| Issue |
61644
|
| Summary |
Generate better code for absl::bit_ceil
|
| Labels |
llvm:instcombine,
missed-optimization
|
| Assignees |
|
| Reporter |
kazutakahirata
|
Compile:
```
// clang -std=c++20 -march=skylake -O3
#include "absl/numeric/bits.h"
unsigned foo(unsigned x) {
return absl::bit_ceil<unsigned>(x);
}
```
I get:
```
%2 = tail call i32 @llvm.ctpop.i32(i32 %0)
%3 = icmp eq i32 %2, 1
%4 = tail call i32 @llvm.ctlz.i32(i32 %0, i1 false)
%5 = select i1 %3, i32 31, i32 32
%6 = sub nsw i32 %5, %4
%7 = shl nuw i32 1, %6
ret i32 %7
```
```
f3 0f b8 c7 popcnt %edi,%eax
31 c9 xor %ecx,%ecx
83 f8 01 cmp $0x1,%eax
0f 94 c1 sete %cl
f3 0f bd c7 lzcnt %edi,%eax
01 c1 add %eax,%ecx
f7 d9 neg %ecx
b8 01 00 00 00 mov $0x1,%eax
c4 e2 71 f7 c0 shlx %ecx,%eax,%eax
```
We could generate the identical code as in https://github.com/llvm/llvm-project/issues/60802:
```
%dec = add i32 %x, -1
%ctlz = tail call i32 @llvm.ctlz.i32(i32 %dec, i1 false)
%sub = sub nsw i32 0, %ctlz
%and = and i32 %1, 31
%sel = shl nuw i32 1, %and
ret i32 %sel
```
https://alive2.llvm.org/ce/z/7aqxmc
![]()
_______________________________________________
llvm-bugs mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
