https://gcc.gnu.org/bugzilla/show_bug.cgi?id=103855
Bug ID: 103855
Summary: Missed optimization: 64bit division used instead of
32bit division
Product: gcc
Version: 12.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: zhaoweiliew at gmail dot com
Target Milestone: ---
Compiler Explorer link: https://gcc.godbolt.org/z/KvW8sMsqz
I compiled the following code with `x86-64 gcc (trunk) -std=c++20 -O3 -Wall
-Wextra -Werror`:
```
unsigned int optimized(unsigned int a, unsigned int b) {
return (unsigned long long)a / b;
}
unsigned int unoptimized(unsigned int a, unsigned int b) {
unsigned long long all = a;
return all / b;
}
```
This is the assembly output:
```
optimized(unsigned int, unsigned int):
mov eax, edi
xor edx, edx
div esi
ret
unoptimized(unsigned int, unsigned int):
mov eax, edi
mov esi, esi
xor edx, edx
div rsi
ret
```
GCC uses a 64-bit divide for `unoptimized()`, when a 32-bit divide would be
equivalent and faster. GCC uses a 32-bit divide for `optimized()`, which is
fine. Note that LLVM does a 32-bit division in both cases.
I would like to tackle this optimization, but am not sure how to go about doing
it. Could someone tell me/point me to resources that tell me what part of the
compiler and codebase I should be looking to optimize? Thanks!
