Comment #1 on issue 2053 by [email protected]: Unexpected performance varaiations on physics simulation
http://code.google.com/p/v8/issues/detail?id=2053

I'm the author of the Chromium bug report 121196.

I do not yet have good profile data to make any good interpretation of the following results I get in Chrome ... but I am normally seeing the model itself running at 550-600 model-steps/s. When Chrome is running slowly the rate drops to 120-150 ... however during one test on Windows 7 (on the same machine) I got a measurement of 1724 model-steps/s.

It would be nice to get that performance regularly!

I'd like to know how to get detailed profiling information when running our physics simulation in Chrome ... but for now to get further I have installed v8 locally and extracted the part of our code running just the molecular dynamics simulation so I can run it with d8 and get better profiling data started this thread on the v8-users list (no responses yet):

  how to interpret the output when running with --trac-deopt on?
http://groups.google.com/group/v8-users/browse_thread/thread/14c339abe646c6c5#

I've created instructions on building v8 and running just the molecular dynamics engine with d8 with various profiling flags turned on:

  https://gist.github.com/2289269

I do not yet know if the data I am collecting using d8 also shows the same kind of huge variation in speeds we are seeing in Chrome -- but is it already clear that something is causing the main integration function to be deoptimized over and over:

  makeIntegrator.integrate()

See the profile.log here: https://raw.github.com/gist/2289269/profiling.log

Information below is for reference for other developers I point to this bug report:

For more info on interpreting the results see Florian Loitsch's series on profiling and optimizing code for V8 starting with this post:

  http://floitsch.blogspot.com/2012/03/optimizing-for-v8-introduction.html

I'm still trying to understand what is causing the deoptimization ...



--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev

Reply via email to