ShuriZma
/
xenia
mirror of https://github.com/xenia-project/xenia.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

[a64] Remove redundant `OPCODE_DOT_PRODUCT_{3,4}` lane-isolation 

The last `FADDP` writes into an `S` register, which automatically masks all the other lanes to zero.
This commit is contained in:
Wunkolo 2024-06-16 15:00:03 -07:00
parent 9c8b0678a5
commit 9c572c3937
1 changed files with 0 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
src/xenia/cpu/backend/a64/a64_sequences.cc
View File

@ -2111,11 +2111,6 @@ struct DOT_PRODUCT_3_V128
e.MOV(dest.toQ().Selem()[3], WZR);
e.FADDP(dest.toQ().S4(), dest.toQ().S4(), dest.toQ().S4());
e.FADDP(dest.toS(), dest.toD().S2());
// Isolate lower lane
e.MOVI(Q0.D2(), RepImm(0b00'00'00'00));
e.INS(Q0.Selem()[0], dest.toQ().Selem()[0]);
e.MOV(dest.toQ().B16(), Q0.B16());
});
}
};
@ -2134,11 +2129,6 @@ struct DOT_PRODUCT_4_V128
e.FMUL(dest.toQ().S4(), src1.S4(), src2.S4());
e.FADDP(dest.toQ().S4(), dest.toQ().S4(), dest.toQ().S4());
e.FADDP(dest.toS(), dest.toD().S2());
// Isolate lower lane
e.MOVI(Q0.D2(), RepImm(0b00'00'00'00));
e.INS(Q0.Selem()[0], dest.toQ().Selem()[0]);
e.MOV(dest.toQ().B16(), Q0.B16());
});
}
};