https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122935
Bug ID: 122935
Summary: Missed optimization: Optimize (a * b) ==/!= 0 to
logical checks for signed/widening multiplications.
Product: gcc
Version: 16.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: target
Assignee: unassigned at gcc dot gnu.org
Reporter: chendongyan at isrc dot iscas.ac.cn
Target Milestone: ---
This is a missed optimization feature report.
GCC currently misses an optimization opportunity to transform multiplication
equality checks (a * b) == 0 (and != 0) into logical checks of the operands.
Generally, multiplication operations have a significantly higher latency (cost)
than logical operations. Implementing this optimization can improve performance
by reducing the critical path latency and increasing instruction-level
parallelism.
Architectures that are known to benefit from this include x86-64 and RISC-V.
Compile the following code with -O3
```c
#include <stdbool.h>
#include <stdint.h>
bool foo(int32_t a, int32_t b) {
return ((int32_t)a * b) == 0;
}
bool foo2(uint8_t a, uint8_t b) {
return ((uint32_t)a * b) == 0;
}
bool foo3(int32_t a, int32_t b) {
return ((int32_t)a * b) != 0;
}
```
x86-64：https://godbolt.org/z/xzEaq7Wa4
risc-v：https://godbolt.org/z/WvTEvohc5
(https://gcc.gnu.org/pipermail/gcc-patches/2025-November/702158.html)
This optimization is already implemented in Clang/LLVM.
