AlexIndustrial/mesa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

nak/sm20: Improve folding of ffma and dfma

Browse Source 
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34619>
This commit is contained in:
Faith Ekstrand
2025-04-19 11:01:19 -05:00
parent 9855467ed0
commit 142fb563c4
1 changed files with 10 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/nouveau/compiler/nak/sm20.rs
+10 -2
View File
@@ -598,7 +598,11 @@ impl SM20Op for OpFFma {
{
b.copy_alu_src(src1, GPR, SrcType::F32);
}
b.copy_alu_src_if_not_reg(src2, GPR, SrcType::F32);
if src_is_reg(src1, GPR) {
b.copy_alu_src_if_imm(src2, GPR, SrcType::ALU);
} else {
b.copy_alu_src_if_not_reg(src2, GPR, SrcType::I32);
}
}
fn encode(&self, e: &mut SM20Encoder<'_>) {
@@ -925,7 +929,11 @@ impl SM20Op for OpDFma {
swap_srcs_if_not_reg(src0, src1, GPR);
b.copy_alu_src_if_not_reg(src0, GPR, SrcType::F64);
b.copy_alu_src_if_f20_overflow(src1, GPR, SrcType::F64);
b.copy_alu_src_if_not_reg(src2, GPR, SrcType::F64);
if src_is_reg(src1, GPR) {
b.copy_alu_src_if_imm(src2, GPR, SrcType::ALU);
} else {
b.copy_alu_src_if_not_reg(src2, GPR, SrcType::I32);
}
}
fn encode(&self, e: &mut SM20Encoder<'_>) {