FLOPS (Floating-Point Operations Per Second) measures the rate at which a processor performs floating-point arithmetic, and is the standard unit for comparing raw compute throughput in HPC. Peak FLOPS is the theoretical maximum based on hardware specs; achieved FLOPS on real code is almost always lower due to memory stalls, branches, and non-arithmetic instructions.

FLOPS figures are meaningless without specifying precision (FP64, FP32, FP16), since throughput can differ by an order of magnitude between precision levels on modern hardware.

Peak FLOPS computation for one x86 core.

Peak = cores × clock (Hz) × FLOPs/cycle
     = 1 × 3.0e9 Hz × (4 lanes × 2 FLOPs × 2 ports) = 48 GFLOPS