fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2025-12-27 04:00:10 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

intel/brw/xe2+: Adjust DPAS lowering to DP4A to accommodate larger GRF and SIMD16

Browse source 
Reviewed-by: Caio Oliveira <caio.oliveira@intel.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28834>
This commit is contained in:
Ian Romanick 2024-03-25 14:22:15 -07:00
parent 77ef241577
commit b6236dd8f3
1 changed files with 5 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

9
src/intel/compiler/brw_fs_lower_dpas.cpp
View file

@ -139,7 +139,7 @@ int8_using_dp4a(const fs_builder &bld, fs_inst *inst)
const fs_reg src1 = retype(inst->src[1], src1_type);
const fs_reg src2 = retype(inst->src[2], src2_type);
const unsigned dest_stride = REG_SIZE;
const unsigned dest_stride = reg_unit(bld.shader->devinfo) * REG_SIZE;
for (unsigned r = 0; r < inst->rcount; r++) {
if (!src0.is_null()) {
@ -152,8 +152,8 @@ int8_using_dp4a(const fs_builder &bld, fs_inst *inst)
for (unsigned s = 0; s < inst->sdepth; s++) {
bld.DP4A(dest,
dest,
byte_offset(src1, s * REG_SIZE),
component(byte_offset(src2, r * REG_SIZE), s))
byte_offset(src1, s * inst->exec_size * 4),
component(byte_offset(src2, r * inst->sdepth * 4), s))
->saturate = inst->saturate;
}
@ -279,7 +279,8 @@ brw_fs_lower_dpas(fs_visitor &v)
if (inst->opcode != BRW_OPCODE_DPAS)
continue;
const fs_builder bld = fs_builder(&v, block, inst).group(8, 0).exec_all();
const unsigned exec_size = v.devinfo->ver >= 20 ? 16 : 8;
const fs_builder bld = fs_builder(&v, block, inst).group(exec_size, 0).exec_all();
if (brw_type_is_float(inst->dst.type)) {
f16_using_mac(bld, inst);