Okay, I have created: https://code.google.com/p/v8/issues/detail?id=3907

Builds are linked there. I'm providing the same links here also for convenience:

Growth: https://dl.dropboxusercontent.com/u/16662598/Ports/DOSBox-web/doom.html?engine=dosbox-growth.js

No growth: https://dl.dropboxusercontent.com/u/16662598/Ports/DOSBox-web/doom.html?engine=dosbox-nogrowth.js

On 2015-02-19 11:15 PM, Alon Zakai wrote:
This is very surprising and a serious issue that we should report to v8.
I would create --profiling builds with and without that option, and file
a bug with those links.

The reason it is surprising is that memory growth affects code in a very
small way. It uses a few more local variables, but that's it. I suppose
it might tip it over some internal limit in TurboFan or something like that.

- Alon


On Thu, Feb 19, 2015 at 2:52 PM, Boris Gjenero <boris.gjen...@gmail.com
<mailto:boris.gjen...@gmail.com>> wrote:

    When running Doom in Em-DOSBox, allowing memory growth results in a
    MASSIVE performance penalty in Chrome 42.0.2305.3 dev (64-bit).

    With -s ALLOW_MEMORY_GROWTH=0 Em-DOSBox runs Doom a lot faster in
    Chrome 42 compared to Chrome 40. Chrome 42 speed is similar to
    Firefox. Without that option, Chrome 42 is much slower than Chrome 40.

    I assume this has something to do with the new TurboFan optimizing
    compiler in the dev branch. Maybe it can't be used with memory
    growth enabled or that disables some optimizations. I realize this
    is the dev branch of Chrome and this issue may be fixed before it is
    released.

--
You received this message because you are subscribed to the Google Groups 
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to emscripten-discuss+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to