Welcome to the GPU Specifications for AI Projects repository!
This project aims to centralize detailed specifications for GPUs, particularly in the context of AI workloads.
Finding comprehensive and accurate GPU specifications for AI can be challenging. Many sources are either incomplete, inconsistent, or unavailable. This repository consolidates information into a single, accessible resource to assist AI researchers, developers, and enthusiasts.
If you believe this project duplicates an existing effort, or if a similar database already exists, please let us know by opening an issue or contacting us directly. Collaboration or linking efforts can benefit the entire community!
The core of this project is a JSON database containing detailed specifications of GPUs:
- Precision support (e.g., FP32, FP16, INT8, etc.)
- Compute & Memory capabilities
- Additional details relevant to AI workloads
The JSON data is rendered into:
If you spot errors, have missing data, or can add reliable sources, we welcome your input! See the CONTRIBUTING.md file for details.
- Add missing specifications for GPUs.
- Expand support to cover GPUs from various vendors (NVIDIA, AMD, etc.).
Below is a sample of the GPU specifications data included in the repository:
| Attribute (Unit) | H100 | L40S | A100 |
|---|---|---|---|
| FP64 (TFLOPS) | 25.6 | 1.4 | 9.7 |
| FP64 Tensor Core (TFLOPS) | 51 | ? | 19.5 |
| FP32 (TFLOPS) | 51.2 | 91.6 | 19.5 |
| TF32 Tensor Core (TFLOPS) | ? | 183 | 156 |
| TF32 Tensor Core with Sparsity (TFLOPS) | 756 | 366 | 312 |
| FP16 (TFLOPS) | 204.9 | 91.6 | 78 |
| FP16 Tensor Core (TFLOPS) | ? | 362 | 312 |
| FP16 Tensor Core with Sparsity (TFLOPS) | ? | 733 | 624 |
| BF16 (TFLOPS) | ? | ? | ? |
| BF16 Tensor Core (TFLOPS) | ? | 362 | 312 |
| BF16 Tensor Core with Sparsity (TFLOPS) | 1513 | 733 | 624 |
| FP8 (TFLOPS) | N/A | N/A | N/A |
| FP8 Tensor Core (TFLOPS) | ? | 733 | N/A |
| FP8 Tensor Core with Sparsity (TFLOPS) | 3026 | 1466 | N/A |
| INT8 (TOPS) | ? | ? | ? |
| INT8 Tensor Core (TOPS) | ? | 733 | 624 |
| INT8 Tensor Core with Sparsity (TOPS) | 3026 | 1466 | 1248 |
| INT4 (TOPS) | N/A | N/A | N/A |
| INT4 Tensor Core (TOPS) | ? | 733 | ? |
| INT4 Tensor Core with Sparsity (TOPS) | ? | 1466 | ? |
| Architecture Details | |||
| GPU Name | H100 | L40S | A100 |
| Manufacturer | NVIDIA | NVIDIA | NVIDIA |
| Architecture | Hopper | Ada Lovelace | Ampere |
| NVIDIA RT Cores | ? | 142 (3rd gen) | ? |
| NVIDIA Tensor Cores | 456 (4th gen) | 568 (4th gen) | 432 (3rd gen) |
| NVIDIA CUDA Cores | 14592 | 18176 | 6912 |
| GPU Memory (GB) | 80 | 48 | 80 |
| Memory Bandwidth (GB/s) | 2048 | 864 | 1935 |
| Interconnect Type | PCIe Gen5 | PCIe Gen4 | PCIe Gen4 |
| Encoders and Decoders | 0, 7 | 3, 3 | 0, 5 |
| CUDA Compute Capability | 9 | 8.9 | 8 |