On 2013-02-04 9:59 PM, Dave Mandelin wrote:
On Friday, February 1, 2013 7:19:04 PM UTC-8, Brian Smith wrote:
Ehsan Akhgari wrote:
Given the above, I'd like to propose the following long-term
solutions:
1. Did we try escalating a support request to Microsoft regarding this issue? I
know it is kind of an odd thing, but it seems like if you are insistent enough
and/or pay enough money, Microsoft engineers from the affected product will get
assigned to help you with the problem [1]. Paid support is how Microsoft
compensates for being closed-source. I would not be surprised if somebody with
knowledge of the internals of the linker/compiler + experience with dealing
with other customers' PGO issues could give us some very helpful advise.
2. AFAICT, we did not seriously investigate the possibility of splitting things
out of libxul more. So far we've tried cutting things off the top of the
dependency tree. Maybe now we need to try cutting things off the bottom of the
dependency tree.
3. What is the performance difference between Visual Studio 2012 PGO builds and Visual
Studio 2010 builds? IMO, before we decide whether to disable PGO on Windows, we need to
get good benchmark results for Visual Studio **2012** PGO builds, to make sure we're not
throwing away wins that could come "just" solving this problem in a different
way + upgrading the compiler.
I was talking to Taras and Naveed about this today, and what also came up was:
4. Do the work to make 64-bit JS jit perf as good as 32-bit JS jit perf, and then switch
to x64 builds for Windows. There are of course many issues involved with such a switch,
but I believe that would fix the linker memory limit problem, and "make x64 jit perf
real good" seems much more useful and rewarding long-term than trying to patch up a
32-bit build process that MS may not even be that interested in supporting any more.
I agree, as long as we do have concrete plans for everything else which
would be involved in such a switch. The JS jit perf is really a small
piece of the puzzle, and a relatively easy one to solve since we control
all of the ends there.
Cheers,
Ehsan
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
