AuroraMiddleware/skia2
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
6cd70ffa98
skia2/tools/nanobench_flags.json

96 lines
1.6 KiB
JSON
Raw Normal View History

Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
{
Add presubmit to run `{dm,nanobench}_flags.py test` when changed. BUG=skia: Review URL: https://codereview.chromium.org/1129443005
2015-05-05 17:28:44 +00:00
"Perf-Android-GCC-NexusPlayer-GPU-PowerVR-x86-Release": [
"--scales",
"1.0",
"1.1",
"--config",
"565",
"8888",
"gpu",
"nonrendering",
"angle",
"hwui",
"--match",
"~blurroundrect",
"~patch_grid",
"~desk_carsvg",
"~desk_unicodetable"
],
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
"Perf-Android-Nexus7-Tegra3-Arm7-Release": [
"--scales",
"1.0",
"1.1",
Add msaa runs to nanobench TBR=mtklein@google.com Review URL: https://codereview.chromium.org/964053002
2015-03-01 00:56:31 +00:00
"--config",
"565",
"8888",
"gpu",
"nonrendering",
"angle",
"hwui",
"msaa4",
"nvprmsaa4",
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
"--match",
"skp"
],
Update [dm|nanobench]_flags for CPU/GPU split TBR=mtklein NOTREECHECKS=true BUG=skia:2073 Review URL: https://codereview.chromium.org/1036223002
2015-03-27 12:42:18 +00:00
"Test-Ubuntu-GCC-ShuttleA-GPU-GTX550Ti-x86_64-Release-Valgrind": [
Run Valgrind bots with --nogpu and --nocpu as appropriate BUG=skia:3506 NOPRESUBMIT=true Review URL: https://codereview.chromium.org/986583004
2015-03-09 13:59:16 +00:00
"--scales",
"1.0",
"1.1",
"--config",
"565",
"8888",
"gpu",
"nonrendering",
"angle",
"hwui",
"msaa16",
"nvprmsaa16",
"--loops",
"1",
"--samples",
Update [dm|nanobench]_flags for CPU/GPU split TBR=mtklein NOTREECHECKS=true BUG=skia:2073 Review URL: https://codereview.chromium.org/1036223002
2015-03-27 12:42:18 +00:00
"1"
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
],
Update [dm|nanobench]_flags for CPU/GPU split TBR=mtklein NOTREECHECKS=true BUG=skia:2073 Review URL: https://codereview.chromium.org/1036223002
2015-03-27 12:42:18 +00:00
"Test-Win7-MSVC-ShuttleA-GPU-HD2000-x86-Debug-ANGLE": [
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
"--scales",
"1.0",
"1.1",
Add msaa runs to nanobench TBR=mtklein@google.com Review URL: https://codereview.chromium.org/964053002
2015-03-01 00:56:31 +00:00
"--config",
"565",
"8888",
"gpu",
"nonrendering",
"angle",
"hwui",
"msaa16",
"nvprmsaa16",
nanobench: split CPU and GPU .SKP tile sizes. Let's make CPU-bound .SKP benching mimic Chrome's tiles. Unfortunately, the CPU code also performs a lot better with those big wide tiles... BUG=skia: Review URL: https://codereview.chromium.org/1189863002
2015-06-16 17:41:27 +00:00
"--GPUbenchTileW",
Only use 256x256 tiles on hd2000 nanobench bots Initial experiments did show that the 256 tile size fixed the hd2000 win7 nanobot failures. However it did not have any effect on other bots, so this change is to move back to the larger tile size on all bots expect for the hd2000. BUG=skia: Review URL: https://codereview.chromium.org/1022083002
2015-03-20 14:03:52 +00:00
"256",
nanobench: split CPU and GPU .SKP tile sizes. Let's make CPU-bound .SKP benching mimic Chrome's tiles. Unfortunately, the CPU code also performs a lot better with those big wide tiles... BUG=skia: Review URL: https://codereview.chromium.org/1189863002
2015-06-16 17:41:27 +00:00
"--GPUbenchTileH",
Only use 256x256 tiles on hd2000 nanobench bots Initial experiments did show that the 256 tile size fixed the hd2000 win7 nanobot failures. However it did not have any effect on other bots, so this change is to move back to the larger tile size on all bots expect for the hd2000. BUG=skia: Review URL: https://codereview.chromium.org/1022083002
2015-03-20 14:03:52 +00:00
"256",
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
"--match",
"~gradient",
"~etc1bitmap"
nanobench: split CPU and GPU .SKP tile sizes. Let's make CPU-bound .SKP benching mimic Chrome's tiles. Unfortunately, the CPU code also performs a lot better with those big wide tiles... BUG=skia: Review URL: https://codereview.chromium.org/1189863002
2015-06-16 17:41:27 +00:00
],
"Test-iOS-Clang-iPad4-GPU-SGX554-Arm7-Debug": [
"--scales",
"1.0",
"1.1",
"--skps",
"ignore_skps",
"--config",
"565",
"8888",
"gpu",
"nonrendering",
"angle",
"hwui",
"msaa16",
"nvprmsaa16",
"--match",
"~blurroundrect",
"~patch_grid",
"~desk_carsvg",
"~keymobi",
"~path_hairline"
Add tools/nanobench_flags.py. This should look suspiciously similar to tools/dm_flags.py. In fact, I tweaked tools/dm_flags.py a bit to make it even more suspiciously similar. I'll leave actually deduping this to future me. I noticed we have an opportunity to make our Valgrind run of nanobench faster, by not only making it not auto-calibrate (--loops 1) but also take only one measurement (--samples 1). Should be 5-10x faster than the default. BUG=skia: Review URL: https://codereview.chromium.org/957503002
2015-02-24 19:45:11 +00:00
]
}
Reference in New Issue Copy Permalink