John McCalpin

APRIL 2, 2020

This data is from the 2007 presentation. It is not surprising that there is a lot of scatter, but the factor of four range in Peak MFLOPS at fixed SPECfp_rate2000/core and the factor of four range in SPECfp_rate2000/core at fixed Peak MFLOPS was higher than I expected… (Also from the 2007 presentation.)

Benchmarking 59

Benchmarking Performance Latency Architecture 59

Brendan Gregg

JANUARY 17, 2018

Short durations can be useful for understanding how well a WSS will fit into the CPU caches (L1/L2/L3, TLB L1/L2, etc). In this case, 28 Mbytes is a little larger than the LLC for this CPU, so may not cache so well (in a single LLC, anyway, and based on this single measurement). This is the working set size metric. 16.570 403.66

Cache 71

Cache Latency C++ Programming 71

Alex Russell

APRIL 29, 2021

Another window into this question is provided by the Web Confluence Metrics project. Since 2007, support for these features has barely improved. For heavily latency-sensitive use-cases like WebXR, this is a critical component in delivering a good experience. This adds to the expense of developing for iOS. Converging Views.

Media 145

Media Games Education Engineering 145

Brendan Gregg

JANUARY 16, 2018

Short durations can be useful for understanding how well a WSS will fit into the CPU caches (L1/L2/L3, TLB L1/L2, etc). In this case, 28 Mbytes is a little larger than the LLC for this CPU, so may not cache so well (in a single LLC, anyway, and based on this single measurement). This is the working set size metric. 16.570 403.66

Cache 40

Cache Latency C++ Programming 40