NVIDIA V100S PCIe 32GB vs NVIDIA V100 SXM2 16GB

The V100S is the newer part: Volta, launched in 2020, against the V100's Volta from 2017. Newer architectures typically add lower-precision formats and better throughput per watt, so check the precision rows your workload actually uses.

Dense throughput for the V100S against the V100: FP64 8.2 vs 7.8 TFLOPS, FP32 16 vs 16 TFLOPS, FP16 130 vs 125 TFLOPS. Sparse figures, where the vendor publishes them, appear under each dense number in the performance table.

Memory is 32 GB against 16 GB, fed at 1.1 TB/s versus 900 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 250 W for the V100S and 300 W for the V100. At FP32 that works out to 0.07 against 0.05 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.