Enterprise adoption of AI is now mainstream, and organizations need end-to-end, AI-ready infrastructure that will accelerate them into this new era.
H800 for mainstream servers comes with a five-year subscription, including enterprise support, to the NVIDIA AI Enterprise software suite, simplifying AI adoption with the highest performance. This ensures organizations have access to the AI frameworks and tools they need to build H800-accelerated AI workflows such as AI chatbots, recommendation engines, vision AI, and more.
Transformational AI training.
H800 features fourth-generation Tensor Cores and the Transformer Engine with FP8 precision that provides up to 9X faster training over the prior generation for mixture-of-experts (MoE) models. The combination of fourth-generation NVlink, which offers 900 gigabytes per second (GB/s) of GPU-to-GPU interconnect; NVLINK Switch System, which accelerates communication by every GPU across nodes; PCIe Gen5; and NVIDIA Magnum IO™ software delivers efficient scalability from small enterprises to massive, unified GPU clusters.
Deploying H800 GPUs at data center scale delivers outstanding performance and brings the next generation of exascale high-performance computing (HPC) and trillion-parameter AI within the reach of all researchers.
| Specification | Details |
|---|---|
| Graphics Processor | |
| GPU Name | GH100 |
| Architecture | Hopper |
| Foundry | TSMC |
| Process Size | 4 nm |
| Transistors | 80,000 million |
| Die Size | 814mm² |
| Graphics Card | |
| Release Date | Mar 2022 |
| Generation | Tesla Hopper |
| Type | Desktop |
| Bus Interface | PCIe 5.0 x16 |
| Clock Speeds | |
| Base Clock | 1095MHz |
| Boost Clock | 1755MHz |
| Memory Clock | 1500MHz |
| Memory | |
| Memory Size | 80GB |
| Memory Type | HBM3 |
| Memory Bus | 5120bit |
| Bandwidth | 1920 GB/s |
| Render Config | |
| Shading Units | 8448 |
| TMUs | 528 |
| ROPs | 24 |
| SM Count | 132 |
| Tensor Cores | 528 |
| L1 Cache | 192 KB (per SM) |
| L2 Cache | 50MB |
| Theoretical Performance | |
| Pixel Rate | 42.12 GPixel/s |
| Texture Rate | 926.6 GTexel/s |
| FP16 (half) | 118.6 TFLOPS |
| FP32 (float) | 29.65 TFLOPS |
| FP64 (double) | 14.83 TFLOPS |
| Board Design | |
| TDP | 700W |
| Suggested PSU | 1100W |
| Outputs | No outputs |
| Power Connectors | 8-pin EPS |
| Graphics Features | |
| DirectX | N/A |
| OpenGL | N/A |
| OpenCL | 3.0 |
| Vulkan | N/A |
| CUDA | 9.0 |
| Shader Model | N/A |
Reviews
There are no reviews yet.