dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2024-11-20 04:18:26 +01:00
Code Issues Actions Packages Projects Releases Wiki Activity

dec_neon: use vld?_lane instead of vset?_lane

Browse Source 
results in fewer instructions, small speed improvement

Change-Id: I61ab48d09a5ce7c5158eac8244d28287457edc7a
This commit is contained in:
James Zern 2014-04-02 23:03:18 -07:00
parent 4b0f2dae6f
commit fa52d7525f
1 changed files with 10 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/dsp/dec_neon.c
View File

@ -114,12 +114,12 @@ static WEBP_INLINE void Load4x16(const uint8_t* const src, int stride,
#else
#define LOAD_LANE_32b(VALUE, LANE) do { \
(VALUE) = vset_lane_u32(*(const uint32_t*)src, (VALUE), (LANE)); \
(VALUE) = vld1_lane_u32((const uint32_t*)src, (VALUE), (LANE)); \
src += stride; \
} while (0)
#define LOADQ_LANE_32b(VALUE, LANE) do { \
(VALUE) = vsetq_lane_u32(*(const uint32_t*)src, (VALUE), (LANE)); \
(VALUE) = vld1q_lane_u32((const uint32_t*)src, (VALUE), (LANE)); \
src += stride; \
} while (0)
@ -1142,10 +1142,10 @@ static void TransformDC(const int16_t* in, uint8_t* dst) {
uint32x2_t dst23 = {0, 0};
// Load the source pixels.
dst01 = vset_lane_u32(*(uint32_t*)(dst + 0 * BPS), dst01, 0);
dst23 = vset_lane_u32(*(uint32_t*)(dst + 2 * BPS), dst23, 0);
dst01 = vset_lane_u32(*(uint32_t*)(dst + 1 * BPS), dst01, 1);
dst23 = vset_lane_u32(*(uint32_t*)(dst + 3 * BPS), dst23, 1);
dst01 = vld1_lane_u32((uint32_t*)(dst + 0 * BPS), dst01, 0);
dst23 = vld1_lane_u32((uint32_t*)(dst + 2 * BPS), dst23, 0);
dst01 = vld1_lane_u32((uint32_t*)(dst + 1 * BPS), dst01, 1);
dst23 = vld1_lane_u32((uint32_t*)(dst + 3 * BPS), dst23, 1);
{
// Convert to 16b.
@ -1248,10 +1248,10 @@ static void TransformAC3(const int16_t* in, uint8_t* dst) {
uint32x2_t dst23 = {0, 0};
// Load the source pixels.
dst01 = vset_lane_u32(*(uint32_t*)(dst + 0 * BPS), dst01, 0);
dst23 = vset_lane_u32(*(uint32_t*)(dst + 2 * BPS), dst23, 0);
dst01 = vset_lane_u32(*(uint32_t*)(dst + 1 * BPS), dst01, 1);
dst23 = vset_lane_u32(*(uint32_t*)(dst + 3 * BPS), dst23, 1);
dst01 = vld1_lane_u32((uint32_t*)(dst + 0 * BPS), dst01, 0);
dst23 = vld1_lane_u32((uint32_t*)(dst + 2 * BPS), dst23, 0);
dst01 = vld1_lane_u32((uint32_t*)(dst + 1 * BPS), dst01, 1);
dst23 = vld1_lane_u32((uint32_t*)(dst + 3 * BPS), dst23, 1);
{
// Convert to 16b.