mesa/src/amd
Georg Lehmann df16f47036 aco: optimize dd[xy]_fine if it's only used by abs
If we can ignore the sign of the derivative, we can swap the lanes
instead of broadcasting per direction. abs(a - b) = abs(b - a).
Shamelessly copied from bifrost.

Foz-DB Navi31:
Totals from 5 (0.01% of 79206) affected shaders:
Instrs: 6191 -> 6184 (-0.11%)
CodeSize: 31960 -> 31920 (-0.13%)
Latency: 111961 -> 111926 (-0.03%)
InvThroughput: 10390 -> 10372 (-0.17%)
VALU: 3286 -> 3279 (-0.21%)

Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30557>
2024-08-08 17:39:55 +00:00
..
addrlib amd: add GFX v11.5.2 support 2024-07-02 12:05:23 +00:00
ci Uprev Piglit to 0453436872b6e4d502c2e87817addb95e0d77e3b 2024-08-08 02:25:45 +00:00
common ac/nir: Shorten the name of ac_nir_calc_io_offset_mapped. 2024-08-08 16:55:02 +00:00
compiler aco: optimize dd[xy]_fine if it's only used by abs 2024-08-08 17:39:55 +00:00
drm-shim build: pass licensing information in SPDX form 2024-06-29 12:42:49 -07:00
llvm amd: switch to derivative intrinsics 2024-08-08 15:26:07 +00:00
registers amd: add gfx12 register definitions 2024-05-11 22:14:05 -04:00
vpelib amd/vpelib: Refactor frontend and backend config callback 2024-08-07 16:46:25 +00:00
vulkan radv: Stop assigning linked driver locations. 2024-08-08 16:55:02 +00:00
meson.build build: pass licensing information in SPDX form 2024-06-29 12:42:49 -07:00