void test_vpaddls16 (void)
{
int32x2_t out_int32x2_t;
int16x4_t arg0_int16x4_t;
out_int32x2_t = vpaddl_s16 (arg0_int16x4_t);
}
/* return the sum of all elements in an array. This works by calculating 4 totals (one for each lane) and adding those at the end to get the final total */
int sum_array(int16_t *array, int size)
{
/* initialize the accumulator vector to zero */
int16x4_t acc = vdup_n_s16(0);
int32x2_t acc1;
int64x1_t acc2;
/* this implementation assumes the size of the array is a multiple of 4 */
assert((size % 4) == 0);
/* counting backwards gives better code */
for (; size != 0; size -= 4)
{
int16x4_t vec;
/* load 4 values in parallel from the array */
vec = vld1_s16(array);
/* increment the array pointer to the next element */
array += 4;
/* add the vector to the accumulator vector */
acc = vadd_s16(acc, vec);
}
/* calculate the total */
acc1 = vpaddl_s16(acc);
acc2 = vpaddl_s32(acc1);
/* return the total as an integer */
return (int)vget_lane_s64(acc2, 0);
}
static WEBP_INLINE uint32_t Select(const uint32_t* const c0,
const uint32_t* const c1,
const uint32_t* const c2) {
const uint8x8_t p0 = vreinterpret_u8_u64(vcreate_u64(*c0));
const uint8x8_t p1 = vreinterpret_u8_u64(vcreate_u64(*c1));
const uint8x8_t p2 = vreinterpret_u8_u64(vcreate_u64(*c2));
const uint8x8_t bc = vabd_u8(p1, p2); // |b-c|
const uint8x8_t ac = vabd_u8(p0, p2); // |a-c|
const int16x4_t sum_bc = vreinterpret_s16_u16(vpaddl_u8(bc));
const int16x4_t sum_ac = vreinterpret_s16_u16(vpaddl_u8(ac));
const int32x2_t diff = vpaddl_s16(vsub_s16(sum_bc, sum_ac));
const int32_t pa_minus_pb = vget_lane_s32(diff, 0);
return (pa_minus_pb <= 0) ? *c0 : *c1;
}
static WEBP_INLINE uint32_t Select(const uint32_t* const c0,
const uint32_t* const c1,
const uint32_t* const c2) {
const uint64x1_t C0 = { *c0, 0 }, C1 = { *c1, 0 }, C2 = { *c2, 0 };
const uint8x8_t p0 = vreinterpret_u8_u64(C0);
const uint8x8_t p1 = vreinterpret_u8_u64(C1);
const uint8x8_t p2 = vreinterpret_u8_u64(C2);
const uint8x8_t bc = vabd_u8(p1, p2); // |b-c|
const uint8x8_t ac = vabd_u8(p0, p2); // |a-c|
const int16x4_t sum_bc = vreinterpret_s16_u16(vpaddl_u8(bc));
const int16x4_t sum_ac = vreinterpret_s16_u16(vpaddl_u8(ac));
const int32x2_t diff = vpaddl_s16(vsub_s16(sum_bc, sum_ac));
int32_t pa_minus_pb;
vst1_lane_s32(&pa_minus_pb, diff, 0);
return (pa_minus_pb <= 0) ? *c0 : *c1;
}
