================
@@ -29,3 +29,102 @@ let Predicates = [HasStdExtZvzip], Constraints =
"@earlyclobber $vd" in {
def VPAIRO_VV : VALUVV<0b001111, OPMVV, "vpairo.vv">;
}
} // Predicates = [HasStdExtZvzip]
+
+defset list<VTypeInfoToWide> AllZvzipVectors = {
+ def : VTypeInfoToWide<VI8MF8, VI8MF4>;
+ def : VTypeInfoToWide<VI8MF4, VI8MF2>;
+ def : VTypeInfoToWide<VI8MF2, VI8M1>;
+ def : VTypeInfoToWide<VI8M1, VI8M2>;
+ def : VTypeInfoToWide<VI8M2, VI8M4>;
+ def : VTypeInfoToWide<VI8M4, VI8M8>;
+
+ def : VTypeInfoToWide<VI16MF4, VI16MF2>;
+ def : VTypeInfoToWide<VI16MF2, VI16M1>;
+ def : VTypeInfoToWide<VI16M1, VI16M2>;
+ def : VTypeInfoToWide<VI16M2, VI16M4>;
+ def : VTypeInfoToWide<VI16M4, VI16M8>;
+
+ def : VTypeInfoToWide<VI32MF2, VI32M1>;
+ def : VTypeInfoToWide<VI32M1, VI32M2>;
+ def : VTypeInfoToWide<VI32M2, VI32M4>;
+ def : VTypeInfoToWide<VI32M4, VI32M8>;
+
+ def : VTypeInfoToWide<VI64M1, VI64M2>;
+ def : VTypeInfoToWide<VI64M2, VI64M4>;
+ def : VTypeInfoToWide<VI64M4, VI64M8>;
+
+ // Floating-point 16-bit
+ def : VTypeInfoToWide<VF16MF4, VF16MF2>;
+ def : VTypeInfoToWide<VF16MF2, VF16M1>;
+ def : VTypeInfoToWide<VF16M1, VF16M2>;
+ def : VTypeInfoToWide<VF16M2, VF16M4>;
+ def : VTypeInfoToWide<VF16M4, VF16M8>;
----------------
wangpc-pp wrote:
Yeah, we should add BF16 and OCP FP8 formats in the future. I need to
investigate how `Zvfbfa` interact with existing instructions.
https://github.com/llvm/llvm-project/pull/186342
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
