NVIDIA H100 PCIe
Data center GPU · Hopper
Launched
Q4 2022
VRAM
80 GB
Mem. bandwidth
2,000 GB/s
On-demand from
Tech specs
VRAM 80 GB HBM2e
Memory bandwidth 2,000 GB/s
Interface PCIe
CUDA cores 14,592
Tensor cores 456 (Gen 4)
TDP 350 W
Supported data types
FP64FP32FP16BF16FP8INT8
Cloud rental prices
9 providers · available in 6 countries · lowest price per GPU, per hour
On-demand
$2.50 – $3.39 /GPU/h
Reserved
$1.18 – $3.88 /GPU/h
| Provider | On-demand | Reserved | Spot |
|---|---|---|---|
 DigitalOcean | $2.99 | — | — |
 Hyperstack | $2.50 | $1.75 | — |
 Lambda | $3.29 | — | — |
 Latitude.sh | $3.37 | $1.18 | — |
 LeaderGPU | — | $2.21 | — |
 Massed Compute | $2.73 | — | — |
 OVHcloud | $3.20 | — | — |
 Runpod | $2.89 | — | — |
| | $3.28 | — | — |
Models that fit in VRAM
Open-weights models that fit in VRAM. Estimated using 🤗 accelerate, plus approximation for up to 8K context.
Peak theoretical performance
FP8 Tensor Core 1,513 TFLOPS
INT8 Tensor Core 1,513 TOPS
BF16 Tensor Core 756 TFLOPS
FP16 Tensor Core 756 TFLOPS
TF32 Tensor Core 378 TFLOPS
FP32 51 TFLOPS
FP64 26 TFLOPS
FP64 Tensor Core 51 TFLOPS
Performance figures assume no sparsity; in cases where only sparse performance figures are published by the manufacturer, these are halved to give approximate dense performance.
Resources
Detailed documentation from the manufacturer.
Similar GPUs
Other accelerators you might compare.