fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-24 11:00:11 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

aco: use and swizzle mask in dpp quad perm

Browse source 
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/21412>
This commit is contained in:
Georg Lehmann 2023-02-23 11:35:13 +01:00 • committed by Marge Bot
parent 8fabde3be4
commit 1c5c2f77c3
1 changed files with 2 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
src/amd/compiler/aco_instruction_selection.cpp
View file

@ -253,10 +253,10 @@ emit_masked_swizzle(isel_context* ctx, Builder& bld, Temp src, unsigned mask)
* because DPP16 supports modifiers and v_permlane
* can't be folded into valu instructions.
*/
if (and_mask == 0x1f && or_mask < 4 && xor_mask < 4) {
if ((and_mask & 0x1c) == 0x1c && or_mask < 4 && xor_mask < 4) {
unsigned res[4] = {0, 1, 2, 3};
for (unsigned i = 0; i < 4; i++)
res[i] = ((res[i] | or_mask) ^ xor_mask) & 0x3;
res[i] = (((res[i] & and_mask) | or_mask) ^ xor_mask) & 0x3;
dpp_ctrl = dpp_quad_perm(res[0], res[1], res[2], res[3]);
} else if (and_mask == 0x1f && !or_mask && xor_mask == 8) {
dpp_ctrl = dpp_row_rr(8);