mirrors/mesa
0
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2024-11-23 02:04:41 +08:00
Code Issues Packages Projects Releases Wiki Activity

ac/nir/ngg: Don't emit dead code with dot_op.

Browse Source 
dot_op would be dead code when v_dot instructions are unavailable.
It was originally added there because ACO didn't have an ILP
scheduler yet, but now it does so let's trust it to do its job.

Signed-off-by: Timur Kristóf <timur.kristof@gmail.com>
Acked-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31973>
This commit is contained in:
Timur Kristóf 2024-11-16 16:44:41 +01:00
parent b9d2b5e100
commit 340ec61984
1 changed files with 2 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
src/amd/common/ac_nir_lower_ngg.c
View File

@ -268,27 +268,25 @@ summarize_repack(nir_builder *b, nir_def *packed_counts, unsigned num_lds_dwords
bool use_dot = b->shader->options->has_udot_4x8;
if (num_lds_dwords == 1) {
nir_def *dot_op = !use_dot ? NULL : nir_ushr(b, nir_ushr(b, nir_imm_int(b, 0x01010101), shift), shift);
/* Broadcast the packed data we read from LDS (to the first 16 lanes, but we only care up to num_waves). */
nir_def *packed = nir_lane_permute_16_amd(b, packed_counts, nir_imm_int(b, 0), nir_imm_int(b, 0));
/* Horizontally add the packed bytes. */
if (use_dot) {
nir_def *dot_op = nir_ushr(b, nir_ushr(b, nir_imm_int(b, 0x01010101), shift), shift);
return nir_udot_4x8_uadd(b, packed, dot_op, nir_imm_int(b, 0));
} else {
nir_def *sad_op = nir_ishl(b, nir_ishl(b, packed, shift), shift);
return nir_msad_4x8(b, sad_op, nir_imm_int(b, 0), nir_imm_int(b, 0));
}
} else if (num_lds_dwords == 2) {
nir_def *dot_op = !use_dot ? NULL : nir_ushr(b, nir_ushr(b, nir_imm_int64(b, 0x0101010101010101), shift), shift);
/* Broadcast the packed data we read from LDS (to the first 16 lanes, but we only care up to num_waves). */
nir_def *packed_dw0 = nir_lane_permute_16_amd(b, nir_unpack_64_2x32_split_x(b, packed_counts), nir_imm_int(b, 0), nir_imm_int(b, 0));
nir_def *packed_dw1 = nir_lane_permute_16_amd(b, nir_unpack_64_2x32_split_y(b, packed_counts), nir_imm_int(b, 0), nir_imm_int(b, 0));
/* Horizontally add the packed bytes. */
if (use_dot) {
nir_def *dot_op = nir_ushr(b, nir_ushr(b, nir_imm_int64(b, 0x0101010101010101), shift), shift);
nir_def *sum = nir_udot_4x8_uadd(b, packed_dw0, nir_unpack_64_2x32_split_x(b, dot_op), nir_imm_int(b, 0));
return nir_udot_4x8_uadd(b, packed_dw1, nir_unpack_64_2x32_split_y(b, dot_op), sum);
} else {