| |
| Product | NVIDIA® HGX™ A100 - 8x NVIDIA® A100 SXM4 GPUs - 640GB HBM2 Memory - 3rd Generation NVLink |
| Action | Select |
| Main Specifications |
| Product Series |
Nvidia A100 |
| Core Type |
NVIDIA TENSOR |
| Host Interface |
4/8 SXM on NVIDIA HGX™ A100 |
| GPU Architecture |
Ampere |
| Detailed Specifications |
| Max Memory Size |
80 GB |
| Max Memory Bandwidth |
2,039 GB/s |
| Peak FP64 |
9.7 TFLOPS |
| Peak FP64 Tensor Core |
19.5 TFLOPS |
| INT8 Tensor Core |
1,248 TOPS (with sparsity) |
| TF32 Tensor Core |
312 TFLOPS (with sparsity) |
| FP32 |
19.5 TFLOPS |
| Peak BFLOAT16 Tensor Core |
624 TFLOPS (with sparsity) |
| Peak FP16 Tensor Core |
624 TFLOPS (with sparsity) |
| Peak INT4 Tensor Core |
2,496 TOPS (with sparsity) |
| Total NVLink Bandwidth |
600 GB/s |
| Multi-Instance GPUs |
7 MIGs at 10GB |
| Cooling |
Passive |
| Max Graphics Card Power (W) |
400W |
| Action | Select |