DEPLOYDatabase

Chip comparison

NVIDIA H100 Tensor Core GPU vs NVIDIA L40S

Side-by-side specs

Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.

FieldNVIDIA H100 Tensor Core GPUNVIDIA L40S
Process nodeTSMC 4NTSMC 4N
Transistors (B)8076.3
Die size814 mm²
TDP700 W350 W
Memory80 GB HBM348 GB GDDR6
Memory bandwidth3,350 GB/s864 GB/s
FP16 (dense)989 TFLOPS362 TFLOPS
BF16 (dense)989 TFLOPS362 TFLOPS
FP8 (dense)1,979 TFLOPS733 TFLOPS
INT8 (dense)1,979 TOPS733 TOPS
Launch price (list)$30,000
Form factorSXM5PCIe
Announced2022-03-222023-08-08
Released2022-10-13

Robots running NVIDIA H100 Tensor Core GPU

No robots publicly running it yet.

Robots running NVIDIA L40S

No robots publicly running it yet.

Data centers with NVIDIA L40S

No data centers publicly running it yet.

Common questions

NVIDIA H100 Tensor Core GPU vs NVIDIA L40S: which is faster for training?

NVIDIA H100 Tensor Core GPU has 2.70x the dense FP8 throughput of the other (NVIDIA H100 Tensor Core GPU: 1,979 TFLOPS; NVIDIA L40S: 733 TFLOPS). Real-world training speed also depends on memory bandwidth, interconnect, and software stack: see the spec diff for those inputs.

NVIDIA H100 Tensor Core GPU vs NVIDIA L40S: which has more memory?

NVIDIA H100 Tensor Core GPU carries more HBM (NVIDIA H100 Tensor Core GPU: 80 GB HBM3; NVIDIA L40S: 48 GB GDDR6). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.

NVIDIA H100 Tensor Core GPU vs NVIDIA L40S: which is more power-efficient?

Dense FP8 TFLOPS per watt: NVIDIA H100 Tensor Core GPU 2.83 (1,979 ÷ 700 W); NVIDIA L40S 2.09 (733 ÷ 350 W). NVIDIA H100 Tensor Core GPU wins at the die level; system-level efficiency also depends on cooling and interconnect.

NVIDIA H100 Tensor Core GPU vs NVIDIA L40S: which is more widely used?

NVIDIA H100 Tensor Core GPU: 0 robots and 7 data centers publicly running it. NVIDIA L40S: 0 robots and 0 data centers. Empty here means no operator has said publicly, not zero adoption in the market.

See also: every chip comparison · NVIDIA H100 Tensor Core GPU full page · NVIDIA L40S full page.