Computer performance microscopy with S him

Abstract

Developers and architects spend a lot of time trying to understand and eliminate performance problems. Unfortunately, the root causes of many problems occur at a fine granularity that existing continuous profiling and direct measurement approaches cannot observe. This paper presents the design and implementation of S him , a continuous profiler that samples at resolutions as fine as 15 cycles; three to five orders of magnitude finer than current continuous profilers. S him 's fine-grain measurements reveal new behaviors, such as variations in instructions per cycle (IPC) within the execution of a single function. A S him observer thread executes and samples autonomously on unutilized hardware. To sample, it reads hardware performance counters and memory locations that store software state. S him improves its accuracy by automatically detecting and discarding samples affected by measurement skew. We measure S him 's observer effects and show how to analyze them. When on a separate core, S him can continuously observe one software signal with a 2% overhead at a ~1200 cycle resolution. At an overhead of 61%, S him samples one software signal on the same core with SMT at a ~15 cycle resolution. Modest hardware changes could significantly reduce overheads and add greater analytical capability to S him . We vary prefetching and DVFS policies in case studies that show the diagnostic power of fine-grain IPC and memory bandwidth results. By repurposing existing hardware, we deliver a practical tool for fine-grain performance microscopy for developers and architects.