mesa/src
Marek Olšák db2d31dab1 radeonsi: use v_mad_f32 for fma
v_fma_f32 runs at FP64 rate (= slow). Alien Isolation and F1 2015 seem
to use fma for all d3d multiply-add instructions, which is silly.

This tries to restore performance for those games.

The main difference between v_mad_f32 and v_fma_f32 is that v_mad doesn't
support denormals, which we don't enable anyway, because they are slow too.

Also, there is code size reduction:
  Totals from affected shaders:
  VGPRS: 109796 -> 109808 (0.01 %)
  Spilled SGPRs: 29995 -> 30022 (0.09 %)
  Spilled VGPRs: 12 -> 13 (8.33 %) <-- it's just one shader going from 12 to 13
  Code Size: 6667596 -> 6476356 (-2.87 %) bytes
  Max Waves: 26931 -> 26899 (-0.12 %)

I've not actually tested real performance.

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
2016-08-03 17:46:46 +02:00
..
compiler nir: fix validation message 2016-08-03 09:31:57 +10:00
egl egl/dri2: Add reference count for dri2_egl_display 2016-07-28 14:08:25 +01:00
gallium radeonsi: use v_mad_f32 for fma 2016-08-03 17:46:46 +02:00
gbm gbm: Removed unused function. 2016-08-01 09:11:14 -07:00
getopt
glx glx: Avoid aliasing violations. 2016-07-26 12:12:27 -07:00
gtest Remove wrongly repeated words in comments 2016-06-23 13:55:03 -07:00
hgl
intel anv: Perform rasterizer discard in the SOL stage instead of the clipper. 2016-07-30 12:06:37 -07:00
loader mesa: Use AC_HEADER_MAJOR to include correct header for major(). 2016-07-26 12:12:41 -07:00
mapi mapi: fix typo in macro name 2016-07-22 10:14:00 -07:00
mesa i965: use mt->offset in intel_miptree_map_movntdqa() 2016-08-03 08:28:52 -07:00
util list: fix list_replace() for empty lists 2016-07-06 10:17:30 -04:00
Makefile.am glsl: reuse main extension table to appropriately restrict extensions 2016-07-23 13:48:04 -04:00
SConscript scons: put the generated git_sha1.h file in top-level src/ directory 2016-06-17 10:33:00 -06:00