SERVCHIP
Interactive view
ARCHITECTURE
Ada Lovelace
SERIES
L40
Memory
48GB GDDR6 ECC
Bandwidth
864 GB/s
TDP
300W
In StockL40·Best for: Visual Computing & AI Inference
NVIDIA L40 GPU
Ada Lovelace
|Data Center GPUsOverview
The NVIDIA L40 GPU is a professional-grade data center accelerator built on the Ada Lovelace architecture, featuring 48GB of GDDR6 ECC memory with 864 GB/s bandwidth. Designed for visual computing, AI inference, and professional graphics rendering, the L40 includes third-generation RT Cores for real-time ray tracing and fourth-generation Tensor Cores for AI acceleration. It is the ideal GPU for virtual workstations, cloud gaming, media production, and mixed AI-graphics workloads in the data center.
Key Features
- 48GB GDDR6 ECC memory with 864 GB/s bandwidth
- Ada Lovelace architecture with 3rd-gen RT Cores
- 4th-generation Tensor Cores
- PCIe 4.0 x16 form factor
- Hardware AV1 encode and decode
Ideal Use Cases
ai inferencevirtualizationhealthcare
Full Specifications
Complete technical specifications for the L40.
Memory
- Memory
- 48GB GDDR6 ECC
- Memory Bandwidth
- 864 GB/s
Compute
- Tensor Cores
- 568 (4th-gen)
- CUDA Cores
- 18176
- FP8 TFLOPS
- 1467
- FP16 TFLOPS
- 733
- TF32 TFLOPS
- 91
Interconnect
- Interconnect
- PCIe 4.0 x16
Physical
- TDP
- 300W
- Form Factor
- PCIe Gen4 dual-slot
- Cooling
- Active Fan
Manufacturing
- Launch Date
- Q4 2022
- Manufacturing Process
- 5nm
