architecture using the GA102 chip, manufactured on Samsung's 8 nm process. This is a massive chip with 28,300 million transistors on a 628 mm² die, yielding a transistor density of 45.1M per mm². The Intel Arc Pro A60 uses the Xe-HPG architecture with the DG2-256 chip, built on TSMC's 6 nm node. It has 11,500 million transistors on a 269 mm² die, for a density of 42.8M per mm². While the CMP 90HX is physically larger and packs more transistors, the Arc Pro A60's newer process node allows for a much higher boost clock.

Feature-wise, the NVIDIA card includes 50 RT cores and 200 tensor cores, which are absent from the Intel card's specifications list, where tensor cores are marked as null. The CMP 90HX has 16 RT cores, but its tensor core count is not provided. Both cards support DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4, so they are matched on API support. The NVIDIA card is end-of-life, released on 2021-07-27, while the Intel card is active and was released on 2023-06-05. This generational gap is critical: the CMP 90HX is a relic of the mining boom with no video outputs, while the Arc Pro A60 is a current professional product with modern display connectivity.

The Verdict

The data paints a clear picture for different buyers. The NVIDIA CMP 90HX is the raw performance winner, posting an 8.7% higher OpenCL score than the Intel Arc Pro A60. Its 21.89 TFLOPS of FP32 performance versus 8.397 TFLOPS for the Intel card is a massive compute advantage. If your only metric is raw compute throughput in a headless environment—and you have the 320 W TDP and 700 W PSU budget to accommodate it—the CMP 90HX is the stronger choice. Its 90th percentile ranking versus the Arc Pro A60's 88th percentile further supports this.

However, the Intel Arc Pro A60 is the only viable option for anyone needing a functional graphics card. The CMP 90HX has no display outputs, making it useless in a standard desktop or workstation. The Arc Pro A60 offers 4x DisplayPort 2.0, 12 GB of memory versus 10 GB, and a 130 W TDP that requires only a 300 W PSU. It also has a higher boost clock of 2050 MHz and supports PCIe 4.0 x16, whereas the CMP 90HX bizarrely lists PCIe 1.0 x4 as its bus interface. For professional use, an active product with modern connectivity and lower power draw is the practical pick, even if it trails in raw compute. The CMP 90HX is a specialized compute card for a bygone era; the Arc Pro A60 is a current, usable workstation GPU.

Specification Differences

The most glaring difference is the bus interface: the NVIDIA CMP 90HX lists PCIe 1.0 x4, while the Intel Arc Pro A60 uses PCIe 4.0 x16. This is a massive gap in data transfer capability, even if the CMP 90HX's compute power is higher. The TDP also diverges sharply: 320 W for the NVIDIA versus 130 W for the Intel, with corresponding suggested PSUs of 700 W and 300 W. The CMP 90HX is dual-slot with two 8-pin connectors, while the Arc Pro A60 is single-slot with no listed power connector requirement.

Memory differs in both quantity and type. The CMP 90HX has 10 GB of GDDR6X with a 320-bit bus, achieving 760.3 GB/s bandwidth. The Arc Pro A60 has 12 GB of GDDR6 on a 192-bit bus, halving bandwidth to 384.0 GB/s. Shader resources are heavily skewed toward the NVIDIA card, which has 6400 shading units, 200 TMUs, and 80 ROPs, versus the Intel card's 2048 shading units, 128 TMUs, and 64 ROPs. Clock speeds tell a different story: the Arc Pro A60 boosts to 2050 MHz from a 900 MHz base, while the CMP 90HX boosts to 1710 MHz from a 1500 MHz base.

Process node and die size also differ: the CMP 90HX is on 8 nm with a 628 mm² die, while the Arc Pro A60 is on 6 nm with a 269 mm² die. Transistor counts are 28,300 million versus 11,500 million. The CMP 90HX has no display outputs; the Arc Pro A60 has 4x DisplayPort 2.0. Production status is end-of-life for the NVIDIA and active for the Intel. The CMP 90HX is 285 mm long and 112 mm tall, while the Arc Pro A60's dimensions are not listed.

Head-to-Head Benchmarks

The only direct comparison available is the Geekbench OpenCL test, where the NVIDIA CMP 90HX wins decisively. It scores 69,000 against the Intel Arc Pro A60's 63,485, a delta of 8.7%. This is a significant margin that reflects the CMP 90HX's superior compute resources: 6400 shading units versus 2048, and 21.89 TFLOPS FP32 versus 8.397 TFLOPS. The CMP 90HX also has a higher pixel rate (136.8 GPixel/s versus 131.2 GPixel/s) and texture rate (342.0 GTexel/s versus 262.4 GTexel/s), though these are not directly benchmarked.

However, the Arc Pro A60 has its own benchmark data that the CMP 90HX lacks. In Geekbench Vulkan, the Arc Pro A60 scores 57,166, which is lower than its OpenCL score of 63,485. This suggests the Intel card performs better in OpenCL workloads than in Vulkan, at least in this test. The CMP 90HX has no Vulkan score listed, so a cross-API comparison is impossible. Looking at rival performance, the CMP 90HX edges out the Intel Arc A770 by 0.3% and the AMD Radeon Instinct MI25 by 0.6%, but trails the AMD Radeon Pro WX 8200 by 1.2% and the NVIDIA Quadro P6000 by 1.4%. The Arc Pro A60, meanwhile, matches the RTX 4090 exactly (0% delta) and beats the AMD Radeon PRO V710 by 2.8%, but falls behind the AMD Radeon Pro W6600M by 2.5%.

Where Each One Wins

The NVIDIA CMP 90HX wins in every compute-heavy metric available. It has 2.6x the FP32 throughput of the Arc Pro A60 (21.89 TFLOPS versus 8.397 TFLOPS), nearly double the memory bandwidth (760.3 GB/s versus 384.0 GB/s), and 3.1x the shading units (6400 versus 2048). It also has a higher pixel rate and texture rate, and its 90th percentile ranking is above the Intel card's 88th. The 8.7% OpenCL benchmark win is the definitive proof: for pure number-crunching, the CMP 90HX is the superior card. This makes it the pick for any headless compute task where power draw and display output are irrelevant.

The Intel Arc Pro A60 wins on every practical front. It has 12 GB of memory versus 10 GB, which is critical for large datasets. It offers 4x DisplayPort 2.0 outputs, making it usable as a display adapter, while the CMP 90HX has none. Its 130 W TDP and 300 W PSU requirement are less than half of the CMP 90HX's 320 W and 700 W demands, saving energy and thermal overhead. It uses a modern PCIe 4.0 x16 interface versus the CMP 90HX's PCIe 1.0 x4, which is a massive IO advantage. The Arc Pro A60 is also an active product with a 2023 release, whereas the CMP 90HX is end-of-life from 2021. For any workstation task involving displays, modern connectivity, or power efficiency, the Arc Pro A60 is the clear winner, despite its lower raw compute scores.