inference/research
draftlow · 0.60updated 2026-10-16

NVIDIA L40 (Ada, 48 GB)

nvidia ada (AD102) 2022
This record is a draft. Not yet verified. Provisional. Treat every number on this page as provisional.

Fields

id
accelerators/nvidia-l40
Vendor
nvidia
Architecture
ada (AD102)
Release year
2022
Form factors
pcie
VRAM (GB)
48
Memory type
gddr6 with ECC
Memory bus (bit)
384
Memory bandwidth (GB/s)
864
Bandwidth basis
GDDR6 18 Gbps x 384-bit = 864 GB/s (NVIDIA L40 datasheet)
TDP (W)
300
Unified memory
no
Consumer part
no
notes
Datacenter Ada inference GPU, 48 GB ECC GDDR6, 300 W, no NVLink. Distinct from the L40S (higher clocks, 48 GB, 350 W). Used-market part in the affordable-datacenter class.

Notes

Datacenter Ada inference GPU, 48 GB ECC GDDR6, 300 W, no NVLink. Distinct from the L40S (higher clocks, 48 GB, 350 W). Used-market part in the affordable-datacenter class.

Cross-references

Interconnect (1)