Is there a way to run random binary / test on perf try bots? |
|||
Issue descriptionI'm trying to measure the cost of time measurements and locking (which can vary wildly depending on the OS and device) to help me design some scheduling primitives. I'd like to be able to leverage perf try bots and the variety of configuration it supports. Ideally I'd like to specify a binary, run it on all perf try bots and get stdout of this binary. Is there a (reasonably easy) way to achieve this (or something with the same effect)?
,
Jan 17 2018
No that isn't supported, and afaik we have no plans to support that use case.
,
Jan 17 2018
Re #3: Do you think that it's possible to hack together something with similar effect? Introducing a new benchmark / writing to stdout in the existing one? I don't need anything fancy, just something that I might be able to make work for me. Thanks!
,
Jan 17 2018
As another binary, off the top of my head I don't think so. You should be able to append to an existing telemetry benchmark (it must be a telemetry benchmark) and run that via the perf try bots.
,
Jan 17 2018
,
Aug 31
,
Aug 31
The bots are all on https://chrome-swarming.appspot.com/ in the chrome.tests.pinpoint pool, so you should be able to manually kick off binaries on them using the isolate and Swarming command-line tools. tools/swarming_client/isolate.py to package the binary and tools/swarming_client/swarming.py to kick off a task on the bots. You can also SSH into the bots directly, and run pretty much anything. Just make sure the bot is idle so as not to mess with the perf numbers! https://chrome-internal.googlesource.com/infra/infra_internal/+/master/doc/ssh.md |
|||
►
Sign in to add a comment |
|||
Comment 1 by perezju@chromium.org
, Jan 17 2018