How to auto-vectorize strided writes with GCC?

Question

When compiled with GCC 5.2 using -std=c99, -O3, and -mavx2, the following code sample auto-vectorizes (assembly here):

#include <stdint.h>

void test(uint32_t *restrict a,
          uint32_t *restrict b) {
  uint32_t *a_aligned = __builtin_assume_aligned(a, 32);
  uint32_t *b_aligned = __builtin_assume_aligned(b, 32);

  for (int i = 0; i < (1L << 10); i += 2) {
    a_aligned[i] = 42 * b_aligned[i];
    a_aligned[i+1] = 3 * a_aligned[i+1];
  }
}

But the following code sample does not auto-vectorize (assembly here):

#include <stdint.h>

void test(uint32_t *restrict a,
          uint32_t *restrict b) {
  uint32_t *a_aligned = __builtin_assume_aligned(a, 32);
  uint32_t *b_aligned = __builtin_assume_aligned(b, 32);

  for (int i = 0; i < (1L << 10); i += 2) {
    a_aligned[i] = 42 * b_aligned[i];
    a_aligned[i+1] = a_aligned[i+1];
  }
}

The only difference between the samples is the scaling factor to a_aligned[i+1].

This was also the case for GCC 4.8, 4.9, and 5.1. Adding volatile to a_aligned's declaration inhibits auto-vectorization completely. The first sample consistently runs faster than the second for us, with a more pronounced speedup for smaller types (e.g. uint8_t instead of uint32_t).

Is there a way to make the second code sample auto-vectorize with GCC?

Answer 1

The following version vectorises, but that's ugly if you ask me...

#include <stdint.h>

void test(uint32_t *a, uint32_t *aa,
          uint32_t *restrict b) {
  #pragma omp simd aligned(a,aa,b:32)
  for (int i = 0; i < (1L << 10); i += 2) {
    a[i] = 2 * b[i];
    a[i+1] = aa[i+1];
  }
}

To compile with -fopenmp and to call with test(a, a, b).