Re: optimizations for 1.0

2008-04-09 Thread Steven Edwards
On Tue, Apr 8, 2008 at 10:54 PM, Steven Edwards <[EMAIL PROTECTED]> wrote: > I have a 64bit linux host and using -m32 with llvmgcc does not seem > to be very well documented. I expect to have problems with icc on > x86-64 also. Preconfigured LLVM packages for 2.2 work out of the box > on OS X

Re: optimizations for 1.0

2008-04-08 Thread Steven Edwards
On Tue, Apr 8, 2008 at 9:26 PM, Dmitry Timoshkov <[EMAIL PROTECTED]> wrote: > Why not start with Linux first? In that case you have much more helping > hands if you are in trouble, you avoid Mac specific code paths that may > distort the picture, and Linux has oprofile - unbeatable choice. I h

Re: optimizations for 1.0

2008-04-08 Thread Dmitry Timoshkov
"Steven Edwards" <[EMAIL PROTECTED]> wrote: > Good point. I just tested LLVM 2.2 on OS X and it produces a totally > broken Wine so thats out. I see ICC has an eval version for OS X so I > am going to give it a shot next. Why not start with Linux first? In that case you have much more helping han

Re: optimizations for 1.0

2008-04-08 Thread Steven Edwards
On Tue, Apr 8, 2008 at 4:19 PM, Scott Ritchie <[EMAIL PROTECTED]> wrote: > I'd settle for someone telling me which GCC version to use. If you are > doing optimization testing, please make sure to test that copy > protection hasn't broken as well :) Good point. I just tested LLVM 2.2 on OS X an

Re: optimizations for 1.0

2008-04-08 Thread Scott Ritchie
Steven Edwards wrote: > Hi, > Given we are coming up close to the freeze, I think this may be a good > time to look in to optimizations. I propose we spend some time trying > different compilers and profiling Wine under different usage patterns > to see what type of speedup we can get. I'm going to

optimizations for 1.0

2008-04-08 Thread Steven Edwards
Hi, Given we are coming up close to the freeze, I think this may be a good time to look in to optimizations. I propose we spend some time trying different compilers and profiling Wine under different usage patterns to see what type of speedup we can get. I'm going to see if its possible to compile