What is Roofline model?
A model expressing attainable performance as the smaller of peak compute and memory bandwidth × arithmetic intensity, used to tell compute-bound from memory-bound work.
Field: AI Hardware & Software Stack
Roofline model in the explanations
In the Roofline model, attainable performance is the smaller of the peak and bandwidth × arithmetic intensity.
On an accelerator with 100 TFLOPS peak and 2 TB/s memory ba…The roofline model plots performance against arithmetic intensity (operations per byte read from memory).
Which model plots whether a program's performance is limite…It shows how much bandwidth is actually usable compared with the theoretical figure and supplies the slope of the slanted line in the Roofline model.
What does the STREAM benchmark mainly measure?
Questions that mention Roofline model (4)
- On an accelerator with 100 TFLOPS peak and 2 TB/s memory bandwidth, a workload with arithmetic intensity 10 FLOP/byte runs. What performance does the Roofline model predict?
- When the Roofline model shows a workload in the memory-bandwidth-limited region, which direction improves performance?
- Which model plots whether a program's performance is limited by compute throughput or memory bandwidth, using arithmetic intensity (operations per byte) on the horizontal axis?
- What does the STREAM benchmark mainly measure?
Related terms
Arithmetic intensity Operator fusion Throughput STREAM benchmark AI compiler AI inference All-reduce Attention mechanism Backpropagation Batch size