Re: [Pixman] [PATCH] SIMD: Try without any CFLAGS before forcing -mcpu=

2010-03-16 Thread Vladimir Vukicevic
Normally I'd agree, but in this case it might be worth it... for windows ce specifically, jeff muizelaar wrote a decompiler a while ago that could generate msvc assembly from compiled gcc code, which seemed to be a good way to go. MSVC doesn't do inline assembly on ARM anyway, so it would need

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 19:22 +0100, Alexander Larsson wrote: > On Tue, 2010-03-16 at 20:17 +0200, Siarhei Siamashka wrote: > > On Tuesday 16 March 2010, Siarhei Siamashka wrote: > > > On Tuesday 16 March 2010, Alexander Larsson wrote: > > > > On Tue, 2010-03-16 at 16:51 +0200, Siarhei Siamashka wrot

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 20:17 +0200, Siarhei Siamashka wrote: > On Tuesday 16 March 2010, Siarhei Siamashka wrote: > > On Tuesday 16 March 2010, Alexander Larsson wrote: > > > On Tue, 2010-03-16 at 16:51 +0200, Siarhei Siamashka wrote: > > > > Regarding the alex's branch and performance, I already me

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Siarhei Siamashka
On Tuesday 16 March 2010, Siarhei Siamashka wrote: > On Tuesday 16 March 2010, Alexander Larsson wrote: > > On Tue, 2010-03-16 at 16:51 +0200, Siarhei Siamashka wrote: > > > Regarding the alex's branch and performance, I already mentioned that > > > it was > > > much slower for over__0565 case

[Pixman] [PATCH 2/2] Add PIXMAN_DEFINE_THREAD_LOCAL() and PIXMAN_GET_THREAD_LOCAL() macros

2010-03-16 Thread Søren Sandmann
From: Søren Sandmann Pedersen These macros hide the various types of thread local support. On Linux and Unix, they expand to just __thread. On Microsoft Visual C++, they expand to __declspec(thread). On OS X and other systems that don't have __thread, they expand to a complicated concoction that

[Pixman] [PATCH 1/2] Add checks for various types of thread local storage.

2010-03-16 Thread Søren Sandmann
From: Søren Sandmann Pedersen OS X does not support __thread, so we have to check for it before using it. It does however support pthread_get/setspecific(), so if we don't have __thread, check if those are available. --- configure.ac | 80

[Pixman] [PATCH 0/2] Thread local support on OS X

2010-03-16 Thread Søren Sandmann
So apparently OS X does not support __thread which means current pixman does not compile there. See this: http://tinderbox.x.org/builds/2010-03-16-0022/logs/pixman/#build The following patches add some macros that expand to various types of thread local support. On Linux, Unix, and Windows

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Siarhei Siamashka
On Tuesday 16 March 2010, Alexander Larsson wrote: > On Tue, 2010-03-16 at 16:51 +0200, Siarhei Siamashka wrote: > > Regarding the alex's branch and performance, I already mentioned that > > it was > > much slower for over__0565 case in my benchmark when compared > > against my > > branch on AR

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 16:51 +0200, Siarhei Siamashka wrote: > > Regarding the alex's branch and performance, I already mentioned that > it was > much slower for over__0565 case in my benchmark when compared > against my > branch on ARM Cortex-A8 (the other cases of scaling are ok). I'm using >

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Siarhei Siamashka
On Tuesday 16 March 2010, Alexander Larsson wrote: > On Tue, 2010-03-16 at 13:26 +0100, Soeren Sandmann wrote: > > * This chunk: > > > > if (unit_x >= 0) > > while ... > > else > > while ... > > > > is repeated multiple times. > > > > Would it be possible

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 13:26 +0100, Soeren Sandmann wrote: > Hi, > > > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > > > > But before it gets committed, some problems with potential fixed > point > > > overflows when dealing with large images need to be addressed. I > have > > >

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 13:26 +0100, Soeren Sandmann wrote: > Hi, > > > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > > > > But before it gets committed, some problems with potential fixed > point > > > overflows when dealing with large images need to be addressed. I > have > > >

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Soeren Sandmann
Hi, > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > > But before it gets committed, some problems with potential fixed point > > overflows when dealing with large images need to be addressed. I have > > made a test program which can expose some of these problems: > > http://cgit

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 12:10 +0200, Siarhei Siamashka wrote: > > Also gcc shows the following warnings: > > pixman-fast-path.c: In > function ‘fast_composite_scaled_nearest_x888_565_SRC_none’: > pixman-fast-path.c:1560: warning: integer constant is too large for > ‘long’ type Hehe... I didn't b

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 12:10 +0200, Siarhei Siamashka wrote: > On Monday 15 March 2010, Alexander Larsson wrote: > > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > > But before it gets committed, some problems with potential fixed > point > > > overflows when dealing with large imag

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 12:10 +0200, Siarhei Siamashka wrote: > On Monday 15 March 2010, Alexander Larsson wrote: > > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > > But before it gets committed, some problems with potential fixed > point > > > overflows when dealing with large imag

Re: [Pixman] [PATCH] SIMD: Try without any CFLAGS before forcing -mcpu=

2010-03-16 Thread Siarhei Siamashka
On Sunday 14 March 2010, Loïc Minier wrote: > On Wed, Mar 10, 2010, Siarhei Siamashka wrote: > > I would prefer a bit more descriptive comment (with the details copied > > from that launchpad page). > > I see you pushed this now; thanks! Yeah, it's not obvious why one > needs to try with the too

Re: [Pixman] fast-scale branch performance improvements

2010-03-16 Thread Siarhei Siamashka
On Monday 15 March 2010, Alexander Larsson wrote: > On Mon, 2010-03-15 at 03:25 +0200, Siarhei Siamashka wrote: > > But before it gets committed, some problems with potential fixed point > > overflows when dealing with large images need to be addressed. I have > > made a test program which can expo

Re: [Pixman] bilinear fast path

2010-03-16 Thread Siarhei Siamashka
On Tuesday 16 March 2010, Alexander Larsson wrote: > On Tue, 2010-03-16 at 09:49 +0200, Siarhei Siamashka wrote: > > On Monday 15 March 2010, Alexander Larsson wrote: > > > On Mon, 2010-03-15 at 17:05 +0200, Siarhei Siamashka wrote: > > > > Really good performance improvements for bilinear scaling

Re: [Pixman] bilinear fast path

2010-03-16 Thread Alexander Larsson
On Tue, 2010-03-16 at 09:49 +0200, Siarhei Siamashka wrote: > On Monday 15 March 2010, Alexander Larsson wrote: > > On Mon, 2010-03-15 at 17:05 +0200, Siarhei Siamashka wrote: > > > Really good performance improvements for bilinear scaling are > going to > > > come > > > from SIMD optimizations. To

Re: [Pixman] bilinear fast path

2010-03-16 Thread Siarhei Siamashka
On Monday 15 March 2010, Alexander Larsson wrote: > On Mon, 2010-03-15 at 17:05 +0200, Siarhei Siamashka wrote: > > Really good performance improvements for bilinear scaling are going to > > come > > from SIMD optimizations. To make it happen, scaler core needs to be > > isolated > > into a small s