Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-31 Thread Oded Gabbay
On Thu, Dec 31, 2015 at 4:13 AM, Roland Scheidegger wrote: > > Am 30.12.2015 um 10:59 schrieb Oded Gabbay: > > On Wed, Dec 30, 2015 at 1:17 AM, Roland Scheidegger > > wrote: > >> The idea looks right to me. > >> Though frankly I don't like our current

Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-31 Thread Roland Scheidegger
Am 31.12.2015 um 19:41 schrieb Roland Scheidegger: > Am 31.12.2015 um 10:15 schrieb Oded Gabbay: >> On Thu, Dec 31, 2015 at 4:13 AM, Roland Scheidegger >> wrote: >>> >>> Am 30.12.2015 um 10:59 schrieb Oded Gabbay: On Wed, Dec 30, 2015 at 1:17 AM, Roland Scheidegger

Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-31 Thread Roland Scheidegger
Am 31.12.2015 um 10:15 schrieb Oded Gabbay: > On Thu, Dec 31, 2015 at 4:13 AM, Roland Scheidegger > wrote: >> >> Am 30.12.2015 um 10:59 schrieb Oded Gabbay: >>> On Wed, Dec 30, 2015 at 1:17 AM, Roland Scheidegger >>> wrote: The idea looks right to

Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-30 Thread Roland Scheidegger
Am 30.12.2015 um 10:59 schrieb Oded Gabbay: > On Wed, Dec 30, 2015 at 1:17 AM, Roland Scheidegger > wrote: >> The idea looks right to me. >> Though frankly I don't like our current setup code too much - in >> particular the mix between c, assembly, and jit code, with some >>

Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-30 Thread Oded Gabbay
On Wed, Dec 30, 2015 at 1:17 AM, Roland Scheidegger wrote: > The idea looks right to me. > Though frankly I don't like our current setup code too much - in > particular the mix between c, assembly, and jit code, with some > duplication (plus the lots of transpose everywhere).

[Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-29 Thread Oded Gabbay
This patch converts the SSE optimization done in do_triangle_ccw to VMX/VSX. I measured the results on POWER8 machine with 32 cores at 3.4GHz and 16GB of RAM. FPS/Score NameBefore AfterDelta glmark2

Re: [Mesa-dev] [PATCH 3/5] llvmpipe: Optimize do_triangle_ccw for POWER8

2015-12-29 Thread Roland Scheidegger
The idea looks right to me. Though frankly I don't like our current setup code too much - in particular the mix between c, assembly, and jit code, with some duplication (plus the lots of transpose everywhere). There's likely optimization potential to be found there. Roland Am 29.12.2015 um 17:12