ALUSaturation

From HPC Wiki
Jump to navigation Jump to search
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Description

The pattern "ALU saturation" describes the performance limitation caused by fully utilizing a functional unit inside a CPU core.

Symptoms

  • Throughput at design limit(s)
  • Good (low) CPI
  • Integral ratio of cycles to specific instruction count(s).

Detection

Use a hardware-counter tool like:

  • LIKWID with performance groups FLOPS_SP, FLOPS_DP, DATA and CLOCK
  • Same information can be provided by perf or PAPI

Possible optimizations and/or fixes