Specifications
MPN
900-2G133-0010-000
Series
NVIDIA L40
Architecture
NVIDIA Ada Lovelace
GPU Memory
48 GB GDDR6 with ECC
Memory Bandwidth
864 GB/s
CUDA Cores
18,176
Tensor Cores
568 (4th generation)
RT Cores
142 (3rd generation)
FP32
90.5 TFLOPS
Tensor TF32
90.5 TFLOPS (181 sparse)
Tensor BF16 / FP16
181 TFLOPS (362 sparse)
Tensor FP8
362 TFLOPS (724 sparse)
INT8 Tensor TOPS
362 (724 sparse)
Interconnect
PCIe Gen4 x16 — 64 GB/s bidirectional
Form Factor
4.4" H x 10.5" L, dual slot
Thermal
Passive
Max Power
300 W
Power Connector
16-pin (12V-2x6)
Display Outputs
4x DisplayPort 1.4a
vGPU Support
Yes
NVLink
Not supported
NVIDIA L40 family comparison
| Product | Architecture | Memory | Memory Bandwidth | Max Power | Form Factor |
|---|---|---|---|---|---|
| NVIDIA L40S | Ada Lovelace | 48 GB GDDR6 ECC | 864 GB/s | 350 W | Dual slot, passive |
| NVIDIA L40 | Ada Lovelace | 48 GB GDDR6 ECC | 864 GB/s | 300 W | Dual slot, passive |
| NVIDIA A100 80GB PCIe | Ampere | 80 GB HBM2e | 1,935 GB/s | 300 W | Dual slot, passive |
Use Case
The L40 is the 300 W member of NVIDIA's Ada Lovelace professional line and the direct predecessor of the L40S. It keeps the same 48 GB GDDR6 ECC frame buffer and 864 GB/s of bandwidth but is tuned for visual computing and virtualisation first: it carries four DisplayPort 1.4a outputs, full vGPU support and the same 18,176-core AD102 die, at a 50 W lower board power than the L40S. It is the usual choice for NVIDIA Omniverse, digital-twin and 3D rendering farms, CUDA virtual workstations and inference of small language models where the extra Tensor Core budget of the L40S is not needed.
Need NVIDIA L40 48GB PCIe GPU Accelerator?
Enterprise hardware — configured, tested, deployed. Contact us for pricing and availability.
Request Quote