Skip to content
AI’s

hardware

FLOPS

Floating Point Operations Per Second, a measure of computing performance. In AI, total FLOPS measures the compute used during training (e.g., 10^25 FLOPS for GPT-4), while FLOPS/s measures hardware throughput.

In practice

NVIDIA's H100 GPU delivers about 989 teraFLOPS of FP16 performance, roughly 3x more than the previous A100.