NVIDIA L40 48GB vs NVIDIA L40S 48GB

Both GPUs launched in 2023: the L40 on Ada Lovelace and the L40S on Ada Lovelace.

Dense throughput for the L40 against the L40S: FP32 91 vs 92 TFLOPS, FP16 181 vs 362 TFLOPS, FP8 362 vs 733 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 48 GB against 48 GB, fed at 864 GB/s versus 864 GB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.

Power budgets are 300 W for the L40 and 350 W for the L40S. At FP32 that works out to 0.30 against 0.26 TFLOPS per watt.

For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.