AMD Radeon RX Vega 64 vs NVIDIA CMP 90HX Comparison

Where Each One Wins

The recorded data shows a clear but narrow split between these two cards, with the NVIDIA CMP 90HX taking the only shared benchmark victory. In the Geekbench OpenCL test, the CMP 90HX scores 69,000 against the RX Vega 64's 62,552, a 10.3% advantage. This is the single head-to-head measurement available, and it favors NVIDIA decisively.

However, the AMD Radeon RX Vega 64 holds its own in other API-level workloads that the database records for it alone. The Vega 64 posts a Geekbench Metal score of 68,750 and a Geekbench Vulkan score of 67,032, numbers that sit remarkably close to the CMP 90HX's OpenCL result. The Vega 64 also has a 3DMark Steel Nomad DX12 score of 1,669, a modern rasterization test that the CMP 90HX does not appear in. This suggests the AMD card can still deliver competitive results in contemporary DirectX 12 workloads, even though it lacks a direct comparison in that specific benchmark.

The use-case split becomes clearer when considering the broader benchmark profile. The CMP 90HX is a mining-oriented GPU with no display outputs, so its entire design targets compute throughput. The data reflects this: its single recorded benchmark is a compute-heavy OpenCL test, and its percentile ranking sits at 90 among all GPUs. The Vega 64, by contrast, is a full consumer graphics card with display outputs, so its benchmark suite spans Metal, Vulkan, and DX12, covering both compute and graphics scenarios. Users needing a card for actual rendering workloads across multiple APIs would find the Vega 64's broader test coverage more relevant, while raw compute performance in OpenCL belongs to the CMP 90HX.

Looking at the percentile rankings, the CMP 90HX sits at the 90th percentile among all GPUs, while the Vega 64 sits at the 86th. The CMP 90HX's average benchmark score is 69,000, while the Vega 64's average is 50,001, a gap driven partly by the fact that the Vega 64's average includes its lower DX12 score of 1,669, which drags the mean down significantly. In practice, the OpenCL comparison remains the only apples-to-apples measurement, and there the NVIDIA card wins by a comfortable margin.

Architecture Differences

The two GPUs come from fundamentally different design philosophies. The NVIDIA CMP 90HX uses the GA102 chip built on Ampere architecture, fabricated on Samsung's 8 nm process. It packs 28,300 million transistors into a 628 mm² die, yielding a transistor density of 45.1 million per square millimeter. The AMD Radeon RX Vega 64 uses the Vega 10 chip with GCN 5.0 architecture, built on GlobalFoundries' 14 nm process. It contains 12,500 million transistors on a 495 mm² die, for a density of 25.3 million per square millimeter. The NVIDIA chip is both larger and denser, reflecting a newer process node and a more complex design.

The compute resources differ sharply. The CMP 90HX has 6,400 shading units, 200 texture mapping units, and 80 render output units. It also includes 50 RT cores and 200 tensor cores, making it a fully-featured Ampere design despite its mining-oriented product positioning. The Vega 64 has 4,096 shading units, 256 TMUs, and 64 ROPs. It has no RT cores and no tensor cores, as it predates hardware ray tracing acceleration in AMD's consumer lineup. The texture rate favors AMD slightly at 395.8 GTexel/s versus 342.0 GTexel/s, but the pixel rate favors NVIDIA at 136.8 GPixel/s versus 98.94 GPixel/s.

Memory architecture is another major divergence. The CMP 90HX uses 10 GB of GDDR6X on a 320-bit bus, delivering 760.3 GB/s of bandwidth. The Vega 64 uses 8 GB of HBM2 on a 2048-bit bus, delivering 483.8 GB/s. The NVIDIA card has both more capacity and significantly higher bandwidth. The Vega 64's HBM2 implementation uses a much wider bus to compensate for its lower memory clock, but the raw bandwidth still falls behind by roughly 57%.

Clock speeds and power behavior also differ. The CMP 90HX runs at a 1500 MHz base clock and 1710 MHz boost, while the Vega 64 runs at 1247 MHz base and 1546 MHz boost. The NVIDIA card draws 320 W TDP with a suggested 700 W PSU, while the Vega 64 draws 295 W with a suggested 600 W PSU. Both are dual-slot cards with 2x 8-pin power connectors. The CMP 90HX has no display outputs, while the Vega 64 offers 1x HDMI 2.0b and 3x DisplayPort 1.4a. The CMP 90HX also uses a PCIe 1.0 x4 bus interface, which is oddly limited for a modern GPU, while the Vega 64 uses PCIe 3.0 x16. The NVIDIA card supports DirectX 12 Ultimate (12_2), while the Vega 64 supports DirectX 12 (12_1). Both support OpenGL 4.6, but the CMP 90HX supports Vulkan 1.4 versus the Vega 64's Vulkan 1.3.

Head-to-Head Benchmarks

The only direct comparison in the database is the Geekbench OpenCL test. The NVIDIA CMP 90HX scores 69,000, and the AMD Radeon RX Vega 64 scores 62,552. The delta is 10.3% in favor of the CMP 90HX. This is a meaningful margin in compute workloads, and it aligns with the architectural advantages the NVIDIA card holds: more shading units, higher clocks, and substantially more memory bandwidth.

Looking at the nearest rival data provides context for both cards. The CMP 90HX's closest competitors include the Intel Arc A770 at 68,809 (0.3% behind), the AMD Radeon Instinct MI25 at 68,562 (0.6% behind), the AMD Radeon Pro WX 8200 at 69,870 (1.2% ahead), and the NVIDIA Quadro P6000 at 69,986 (1.4% ahead). This places the CMP 90HX in a tight cluster of professional and high-end consumer GPUs, all within roughly 1.5% of each other. The Vega 64's nearest rivals include the NVIDIA GeForce RTX 5070 Ti at 49,957 (0.1% ahead), the Intel Arc A550M at 49,737 (0.5% ahead), the AMD Radeon RX 6900 XT at 50,951 (1.9% ahead), and the AMD Radeon RX 6800 XT at 48,477 (3.1% behind). These rival scores are based on the Vega 64's average benchmark score of 50,001, which includes the low DX12 result. In the OpenCL test specifically, the Vega 64's 62,552 is well above its average, indicating that the GPU performs much better in compute-heavy OpenCL workloads than its overall average suggests.

The data indicates that in OpenCL compute, the CMP 90HX is the stronger card, and by a margin that would be noticeable in real workloads. The Vega 64, however, shows competitive Metal and Vulkan scores that suggest it remains viable in those API environments, even if no direct head-to-head data exists for those tests.

FAQ

Q: Which card is faster in OpenCL compute?

A: The NVIDIA CMP 90HX scores 69,000 in Geekbench OpenCL versus the AMD Radeon RX Vega 64's 62,552, a 10.3% advantage for the NVIDIA card.

Q: Does the AMD Radeon RX Vega 64 support hardware ray tracing?

A: No. The Vega 64 has no RT cores, while the NVIDIA CMP 90HX includes 50 RT cores as part of its Ampere architecture.

Q: How do the memory configurations compare?

A: The CMP 90HX has 10 GB of GDDR6X on a 320-bit bus with 760.3 GB/s bandwidth. The Vega 64 has 8 GB of HBM2 on a 2048-bit bus with 483.8 GB/s bandwidth. The NVIDIA card offers more capacity and higher bandwidth.

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

A: No. The CMP 90HX has no display outputs. The Vega 64 has 1x HDMI 2.0b and 3x DisplayPort 1.4a.

Q: What is the power draw difference?

A: The CMP 90HX has a 320 W TDP with a suggested 700 W PSU. The Vega 64 has a 295 W TDP with a suggested 600 W PSU. Both use 2x 8-pin power connectors.

Q: How do the DirectX and Vulkan versions compare?

A: The CMP 90HX supports DirectX 12 Ultimate (12_2) and Vulkan 1.4. The Vega 64 supports DirectX 12 (12_1) and Vulkan 1.3.

Specification Differences

| Specification | NVIDIA CMP 90HX | AMD Radeon RX Vega 64 |

|---|---|---|

| Architecture | Ampere | GCN 5.0 |

| Process Node | 8 nm | 14 nm |

| Foundry | Samsung | GlobalFoundries |

| Transistors | 28,300 million | 12,500 million |

| Die Size | 628 mm² | 495 mm² |

| Transistor Density | 45.1M / mm² | 25.3M / mm² |

| Base Clock | 1500 MHz | 1247 MHz |

| Boost Clock | 1710 MHz | 1546 MHz |

| Memory Size | 10 GB | 8 GB |

| Memory Type | GDDR6X | HBM2 |

| Memory Bus Width | 320 bit | 2048 bit |

| Memory Bandwidth | 760.3 GB/s | 483.8 GB/s |

| Shading Units | 6400 | 4096 |

| TMUs | 200 | 256 |

| ROPs | 80 | 64 |

| RT Cores | 50 | None |

| Tensor Cores | 200 | None |

| Pixel Rate | 136.8 GPixel/s | 98.94 GPixel/s |

| Texture Rate | 342.0 GTexel/s | 395.8 GTexel/s |

| FP32 Performance | 21.89 TFLOPS | 12.66 TFLOPS |

| FP16 Performance | 21.89 TFLOPS (1:1) | 25.33 TFLOPS (2:1) |

| TDP | 320 W | 295 W |

| Suggested PSU | 700 W | 600 W |

| Bus Interface | PCIe 1.0 x4 | PCIe 3.0 x16 |

| Display Outputs | No outputs | 1x HDMI 2.0b, 3x DisplayPort 1.4a |

| DirectX | 12 Ultimate (12_2) | 12 (12_1) |

| Vulkan | 1.4 | 1.3 |

| Dimensions | 285 mm x 112 mm | 280 mm x 111 mm x 40 mm |

| Release Date | 2021-07-27 | 2017-08-06 |

| Launch MSRP | None recorded | 499 USD |