Missed Optimization: Fold smin(a - b, -1) + b into a clamp — select(b < 1, 0, smin(a, b - 1)) #166885
Description
define i16 @src(i16 %arg0, i16 %arg1) {
%v0 = sub nsw i16 0, %arg1
%v1 = sub nsw i16 %arg0, %arg1
%v2 = icmp slt i16 %arg1, 1
%v3 = tail call i16 @llvm.smin.i16(i16 %v1, i16 -1)
%v4 = select i1 %v2, i16 %v0, i16 %v3
%v5 = add nsw i16 %v4, %arg1
ret i16 %v5
}
define i16 @tgt(i16 %arg0, i16 %arg1) {
%v2 = icmp slt i16 %arg1, 1
%v3 = sub nsw i16 %arg1, 1
%v4 = tail call i16 @llvm.smin.i16(i16 %arg0, i16 %v3)
%v5 = select i1 %v2, i16 0, i16 %v4
ret i16 %v5
}
alive2: https://alive2.llvm.org/ce/z/xfxPEd
godbolt: https://godbolt.org/z/KcYKdf3Gj
pattern found in: https://github.com/dtcxzyw/llvm-opt-benchmark/blob/main/bench/ffmpeg/optimized/hcadec.ll