projects / platform / upstream / mesa.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 78bf539) | patch
intel/fs: Use new F16TO32 helpers for unpack_half_split_* opcodes
authorKenneth Graunke <kenneth@whitecape.org>
Wed, 8 Mar 2023 04:47:02 +0000 (20:47 -0800)
committerMarge Bot <emma+marge@anholt.net>
Thu, 9 Mar 2023 23:26:17 +0000 (23:26 +0000)
commit309ec3725a54f7ac037a102b1a90057fb7803d88
tree02abcfee3b5985033b50f102f0615e90d60a806ftree | snapshot
parent78bf53904e85b4578fbd5371efdf57c95ea523e1commit | diff
intel/fs: Use new F16TO32 helpers for unpack_half_split_* opcodes

This gets us a MOV at the IR level on Gfx8+ which should be more
optimizable than F16TO32.  It also removes confusion about which
pipe which the instruction will run on.

Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Sagar Ghuge <sagar.ghuge@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21783>
src/intel/compiler/brw_fs_nir.cpp diff | blob | history
Domain: Graphics System / GL;