SIMD intrinsic _mm512_mask_cmp_epi8_mask returns wrong result with -O3 compile option

[howin]

Running the following code gets wrong result with `-O3` compilation option.

SIMD intrinsic `_mm512_mask_cmp_epi8_mask` returns incorrect result.

OS: CentOS Linux release 7.9 (Final)

CPU: Intel(R) Xeon(R) Platinum 8255C CPU @ 2.50GHz

G++: 7.3.1

```

#include <immintrin.h>

#include <bitset>

#include <iostream>

#include <vector>

int main() {

std::vector<int8_t> a_data(64, 5);

std::vector<int8_t> b_data(64, 10);

__mmask64 mask = ((uint64_t)1 << 63) | (((uint64_t)1 << 63) - 1);

__m512i a = _mm512_loadu_si512(a_data.data());

__m512i b = _mm512_loadu_si512(b_data.data());

__mmask64 res1 = _mm512_mask_cmp_epi8_mask(mask, a, b, _MM_CMPINT_LT);

__mmask64 res2 = _mm512_mask_cmplt_epi8_mask(mask, a, b);

std::cout << "res1: " << std::hex << res1 << '\n' << "res2: " << std::hex << res2 << '\n';

}

```

[HemanthCH_Intel]

Hi,

Thanks for posting in Intel communities.

We did not observe any changes in the result even with -the O3 flag enabled using Intel compilers(icpc,icpx). Could you please try with the Intel Compilers and let us know if you face any issues?

Please refer to the above screenshot for  more details.

Thanks & Regards,

Hemanth

[howin]

How about using G++ 7.3.1？What I'm confused about is only this SIMD instrinsic returns an unexpected result.

[HemanthCH_Intel]

Hi,

As we can see that you are compiling the code with Gcc, but works as expected with Intel C++ compilers. This forum is intended to support the queries related to the Intel C++ Compilers and other Intel oneAPI products. So, could you please let us know whether you have any issues with Intel oneAPI products?

Thanks & Regards,

Hemanth

[HemanthCH_Intel]

Hi,

We assume that your issue is resolved. If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Thanks & Regards,

Hemanth