fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-24 21:50:12 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

nir/lower_alu: drop unnecessary iand on uadd_carry result

Browse source 
uadd_carry returns 1 or 0, so ANDing with 1 is unnecessary. Probably
this was implemented thinking that it was returning a boolean value.

shader-db results for V3D:

total instructions in shared programs: 12463571 -> 12462964 (<.01%)
instructions in affected programs: 28994 -> 28387 (-2.09%)
helped: 110
HURT: 1

total uniforms in shared programs: 3704591 -> 3704588 (<.01%)
uniforms in affected programs: 247 -> 244 (-1.21%)
helped: 3
HURT: 0

total max-temps in shared programs: 2148138 -> 2148117 (<.01%)
max-temps in affected programs: 729 -> 708 (-2.88%)
helped: 23
HURT: 2

total sfu-stalls in shared programs: 21230 -> 21232 (<.01%)
sfu-stalls in affected programs: 0 -> 2
helped: 0
HURT: 2

Reviewed-by: Alyssa Rosenzweig <alyssa@collabora.com>
Reviewed-by: Jason Ekstrand <jason.ekstrand@collabora.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/17903>
This commit is contained in:
Iago Toral Quiroga 2022-08-04 10:29:16 +02:00 • committed by Marge Bot
parent 8ecea47f06
commit 9d6770d20a
1 changed files with 4 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/compiler/nir/nir_lower_alu.c
View file

@ -147,7 +147,6 @@ lower_alu_instr(nir_alu_instr *instr, nir_builder *b)
nir_ssa_def *dest_shifted = nir_ishr(b, dest_32, nir_imm_int(b, src0->bit_size));
lowered = nir_build_alu(b, downscast_op, dest_shifted, NULL, NULL, NULL);
} else {
nir_ssa_def *c1 = nir_imm_intN_t(b, 1, src0->bit_size);
nir_ssa_def *cshift = nir_imm_int(b, src0->bit_size / 2);
nir_ssa_def *cmask = nir_imm_intN_t(b, (1ull << (src0->bit_size / 2)) - 1, src0->bit_size);
nir_ssa_def *different_signs = NULL;
@ -180,12 +179,12 @@ lower_alu_instr(nir_alu_instr *instr, nir_builder *b)
nir_ssa_def *tmp;
tmp = nir_ishl(b, m1, cshift);
hi = nir_iadd(b, hi, nir_iand(b, nir_uadd_carry(b, lo, tmp), c1));
hi = nir_iadd(b, hi, nir_uadd_carry(b, lo, tmp));
lo = nir_iadd(b, lo, tmp);
hi = nir_iadd(b, hi, nir_ushr(b, m1, cshift));
tmp = nir_ishl(b, m2, cshift);
hi = nir_iadd(b, hi, nir_iand(b, nir_uadd_carry(b, lo, tmp), c1));
hi = nir_iadd(b, hi, nir_uadd_carry(b, lo, tmp));
lo = nir_iadd(b, lo, tmp);
hi = nir_iadd(b, hi, nir_ushr(b, m2, cshift));
@ -195,14 +194,11 @@ lower_alu_instr(nir_alu_instr *instr, nir_builder *b)
* high 32-bits. Consider -3 * 2. The high 32-bits is 0, but the
* desired result is -1, not -0! Recall -x == ~x + 1.
*/
nir_ssa_def *c1 = nir_imm_intN_t(b, 1, src0->bit_size);
hi = nir_bcsel(b, different_signs,
nir_iadd(b,
nir_inot(b, hi),
nir_iand(b,
nir_uadd_carry(b,
nir_inot(b, lo),
c1),
nir_imm_intN_t(b, 1, src0->bit_size))),
nir_uadd_carry(b, nir_inot(b, lo), c1)),
hi);
}