| 1 | #include <iostream> |
|---|
| 2 | #include <iomanip> |
|---|
| 3 | |
|---|
| 4 | #include <x86intrin.h> |
|---|
| 5 | |
|---|
| 6 | int main() |
|---|
| 7 | { |
|---|
| 8 | #ifdef __SSE__ |
|---|
| 9 | { |
|---|
| 10 | float __attribute__ ((aligned(16))) Q[4]={1.f,2.f,3.f,4.f}; |
|---|
| 11 | float __attribute__ ((aligned(16))) Z[4]; |
|---|
| 12 | __m128 v4Q = _mm_load_ps(Q); |
|---|
| 13 | __m128 v4X = _mm_mul_ps(v4Q,v4Q); |
|---|
| 14 | _mm_store_ps(Z,v4X); |
|---|
| 15 | for(int i=0; i!=4; ++i) |
|---|
| 16 | std::cout << " Q["<<i<<"]="<<std::setw(6)<<Q[i] |
|---|
| 17 | << " Z["<<i<<"]="<<std::setw(6)<<Z[i]<<'\n'; |
|---|
| 18 | } |
|---|
| 19 | #endif |
|---|
| 20 | #ifdef __AVX__ |
|---|
| 21 | { |
|---|
| 22 | float __attribute__ ((aligned(32))) Q[8]={1.f,2.f,3.f,4.f,5.f,6.f,7.f,8.f}; |
|---|
| 23 | float __attribute__ ((aligned(32))) Z[8]; |
|---|
| 24 | __m256 v8Q = _mm256_load_ps(Q); |
|---|
| 25 | __m256 v8X = _mm256_mul_ps(v8Q,v8Q); |
|---|
| 26 | _mm256_store_ps(Z,v8X); |
|---|
| 27 | for(int i=0; i!=8; ++i) |
|---|
| 28 | std::cout << " Q["<<i<<"]="<<std::setw(6)<<Q[i] |
|---|
| 29 | << " Z["<<i<<"]="<<std::setw(6)<<Z[i]<<'\n'; |
|---|
| 30 | } |
|---|
| 31 | #else |
|---|
| 32 | # warning __AVX__ not #defined: this will not test cctools as |
|---|
| 33 | # warning did you compile with -march=native? |
|---|
| 34 | # warning if so, does your computer support AVX instructions? |
|---|
| 35 | # warning (it will if "sysctl machdep.cpu.features" has "AVX1.0)" |
|---|
| 36 | #endif |
|---|
| 37 | } |
|---|
