NVIDIA H800 PCIe 80GB vs NVIDIA H100 SXM5 80GB

The H100 delivers 31% more FP8 throughput than the H800 (1,979 vs 1,513 TFLOPS dense).

H800: Hopper, 2022 H100: Hopper, 2022
FP8 dense lead
+31%
H100: 1,979 vs 1,513 TFLOPS
VRAM
80 vs 80 GB
Same capacity
Bandwidth
2.0 TB/s vs 3.4 TB/s
H100 moves data faster
TDP
350 W vs 700 W
H800 draws 350 W less

Specifications

H800
H100
Architecture
HopperHopper
Launch Year
20222022
Form Factor
PCIeSXM
VRAM
80 GB80 GB
Memory Bandwidth
2.0 TB/s3.4 TB/s
TDP
350 W700 W
Process Node
4nm4nm

Performance (TFLOPS)

H800
H100
FP64
0.8 TFlops 34.0 TFlops
FP32
51.0 TFlops 67.0 TFlops
FP16
756.5 TFlops
sparse not published
990.0 TFlops
1979.0 TFLOPS sparse
BF16
756.5 TFlops
sparse not published
990.0 TFlops
1979.0 TFLOPS sparse
FP8
1513.0 TFlops
sparse not published
1979.0 TFlops
3958.0 TFLOPS sparse
INT8
1513.0 TFlops
sparse not published
1979.0 TFlops
3958.0 TOPS sparse
Tensor Core
FP64 (TC)
67.0 TFlops

FLOPS by Precision

What actually differs

Both GPUs launched in 2022: the H800 on Hopper and the H100 on Hopper.

Dense throughput for the H800 against the H100: FP64 0.8 vs 34 TFLOPS, FP32 51 vs 67 TFLOPS, FP16 757 vs 990 TFLOPS, FP8 1,513 vs 1,979 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 80 GB against 80 GB, fed at 2.0 TB/s versus 3.4 TB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

Power budgets are 350 W for the H800 and 700 W for the H100. At FP32 that works out to 0.15 against 0.10 TFLOPS per watt.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.

Frequently asked questions

Is the H800 faster than the H100?

At FP8 precision the H100 reaches 1,979 TFLOPS dense against 1,513 TFLOPS for the H800. The performance table on this page lists every published precision for both GPUs.

Which has more memory, the H800 or the H100?

Both GPUs carry 80 GB of VRAM. Memory bandwidth is 2.0 TB/s for the H800 and 3.4 TB/s for the H100.

How much power do the H800 and the H100 draw?

The H800 is rated at 350 W TDP and the H100 at 700 W. On FP32 throughput per watt, the H800 is the more efficient part.

Can I rent the H800 or the H100 in the cloud?

Yes. Live cloud listings tracked by Flopper start at $1.63 per GPU hour. The rental pricing section on this page lists current providers and rates for both GPUs.

Where to Rent

NVIDIA H100 SXM5 80GB

ProviderConfigurationPrice/GPU-hrChecked
DataCrunch
1× H100 SXM5 Spot $1.63 2h ago View →
Cudo Compute
1× H100 SXM5 $1.79 2h ago View →
Nebius
1× H100 SXM5 Spot $2.15 2h ago View →
CoreWeave
8× H100 SXM5 Spot $2.46 2h ago View →
RunPod
1× H100 SXM5 Community $2.69 2h ago View →
Massed Compute
2× H100 SXM5 $2.73 2h ago View →
Together AI
1× H100 SXM5 Reserved $3.19 2h ago View →
Hyperstack
1× H100 SXM5 $3.20 2h ago View →
DataCrunch
1× H100 SXM5 $3.25 2h ago View →
RunPod
1× H100 SXM5 $3.29 2h ago View →
AceCloud
1× H100 SXM5 $3.63 2h ago View →
Scaleway
2× H100 SXM5 $3.84 2h ago View →
Nebius
1× H100 SXM5 $3.85 2h ago View →
Crusoe
1× H100 SXM5 $3.90 2h ago View →
Together AI
1× H100 SXM5 $3.99 2h ago View →
Lambda Labs
1× H100 SXM5 $4.36 2h ago View →
DigitalOcean
1× H100 SXM5 $4.41 2h ago View →
DigitalOcean
8× H100 SXM5 $4.41 2h ago View →
Paperspace
1× H100 SXM5 $5.99 2h ago View →
CoreWeave
8× H100 SXM5 $6.16 2h ago View →
Oracle Cloud (OCI)
1× H100 SXM5 $10.75 2h ago View →
Microsoft Azure
8× H100 SXM5 $11.06 2h ago View →
Microsoft Azure
8× H100 SXM5 $11.06 2h ago View →
Microsoft Azure
8× H100 SXM5 $11.06 2h ago View →
Microsoft Azure
8× H100 SXM5 $11.06 2h ago View →
Microsoft Azure
8× H100 SXM5 $12.29 2h ago View →
Microsoft Azure
8× H100 SXM5 $12.29 2h ago View →
All H100 listings and price history →

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.