mirror of
https://gitlab.freedesktop.org/mesa/mesa.git
synced 2026-05-05 13:58:04 +02:00
llvmpipe: fix nir dot products (fsum op)
When the dot product uses a source which can be optimized to a scalar, after a bunch of nir optimization steps the source to fsum will be a scalar with a x replicate swizzle. Hence nir_src_num_components is just 1 and the fsum was just a no-op which is not correct. Arguably this could be optimized a bit better, but just determine the number of addends by using nir_op_infos instead (the operand fetch was fixed already by39a938ecf4doing the same). Fixes:4eb0475b5a("gallivm/nir: add fsum support") Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com> Reviewed-by: Brian Paul <brianp@vmware.com> Reviewed-by: Dave Airlie <airlied@redhat.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/12066>
This commit is contained in:
parent
6691a48bcc
commit
cac5711d43
1 changed files with 1 additions and 1 deletions
|
|
@ -1021,7 +1021,7 @@ static void visit_alu(struct lp_build_nir_context *bld_base, const nir_alu_instr
|
|||
result[i] = cast_type(bld_base, src[i], nir_op_infos[instr->op].input_types[i], src_bit_size[i]);
|
||||
}
|
||||
} else if (instr->op == nir_op_fsum4 || instr->op == nir_op_fsum3 || instr->op == nir_op_fsum2) {
|
||||
for (unsigned c = 0; c < nir_src_num_components(instr->src[0].src); c++) {
|
||||
for (unsigned c = 0; c < nir_op_infos[instr->op].input_sizes[0]; c++) {
|
||||
LLVMValueRef temp_chan = LLVMBuildExtractValue(gallivm->builder,
|
||||
src[0], c, "");
|
||||
temp_chan = cast_type(bld_base, temp_chan, nir_op_infos[instr->op].input_types[0], src_bit_size[0]);
|
||||
|
|
|
|||
Loading…
Add table
Reference in a new issue