H100 PCIe vs CMP 90HX
Contents
Primary details
GPU architecture, market segment, value for money and other general parameters compared.
| Place in the ranking | not rated | not rated |
| Place by popularity | not in top-100 | not in top-100 |
| Architecture | Ampere (2020−2025) | Hopper (2022−2024) |
| GPU code name | GA102 | GH100 |
| Market segment | Workstation | Workstation |
| Release date | 28 July 2021 (4 years ago) | 22 March 2022 (4 years ago) |
Detailed specifications
General parameters such as number of shaders, GPU core base clock and boost clock speeds, manufacturing process, texturing and calculation speed. Note that power consumption of some graphics cards can well exceed their nominal TDP, especially when overclocked.
| Pipelines / CUDA cores | 6400 | 7296 |
| Core clock speed | 1500 MHz | 1065 MHz |
| Boost clock speed | 1710 MHz | 1650 MHz |
| Number of transistors | 28,300 million | 80,000 million |
| Manufacturing process technology | 8 nm | 4 nm |
| Power consumption (TDP) | 320 Watt | 350 Watt |
| Texture fill rate | 342.0 | 752.4 |
| Floating-point processing power | 21.89 TFLOPS | 24.08 TFLOPS |
| ROPs | 80 | 24 |
| TMUs | 200 | 456 |
| Tensor Cores | 200 | 456 |
| Ray Tracing Cores | 50 | no data |
| L1 Cache | 6.3 MB | 21.4 MB |
| L2 Cache | 5 MB | 50 MB |
Form factor & compatibility
Information on compatibility with other computer components. Useful when choosing a future computer configuration or upgrading an existing one. For desktop graphics cards it's interface and bus (motherboard compatibility), additional power connectors (power supply compatibility).
| Interface | PCIe 1.0 x4 | PCIe 5.0 x16 |
| Length | 285 mm | 267 mm |
| Width | 2-slot | 2-slot |
| Supplementary power connectors | 2x 8-pin | 8-pin EPS |
VRAM capacity and type
Parameters of VRAM installed: its type, size, bus, clock and resulting bandwidth. Integrated GPUs have no dedicated video RAM and use a shared part of system RAM.
| Memory type | GDDR6X | HBM2e |
| Maximum RAM amount | 10 GB | 80 GB |
| Memory bus width | 320 Bit | 5120 Bit |
| Memory clock speed | 1188 MHz | 1000 MHz |
| Memory bandwidth | 760.3 GB/s | 1,280 GB/s |
| Resizable BAR | + | + |
Connectivity and outputs
This section shows the types and number of video connectors on each GPU. The data applies specifically to desktop reference models (for example, NVIDIA’s Founders Edition). OEM partners often modify both the number and types of ports. On notebook GPUs, video‐output options are determined by the laptop’s design rather than the graphics chip itself.
| Display Connectors | No outputs | No outputs |
API and SDK support
List of supported 3D and general-purpose computing APIs, including their specific versions.
| DirectX | 12 Ultimate (12_2) | N/A |
| Shader Model | 6.8 | N/A |
| OpenGL | 4.6 | N/A |
| OpenCL | 3.0 | 3.0 |
| Vulkan | 1.3 | N/A |
| CUDA | 8.6 | 9.0 |
| DLSS | + | + |
Synthetic benchmarks
Non-gaming benchmark results comparison. The combined score is measured on a 0-100 point scale.
GeekBench 5 OpenCL
Geekbench 5 is a widespread graphics card benchmark combined from 11 different test scenarios. All these scenarios rely on direct usage of GPU's processing power, no 3D rendering is involved. This variation uses OpenCL API by Khronos Group.
Pros & cons summary
| Recency | 28 July 2021 | 22 March 2022 |
| Maximum RAM amount | 10 GB | 80 GB |
| Chip lithography | 8 nm | 4 nm |
| Power consumption (TDP) | 320 Watt | 350 Watt |
CMP 90HX has 9% lower power consumption.
H100 PCIe, on the other hand, has an age advantage of 7 months, a 700% higher maximum VRAM amount, and a 100% more advanced lithography process.
We couldn't decide between CMP 90HX and H100 PCIe. We've got no test results to judge.
Other comparisons
We selected several comparisons of graphics cards with performance close to those reviewed, providing you with more options to consider.
