[clang] [llvm] [RISCV][P-ext] Add scalar merge intrinsics (PR #239119)

via cfe-commits cfe-commits at lists.llvm.org
Thu Nov 5 05:16:33 PST 2026


================
@@ -265,6 +265,11 @@ def pmerge_u32x2 : RISCVBuiltin<"_Vector<2, unsigned int>(_Vector<2, unsigned in
 let Features = "experimental-p,64bit" in
 def rev16_64 : RISCVBuiltin<"uint64_t(uint64_t)">;
 
+// Scalar Merge
+def merge_u32 : RISCVBuiltin<"uint32_t(uint32_t, uint32_t, uint32_t)">;
+let Features = "experimental-p,64bit" in
+def merge_u64 : RISCVBuiltin<"uint64_t(uint64_t, uint64_t, uint64_t)">;
----------------
sihuan wrote:

Should this be RV64-only? The spec lists the i64 merge as "merge/mvm/mvmn (RV64), 2x merge/mvm/mvmn (RV32)". The packed pmerge lowering already does that split, so the i64 form should be doable the same way. Is RV32 intended as a follow-up?

https://github.com/llvm/llvm-project/pull/239119


More information about the cfe-commits mailing list