liveBook · Manning

Floating-point operations per second

As explained in GPU Programming with Triton

Floating-point operations per second measures the rate at which a processor or system performs floating-point arithmetic.

Floating-point operations per second (FLOPS or FLOP/s) measures how many floating-point operations a processor or system performs each second. A FLOP counts one operation, such as an addition or multiplication; FLOPs can mean the total operation count. section 3.1.2

Calculating a rate

A loop that performs 1,000 floating-point additions in 0.001 seconds runs at 1,000,000 FLOPS. A GPU rated at 10 TFLOPS has a theoretical peak of about 10 × 10¹² operations per second when fully utilized. section 3.1.2

How GPU Programming with Triton covers it

How deep it goes
Defines individual operations, total operation counts, and rates, then works through a simple timing calculation.
Examples
Uses a 1,000-add loop and a GPU rated at 10 TFLOPS to illustrate measured and peak throughput.
What it assumes
Requires no stated GPU background; defines individual floating-point operations and total operation counts.
What you can do after
Evaluate throughput by dividing a floating-point operation count by elapsed time.
How it connects
Distinguishes floating-point operation counts from throughput and relates throughput to GPU performance.

Also explained in

Compare how 3 books explain floating-point operations per second →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of GPU Programming with Triton. The section links point to the parts of the book it draws on.