skia2/dm
commit-bot@chromium.org e3ff558a4b SkRecord strawman
Record performance as measured by bench_record (out/Release/bench_record --skr) improves by at least 1.9x, at most 6.7x, arithmetic mean 2.6x, geometric mean 3.0x.  So, good.

Correctness as measured by DM (out/Debug/dm --skr) is ~ok.  One GM (shadertext2) fails because we're assuming all paint effects are immutable, but SkShaders are still mutable.

To do after this CL:
  - measure playback speed
  - catch up feature-wise to SkPicture
  - match today's playback speed


BUG=skia:
R=robertphillips@google.com, bsalomon@google.com, reed@google.com, mtklein@google.com

Author: mtklein@chromium.org

Review URL: https://codereview.chromium.org/206313003

git-svn-id: http://skia.googlecode.com/svn/trunk@14010 2bbb7eff-a529-9590-31e7-b0007b416f81
2014-04-01 16:24:06 +00:00
..
DM.cpp Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMBenchTask.cpp Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMBenchTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMCpuGMTask.cpp SkRecord strawman 2014-04-01 16:24:06 +00:00
DMCpuGMTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMExpectations.h Add support for reading a directory of images with --expectations (-r). 2013-12-10 14:53:16 +00:00
DMExpectationsTask.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMExpectationsTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMGpuGMTask.cpp Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMGpuGMTask.h Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMGpuSupport.h Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMPipeTask.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMPipeTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMRecordTask.cpp SkRecord strawman 2014-04-01 16:24:06 +00:00
DMRecordTask.h SkRecord strawman 2014-04-01 16:24:06 +00:00
DMReplayTask.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMReplayTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMReporter.cpp DM: tweak output. 2014-03-03 15:44:56 +00:00
DMReporter.h DM: tweak output. 2014-03-03 15:44:56 +00:00
DMSerializeTask.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMSerializeTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMTask.cpp Make DM play nice with the plan to split CPU and GPU work. 2014-03-31 19:59:44 +00:00
DMTask.h Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMTaskRunner.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMTaskRunner.h Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMTestTask.cpp Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMTestTask.h Let DM work without a GPU. 2014-03-26 21:26:15 +00:00
DMTileGridTask.cpp DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMTileGridTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
DMUtil.cpp DM: also run benches once. 2014-02-26 23:01:57 +00:00
DMUtil.h DM: also run benches once. 2014-02-26 23:01:57 +00:00
DMWriteTask.cpp DM: read image files without an extra copy 2014-03-05 15:37:11 +00:00
DMWriteTask.h DM: make GPU tasks multithreaded again. Big refactor. 2014-02-28 20:31:31 +00:00
README Reverting r12427 2013-12-02 13:50:38 +00:00

DM is like GM, but multithreaded.  It doesn't do everything GM does yet.

Current approximate list of missing features:
  --config pdf
  --mismatchPath
  --missingExpectationsPath
  --writePicturePath

  --deferred


DM's design is based around Tasks and a TaskRunner.

A Task represents an independent unit of work that might fail.  We make a task
for each GM/configuration pair we want to run.  Tasks can kick off new tasks
themselves.  For example, a CpuTask can kick off a ReplayTask to make sure
recording and playing back an SkPicture gives the same result as direct
rendering.

The TaskRunner runs all tasks on one of two threadpools, whose sizes are
configurable by --cpuThreads and --gpuThreads.  Ideally we'd run these on a
single threadpool but it can swamp the GPU if we shove too much work into it at
once.  --cpuThreads defaults to the number of cores on the machine.
--gpuThreads defaults to 1, but you may find 2 or 4 runs a little faster.

So the main flow of DM is:

    for each GM:
        for each configuration:
            kick off a new task
    < tasks run, maybe fail, and maybe kick off new tasks >
    wait for all tasks to finish
    report failures