DEPLOYDatabase

Chip comparison

Huawei Ascend 910D 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.

FieldHuawei Ascend 910DNVIDIA L40S
Process nodeSMIC N+2 (~7nm)TSMC 4N
Transistors (B)76.3
TDP800 W350 W
Memory192 GB HBM348 GB GDDR6
Memory bandwidth4,000 GB/s864 GB/s
FP16 (dense)1,200 TFLOPS362 TFLOPS
BF16 (dense)1,200 TFLOPS362 TFLOPS
FP8 (dense)2,400 TFLOPS733 TFLOPS
INT8 (dense)733 TOPS
Form factorOAMPCIe
Announced2025-04-012023-08-08
Released2025-12-01

Robots running Huawei Ascend 910D

No robots publicly running it yet.

Robots running NVIDIA L40S

No robots publicly running it yet.

Data centers with Huawei Ascend 910D

No data centers publicly running it yet.

Data centers with NVIDIA L40S

No data centers publicly running it yet.

Common questions

Huawei Ascend 910D vs NVIDIA L40S: which is faster for training?

Huawei Ascend 910D has 3.27x the dense FP8 throughput of the other (Huawei Ascend 910D: 2,400 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.

Huawei Ascend 910D vs NVIDIA L40S: which has more memory?

Huawei Ascend 910D carries more HBM (Huawei Ascend 910D: 192 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.

Huawei Ascend 910D vs NVIDIA L40S: which is more power-efficient?

Dense FP8 TFLOPS per watt: Huawei Ascend 910D 3.00 (2,400 ÷ 800 W); NVIDIA L40S 2.09 (733 ÷ 350 W). Huawei Ascend 910D wins at the die level; system-level efficiency also depends on cooling and interconnect.

See also: every chip comparison · Huawei Ascend 910D full page · NVIDIA L40S full page.