Re: [Mesa-dev] softpipe and tgsi_exec.c FAST_MATH

2012-01-03 Thread Roland Scheidegger
Am 31.12.2011 17:46, schrieb Dave Airlie: > Hi, > > So in the interest of correctness > perf wrt softpipe, flicking > tgsi_exec.c FAST_MATH to 0 fixes 15 piglit tests with 0 regressions, > > I could probably be persuaded to make a SOFTPIPE_FAST_MATH run time > set to 0 by default, anyone care one

Re: [Mesa-dev] softpipe and tgsi_exec.c FAST_MATH

2012-01-02 Thread Brian Paul
On 12/31/2011 09:46 AM, Dave Airlie wrote: Hi, So in the interest of correctness> perf wrt softpipe, flicking tgsi_exec.c FAST_MATH to 0 fixes 15 piglit tests with 0 regressions, I could probably be persuaded to make a SOFTPIPE_FAST_MATH run time set to 0 by default, anyone care one way or ano

[Mesa-dev] softpipe and tgsi_exec.c FAST_MATH

2011-12-31 Thread Dave Airlie
Hi, So in the interest of correctness > perf wrt softpipe, flicking tgsi_exec.c FAST_MATH to 0 fixes 15 piglit tests with 0 regressions, I could probably be persuaded to make a SOFTPIPE_FAST_MATH run time set to 0 by default, anyone care one way or another? Dave.