AMD Radeon Pro Vega II Duo vs NVIDIA CMP 90HX Comparison

Where Each One Wins

The recorded benchmark data presents a clear, though limited, competitive picture between these two very different graphics cards. The AMD Radeon Pro Vega II Duo and the NVIDIA CMP 90HX were designed for entirely separate purposes, and the single shared benchmark in the database reflects that divergence.

The AMD Radeon Pro Vega II Duo is the definitive winner in the only head-to-head comparison available. In the Geekbench OpenCL test, it achieves a score of 98432, while the NVIDIA CMP 90HX manages 69000. This represents a 42.7% advantage for the AMD card, a substantial margin that speaks to its professional compute capabilities. The AMD card also holds a higher overall percentile ranking at 94, compared to the NVIDIA card's 90, indicating that across the entire database of GPUs, the Radeon Pro Vega II Duo sits higher in the performance distribution.

However, the NVIDIA CMP 90HX has its own domain of superiority, though it is not measured by the benchmark scores in this database. The CMP 90HX is a dedicated mining GPU, as indicated by its "Mining GPUs" generation and its complete lack of display outputs. Its architecture is optimized for the specific workloads of cryptocurrency hashing, not general-purpose compute or graphics rendering. The data shows that it has a higher raw FP32 throughput at 21.89 TFLOPS compared to the AMD card's 14.09 TFLOPS. This suggests that in pure compute tasks that are not memory-bandwidth bound, the NVIDIA card could outperform the AMD card, but this is not reflected in the Geekbench OpenCL score, where the AMD card's massive memory bandwidth likely plays a decisive role.

The AMD card wins in scenarios that leverage its enormous memory subsystem and its professional feature set. Its 1.02 TB/s bandwidth and 32 GB of HBM2 memory are ideal for large datasets, machine learning inference, and high-resolution texture workloads. The NVIDIA card, with its 10 GB GDDR6X memory and 760.3 GB/s bandwidth, is better suited for workloads that fit within that capacity but require high-throughput integer math, which is common in mining operations. The data indicates a fundamental split: the AMD card is a general-purpose professional compute powerhouse, while the NVIDIA card is a specialized, single-purpose accelerator.

Architecture Differences

The architectural chasm between these two cards is vast, starting with the fundamental design philosophy. The AMD Radeon Pro Vega II Duo is built on the Vega 20 chip using the GCN 5.1 architecture, fabricated on a 7 nm process at TSMC. It packs 13,230 million transistors onto a 331 mm² die, resulting in a transistor density of 40.0M per mm². This is a dual-die design, as the "Duo" in its name suggests, though the database records it as a single entity with 4096 shading units, 256 texture mapping units (TMUs), and 64 render output units (ROPs).

In contrast, the NVIDIA CMP 90HX is built on the GA102 chip using the Ampere architecture, fabricated on an 8 nm process at Samsung. It contains a staggering 28,300 million transistors on a 628 mm² die, giving it a slightly higher transistor density of 45.1M per mm². The NVIDIA card features 6400 shading units, 200 TMUs, and 80 ROPs. Crucially, the Ampere architecture includes dedicated hardware that the Vega architecture lacks entirely: 50 RT cores for ray tracing and 200 tensor cores for AI acceleration. The AMD card has null values for both RT and tensor cores, confirming its older, compute-oriented design.

The memory architectures are equally divergent. The AMD card uses 32 GB of HBM2 on a 4096-bit bus, achieving a bandwidth of 1.02 TB/s. The NVIDIA card uses 10 GB of GDDR6X on a 320-bit bus, achieving 760.3 GB/s. The AMD card's memory clock is listed as 1000 MHz with 2 Gbps effective, while the NVIDIA card runs at 1188 MHz with 19 Gbps effective. The bus interface also differs dramatically: the AMD card uses Apple MPX, a proprietary connector for Apple Mac Pro systems, while the NVIDIA card uses PCIe 1.0 x4, which is a legacy interface likely chosen to limit its use in standard gaming PCs and push it toward mining rigs. The NVIDIA card has no display outputs, while the AMD card offers 1x HDMI 2.0b and 4x Thunderbolt.

Head-to-Head Benchmarks

The database contains only one direct comparison between these two cards: the Geekbench OpenCL test. In this benchmark, the AMD Radeon Pro Vega II Duo scores 98432, while the NVIDIA CMP 90HX scores 69000. This is a decisive 42.7% victory for the AMD card. To put this in context, the AMD card's average benchmark score across all tests in the database is 106750, while the NVIDIA card's average is 69000, which is lower than its OpenCL score alone because the AMD card has three benchmarks recorded (Geekbench Metal at 122522 and Geekbench Vulkan at 99296) that pull its average up.

Looking at the nearest rivals for each card provides further context. The AMD Radeon Pro Vega II Duo sits within a tight cluster of professional cards. It is 0.6% behind the AMD Radeon Pro W6600X (which scores 107342), 2.6% behind the AMD Radeon Pro Vega II (109617), and 3.6% behind the AMD Radeon PRO W7900 (110725). However, it is 4.8% ahead of the NVIDIA Quadro RTX 6000 (101872). This shows that the Duo, despite its age, remains competitive with newer professional cards in this specific benchmark suite.

The NVIDIA CMP 90HX is positioned very differently. It is 0.3% ahead of the Intel Arc A770 (68809) and 0.6% ahead of the AMD Radeon Instinct MI25 (68562). However, it trails the AMD Radeon Pro WX 8200 by 1.2% (which scores 69870) and the NVIDIA Quadro P6000 by 1.4% (69986). This places the CMP 90HX in a mid-range performance tier, far below the professional compute cards in the AMD Duo's neighborhood. The 42.7% delta in the head-to-head test is consistent with this positioning: the AMD card operates in a completely different performance class.

Specification Differences

The most striking specification differences between the two cards are as follows:

• Process Node: AMD uses 7 nm (TSMC), NVIDIA uses 8 nm (Samsung).
• Transistors: AMD has 13,230 million, NVIDIA has 28,300 million.
• Die Size: AMD is 331 mm², NVIDIA is 628 mm².
• Shading Units: AMD has 4096, NVIDIA has 6400.
• TMUs: AMD has 256, NVIDIA has 200.
• ROPs: AMD has 64, NVIDIA has 80.
• RT Cores: AMD has none, NVIDIA has 50.
• Tensor Cores: AMD has none, NVIDIA has 200.
• Memory Size: AMD has 32 GB, NVIDIA has 10 GB.
• Memory Type: AMD uses HBM2, NVIDIA uses GDDR6X.
• Memory Bus Width: AMD is 4096-bit, NVIDIA is 320-bit.
• Memory Bandwidth: AMD is 1.02 TB/s, NVIDIA is 760.3 GB/s.
• Base Clock: AMD is 1400 MHz, NVIDIA is 1500 MHz.
• Boost Clock: AMD is 1720 MHz, NVIDIA is 1710 MHz.
• FP32 Performance: AMD is 14.09 TFLOPS, NVIDIA is 21.89 TFLOPS.
• FP16 Performance: AMD is 28.18 TFLOPS (2:1), NVIDIA is 21.89 TFLOPS (1:1).
• TDP: AMD is 475 W, NVIDIA is 320 W.
• Slot Width: AMD is Quad-slot, NVIDIA is Dual-slot.
• Power Connectors: AMD has none listed, NVIDIA has 2x 8-pin.
• Suggested PSU: AMD is 850 W, NVIDIA is 700 W.
• Bus Interface: AMD uses Apple MPX, NVIDIA uses PCIe 1.0 x4.
• Display Outputs: AMD has 1x HDMI 2.0b and 4x Thunderbolt, NVIDIA has no outputs.
• DirectX Support: AMD supports 12 (12_1), NVIDIA supports 12 Ultimate (12_2).
• Vulkan Support: AMD supports 1.3, NVIDIA supports 1.4.
• Release Date: AMD is 2019-06-02, NVIDIA is 2021-07-27.

FAQ

Q: Which card has more raw compute power in FP32 operations?

A: The NVIDIA CMP 90HX has a higher FP32 throughput at 21.89 TFLOPS, compared to the AMD Radeon Pro Vega II Duo's 14.09 TFLOPS.

Q: Is the AMD card faster in the only shared benchmark?

A: Yes, the AMD Radeon Pro Vega II Duo scores 98432 in Geekbench OpenCL, which is 42.7% higher than the NVIDIA CMP 90HX's score of 69000.

Q: What are the dedicated hardware features on the NVIDIA card that the AMD card lacks?

A: The NVIDIA CMP 90HX has 50 RT cores and 200 tensor cores, while the AMD Radeon Pro Vega II Duo has none of these dedicated units.

Q: How does the memory configuration differ between the two cards?

A: The AMD card has 32 GB of HBM2 memory on a 4096-bit bus with 1.02 TB/s bandwidth, while the NVIDIA card has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth.

Q: Can the NVIDIA CMP 90HX be used for display output?

A: No, the NVIDIA CMP 90HX has no display outputs at all, whereas the AMD Radeon Pro Vega II Duo provides 1x HDMI 2.0b and 4x Thunderbolt ports.

Q: Which card has a higher overall ranking among all GPUs in the database?

A: The AMD Radeon Pro Vega II Duo is in the 94th percentile, while the NVIDIA CMP 90HX is in the 90th percentile, indicating the AMD card is better positioned overall.