AMD Radeon Pro W5700X vs NVIDIA CMP 40HX Comparison

Head-to-Head Benchmarks

The benchmark data is decisive. The NVIDIA CMP 40HX wins both recorded head-to-head comparisons, and it does so by a wide margin. In Geekbench OpenCL, the CMP 40HX scores 93,395 against the Radeon Pro W5700X's 43,810, a 113.2% advantage. In Geekbench Vulkan, the lead narrows but remains substantial: 77,879 versus 45,246, a 72.1% delta. These are not marginal wins; they represent a generational gap in compute throughput as measured by these workloads.

The average benchmark score tells a similar story. The CMP 40HX averages 85,637 across its recorded tests, while the W5700X averages 54,828. That puts the NVIDIA part roughly 56% higher in aggregate. The CMP 40HX also sits at the 93rd percentile of all GPUs in the database, while the W5700X sits at the 87th. Both are above average, but the NVIDIA card is clearly further up the distribution.

Context from nearest rivals reinforces the gap. The CMP 40HX's average score is 4.4% above the AMD Radeon PRO W6600 and 5.8% above the AMD Radeon Pro Vega 64X, while sitting just 1.7% below the AMD Radeon PRO W7600. The W5700X, by contrast, lands within 1.7% of the AMD Radeon 8060S and within 1.6% of the AMD Radeon RX 6750 GRE 12 GB. The two cards are not competing in the same performance tier according to the database's aggregate scoring.

One important caveat in the data: the W5700X has a Geekbench Metal score of 75,427, a test the CMP 40HX does not have recorded. This means the W5700X has a broader benchmark footprint, but its OpenCL and Vulkan results are what allow a direct comparison. In those shared tests, the NVIDIA card wins outright.

Architecture Differences

The two GPUs come from different design philosophies and process generations. The NVIDIA CMP 40HX uses the TU106 chip on the Turing architecture, built on a 12 nm process at TSMC. The AMD Radeon Pro W5700X uses the Navi 10 chip on RDNA 1.0, built on a 7 nm process, also at TSMC. The process node difference is significant: 12 nm versus 7 nm, which helps explain the AMD card's higher transistor density. The CMP 40HX packs 10,800 million transistors across a 445 mm² die, yielding a density of 24.3 million transistors per square millimeter. The W5700X has 10,300 million transistors on a much smaller 251 mm² die, for a density of 41.0 million per square millimeter.

Clock behavior also diverges sharply. The CMP 40HX has a base clock of 1470 MHz and a boost of 1650 MHz, a modest spread. The W5700X has a base of 1243 MHz but boosts to 2040 MHz, a much larger dynamic range. The AMD part's boost clock is 390 MHz higher than the NVIDIA card's, which contributes to its higher peak throughput in raw terms.

Compute resources differ in configuration. The CMP 40HX has 2304 shading units, 144 texture mapping units, and 64 ROPs. The W5700X has 2560 shading units, 160 TMUs, and 64 ROPs. The AMD card has more shaders and TMUs, but the NVIDIA card counters with dedicated hardware: 36 RT cores and 288 tensor cores. The W5700X has neither, reflecting its lack of hardware ray tracing and tensor acceleration. The CMP 40HX also supports DirectX 12 Ultimate (12_2), while the W5700X only reaches DirectX 12 (12_1), a meaningful feature gap for applications that use the latest graphics API features.

Memory configurations are split on capacity but identical on bandwidth. Both use GDDR6 with a 256 bit bus and 448.0 GB/s of bandwidth. The CMP 40HX has 8 GB, the W5700X has 16 GB. The AMD card's doubled capacity gives it a clear advantage for large datasets that exceed 8 GB, even though raw bandwidth is equal.

Physical and power characteristics also differ. The CMP 40HX is a dual-slot card, 229 mm long, 111 mm tall, and 35 mm wide, with a 185 W TDP and a single 8-pin power connector. The W5700X is a quad-slot card, 305 mm long, with a 205 W TDP and no listed power connectors because it uses the Apple MPX bus interface. The CMP 40HX has no display outputs, while the W5700X provides 1x HDMI 2.0b and 4x Thunderbolt outputs. The NVIDIA card is a mining part, the AMD card is a Mac workstation part, and the interface choices reflect that split.

The Verdict

The data points to a clear split by workload and platform. For raw compute in OpenCL and Vulkan, the NVIDIA CMP 40HX is the stronger card by a wide margin. Its 113.2% OpenCL lead and 72.1% Vulkan lead are too large to ignore. Anyone running these types of workloads on a compatible system should favor the CMP 40HX based on the recorded measurements.

The AMD Radeon Pro W5700X is the better choice when memory capacity and display output matter. It has 16 GB of GDDR6 versus 8 GB, and it provides HDMI and Thunderbolt outputs. The CMP 40HX has no display outputs at all, making it unsuitable for any workstation use that requires a monitor connection. The W5700X also uses the Apple MPX bus interface, which makes it the natural fit for Mac systems, while the CMP 40HX uses a PCIe 1.0 x4 interface, an odd and limiting choice for a modern card.

The CMP 40HX sits at the 93rd percentile of all GPUs, the W5700X at the 87th. If the question is which card is faster in the database's aggregate measure, the answer is the CMP 40HX. If the question is which card fits a Mac Pro workstation with 16 GB of memory and full display connectivity, the answer is the W5700X. The benchmark results do not support a single universal winner; they support a use-case-based decision.

FAQ

Q: Which card has the higher average benchmark score?

A: The NVIDIA CMP 40HX, with an average score of 85,637 compared to the AMD Radeon Pro W5700X's 54,828.

Q: How large is the OpenCL performance gap?

A: The CMP 40HX scores 93,395 versus 43,810 in Geekbench OpenCL, a 113.2% lead.

Q: Does the AMD card have any benchmark that the NVIDIA card lacks?

A: Yes, the W5700X has a Geekbench Metal score of 75,427. The CMP 40HX has no recorded Metal benchmark.

Q: Which card has more memory?

A: The AMD Radeon Pro W5700X has 16 GB of GDDR6, while the NVIDIA CMP 40HX has 8 GB.

Q: Do both cards support the same DirectX version?

A: No, the CMP 40HX supports DirectX 12 Ultimate (12_2), while the W5700X supports DirectX 12 (12_1).

Q: Can the CMP 40HX drive a display?

A: No, the CMP 40HX has no display outputs, while the W5700X offers 1x HDMI 2.0b and 4x Thunderbolt.

Where Each One Wins

The NVIDIA CMP 40HX wins in compute-bound workloads that rely on OpenCL or Vulkan. Its 93,395 OpenCL score and 77,879 Vulkan score dwarf the W5700X's 43,810 and 45,246 respectively. It also has hardware features the AMD card lacks: 36 RT cores and 288 tensor cores, which make it more capable in ray tracing and tensor-accelerated tasks, even though no benchmark in the database directly measures those. The CMP 40HX also supports DirectX 12 Ultimate, giving it a forward-looking API profile.

The AMD Radeon Pro W5700X wins in memory capacity and workstation integration. Its 16 GB frame buffer is double the CMP 40HX's 8 GB, which matters for large models, high-resolution textures, or multi-application workflows. Its display outputs, including 4x Thunderbolt, make it a practical card for a Mac Pro environment. The Apple MPX bus interface means it is designed for that ecosystem, and the card's 205 W TDP with a 550 W suggested PSU reflects a workstation power envelope. The CMP 40HX, with no outputs and a PCIe 1.0 x4 interface, cannot serve that role.

The W5700X also holds an advantage in transistor efficiency. Built on 7 nm, it achieves 41.0 million transistors per square millimeter versus the CMP 40HX's 24.3 million, and it does so with a smaller die at 251 mm² versus 445 mm². For power-constrained or space-constrained systems that can accommodate a quad-slot card, the AMD part is the more modern silicon.

Specification Differences

The two cards differ across nearly every major specification category. The CMP 40HX uses the TU106 chip on Turing architecture at 12 nm, while the W5700X uses Navi 10 on RDNA 1.0 at 7 nm. Transistor counts are close, 10,800 million versus 10,300 million, but die size is not: 445 mm² versus 251 mm². Transistor density favors the AMD card at 41.0M per square millimeter versus 24.3M.

Clock speeds differ in both base and boost. The CMP 40HX runs at 1470 MHz base and 1650 MHz boost. The W5700X runs at 1243 MHz base and 2040 MHz boost. Memory clocks are identical at 1750 MHz, 14 Gbps effective, with both cards using GDDR6 on a 256 bit bus for 448.0 GB/s of bandwidth.

Shading units favor AMD: 2560 versus 2304. TMUs also favor AMD: 160 versus 144. ROPs are tied at 64. The CMP 40HX has 36 RT cores and 288 tensor cores, while the W5700X has none. Pixel rate favors the W5700X at 130.6 GPixel/s versus 105.6 GPixel/s. Texture rate favors the W5700X at 326.4 GTexel/s versus 237.6 GTexel/s. FP32 compute favors the W5700X at 10.44 TFLOPS versus 7.603 TFLOPS, and FP16 favors the W5700X at 20.89 TFLOPS versus 15.21 TFLOPS.

TDP differs by 20 W: 185 W for the CMP 40HX, 205 W for the W5700X. The suggested PSU is 450 W for the NVIDIA card and 550 W for the AMD card. The CMP 40HX is dual-slot with a 1x 8-pin connector; the W5700X is quad-slot with no listed power connectors. The CMP 40HX has no display outputs; the W5700X has 1x HDMI 2.0b and 4x Thunderbolt. Bus interfaces differ entirely: PCIe 1.0 x4 for the NVIDIA card, Apple MPX for the AMD card. Dimensions diverge as well: the CMP 40HX is 229 mm long, 111 mm tall, and 35 mm wide, while the W5700X is 305 mm long with no recorded height or width. Both cards are end-of-life, with the CMP 40HX released on 2021-02-24 and the W5700X on 2019-12-10.