More micro-opts for AudioDelayDspKernel
Benchmarks (https://spotify.github.io/web-audio-bench/) using default number of runs and duration: Without this CL: TEST μs MIN Q1 MEDIAN Q3 MAX Baseline 708 708 792 804 817 1595 Delay-default 372 372 472 499 536 4723 Delay-0.1 409 409 498 527 559 1639 DelayAutomation-a-rate 640 640 814 842 874 1018 DelayAutomation-k-rate 415 415 500 524 559 1676 With this CL TEST μs MIN Q1 MEDIAN Q3 MAX Baseline 767 767 793 804 821 1295 Delay-default 262 262 332 354 388 3840 Delay-0.1 284 284 355 382 415 1773 DelayAutomation-a-rate 640 640 741 761 797 1012 DelayAutomation-k-rate 264 264 367 388 421 1812 So we see that delay tests are improved by about 30%, except for the a-rate case. The changes for a-rate didn't have any measurable impact. Bug: 1054416 Change-Id: I4d3428fad6a13fd0632e09f7c03dca051ae09b87 Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1867377 Commit-Queue: Raymond Toy <rtoy@chromium.org> Reviewed-by:Hongchan Choi <hongchan@chromium.org> Cr-Commit-Position: refs/heads/master@{#743362}
Showing
Please register or sign in to comment