Buying in volume? Request a bulk quote
We source this item through our supplier network and do not hold inventory. Price and availability are confirmed with you before any order is accepted.
Questions about this part? Talk to us.
Phone +1 872 293 4196 · Email supply@allchipsupply.com
Location: USA · Shipping: Worldwide Shipping
The NVIDIA L40S is a universal data center GPU built on the Ada Lovelace architecture, designed for generative AI inference, fine-tuning, graphics and video workloads. It has 18,176 CUDA cores and 48 GB of GDDR6 memory with ECC, with 864 GB/s of bandwidth.
NVIDIA lists 91.6 TFLOPS of FP32 performance and 1,466 TFLOPS of FP8 with sparsity. The card uses PCIe 4.0 x16, draws up to 350 W and is a dual-slot, passively cooled design measuring 4.4 in by 10.5 in, with four DisplayPort outputs. Because it fits standard PCIe servers, it is often chosen where SXM-based HGX systems are not practical. Confirm your server supports it before ordering.
| Architecture | NVIDIA Ada Lovelace |
|---|---|
| CUDA cores | 18,176 |
| GPU memory | 48 GB GDDR6 with ECC |
| Memory bandwidth | 864 GB/s |
| Interface | PCIe 4.0 x16 |
| FP32 | 91.6 TFLOPS |
| FP8 (sparsity) | 1,466 TFLOPS |
| Max power | 350 W |
| Outputs | 4x DisplayPort |
| Form factor | Dual-slot, 4.4" H x 10.5" L |
Specifications are summarized from manufacturer and listing information; confirm against the manufacturer datasheet before purchase.
Manufacturer datasheet / product page (opens the manufacturer's website)
NVIDIA Ada Lovelace · 24 GB GDDR6 · 300 GB/s
NVIDIA Ampere · 80 GB HBM2e, ECC · ~1.9 TB/s
NVIDIA Hopper · 80 GB HBM2e · 2.0 TB/s
NVIDIA Hopper · 141 GB HBM3e · 4.8 TB/s
