Status: New
Owner: ----

New issue 3907 by [email protected]: Allowing asm.js memory growth causes big slowdown
https://code.google.com/p/v8/issues/detail?id=3907

Version: Google Chrome 42.0.2305.3 dev (64-bit)
OS: Ubuntu 14.10
Architecture: x64

What steps will reproduce the problem?
1. Compile Em-DOSBox https://github.com/dreamlayers/em-dosbox using Emscripten with memory growth enabled. (Link time switch "-s ALLOW_MEMORY_GROWTH=1" is now the default in Emscripten incoming.) 2. Run DOS program using max cycles, such as Doom. A compiled version running Doom can be found at: https://dl.dropboxusercontent.com/u/16662598/Ports/DOSBox-web/doom.html?engine=dosbox-growth.js
3. Observe performance

What is the expected output?
When memory growth is disabled via "-s ALLOW_MEMORY_GROWTH=0" at link time, Em-DOSBox runs quickly. Chrome 42 is faster than Chrome 40 and close to Firefox asm.js performance. A version compiled without memory growth can be run at: https://dl.dropboxusercontent.com/u/16662598/Ports/DOSBox-web/doom.html?engine=dosbox-nogrowth.js . Considering that allowing memory growth only makes minor changes and that it doesn't cause a big slowdown in Chrome 40 and Firefox 35, there shouldn't be a big difference in Chrome 42.

What do you see instead?
With memory growth enabled, Em-DOSBox runs much slower, and even significantly slower than in Chrome 40.

Both of those JS files were built with profiling enabled. I didn't notice any glaring issues in profiling output. It seems as if the whole thing is running slower.


--
You received this message because this project is configured to send all issue notifications to this address.
You may adjust your notification preferences at:
https://code.google.com/hosting/settings

--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
--- You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to