[llvm] [RISCV] Set AllocationPriority in line with LMUL (PR #131176)

Pengcheng Wang via llvm-commits llvm-commits at lists.llvm.org
Thu Mar 13 20:58:32 PDT 2025


================
@@ -574,62 +559,85 @@ define {<8 x i64>, <8 x i64>, <8 x i64>, <8 x i64>, <8 x i64>, <8 x i64>} @load_
 ; RV32-NEXT:    lui a1, %hi(.LCPI8_9)
 ; RV32-NEXT:    addi a1, a1, %lo(.LCPI8_9)
 ; RV32-NEXT:    vsetivli zero, 16, e16, m2, ta, ma
-; RV32-NEXT:    vle16.v v28, (a2)
+; RV32-NEXT:    vle16.v v14, (a2)
+; RV32-NEXT:    csrr a2, vlenb
+; RV32-NEXT:    li a3, 72
+; RV32-NEXT:    mul a2, a2, a3
+; RV32-NEXT:    add a2, sp, a2
+; RV32-NEXT:    addi a2, a2, 16
+; RV32-NEXT:    vs2r.v v14, (a2) # Unknown-size Folded Spill
----------------
wangpc-pp wrote:

Use less memory but cause more spills?

https://github.com/llvm/llvm-project/pull/131176


More information about the llvm-commits mailing list