[llvm-bugs] [Bug 47200] New: Clang failed to inline memset for SSE2 and AVX2

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=47200

            Bug ID: 47200
           Summary: Clang failed to inline memset for SSE2 and AVX2
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: zufuliu at 163.com
                CC: craig.topper at gmail.com, llvm-bugs at lists.llvm.org,
                    llvm-dev at redking.me.uk, spatel+llvm at rotateright.com

following code (online at https://godbolt.org/z/9eWq78), when CNT is larger
than 1 (for both AVX2 and SSE2), clang emits call to memset.
gcc inline the memset call when buffer size is <= 256, e.g. CNT <= 8 for AVX2,
or CNT <= 16 for SSE2.


#include <string.h>
#include <immintrin.h>

#define CNT 2

int IsUTF7(const char *data, size_t length) {
#if defined(__AVX2__)
    char buffer[CNT*sizeof(__m256i)] = {0};
    memcpy(buffer, data, length);
    __m256i chunk = _mm256_loadu_si256((__m256i *)buffer);
    return _mm256_movemask_epi8(chunk) == 0;
#else
    char buffer[CNT*sizeof(__m128i)] = {0};
    memcpy(buffer, data, length);
    __m128i chunk = _mm_loadu_si128((__m128i *)buffer);
    return _mm_movemask_epi8(chunk) == 0;
#endif
}

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20200817/10c00407/attachment.html>