Ada Lovelace universal GPU for generative AI inference, graphics and Omniverse workloads.
The NVIDIA L40S is a universal Ada Lovelace data center GPU that combines AI compute with graphics and media acceleration. NVIDIA rates it at up to 5x the inference performance of the previous-generation A40.
Designed for 24/7 enterprise operation with passive cooling, secure boot and NEBS Level 3 readiness, it is widely deployed for generative AI, rendering and NVIDIA Omniverse.
Structural sparsity and TF32 support for faster training and inference.
Automatically switches between FP8 and FP16 to speed up AI.
Up to 2x the real-time ray-tracing performance of the previous generation.
Enough memory for LLM and generative AI workloads.
Three NVENC and three NVDEC engines with AV1 encode and decode.
Secure boot with root of trust, NEBS Level 3 ready, passive design.
| Architecture | NVIDIA Ada Lovelace |
|---|---|
| CUDA cores | 18,176 |
| Tensor Cores | 568 (4th generation) |
| RT Cores | 142 (3rd generation) |
| FP32 | 91.6 TFLOPS |
|---|---|
| TF32 Tensor Core | 183 | 366 TFLOPS (with sparsity) |
| FP16 Tensor Core | 362 | 733 TFLOPS (with sparsity) |
| FP8 Tensor Core | 733 | 1,466 TFLOPS (with sparsity) |
| INT8 / INT4 Tensor | 733 | 1,466 TOPS (with sparsity) |
| RT Core performance | 212 TFLOPS |
| GPU memory | 48 GB GDDR6 with ECC |
|---|---|
| Memory bandwidth | 864 GB/s |
| Interconnect | PCIe Gen4 x16 (64 GB/s bidirectional) |
| Max power | 350 W |
| Power connector | 16-pin |
| Form factor | 4.4" H × 10.5" L, dual-slot, passive |
| Display | 4× DisplayPort 1.4a |
| Media engines | 3× NVENC / 3× NVDEC (AV1 encode and decode) |
| vGPU | Supported |
| MIG / NVLink | Not supported |
Universal data center GPU for agentic AI, physical AI, scientific computing, rendering and video.
Energy-efficient multi-workload accelerator for AI inference, data science, video and visual computing.
Hopper-generation PCIe GPU with large HBM3e memory for LLM inference and HPC in air-cooled racks.
Tell us your workload and timeline for the NVIDIA L40S. We reply within one business day with a configuration, pricing and lead time.