diff options
| author | Yunqing Wang <yunqingwang@google.com> | 2013-03-21 09:22:22 -0700 |
|---|---|---|
| committer | Yunqing Wang <yunqingwang@google.com> | 2013-03-21 11:44:05 -0700 |
| commit | ec3100661c090c4f46fe231cef51897dc6660551 (patch) | |
| tree | 99315d5df54cb2f9c254203e695611a0102e6a31 /vp9/decoder | |
| parent | 66eff0aa3878c0f3aa6a93d0f171ffd63356c2f7 (diff) | |
| download | libvpx-ec3100661c090c4f46fe231cef51897dc6660551.tar libvpx-ec3100661c090c4f46fe231cef51897dc6660551.tar.gz libvpx-ec3100661c090c4f46fe231cef51897dc6660551.tar.bz2 libvpx-ec3100661c090c4f46fe231cef51897dc6660551.zip | |
Optimize 16x16 idct function
Wrote sse2 version of vp9_short_idct16x16 function. Compared to c
version, the sse2 version is over 2.5X faster.
Change-Id: I38536e2b846427a2cc5c5423aaf305fd0e605d61
Diffstat (limited to 'vp9/decoder')
| -rw-r--r-- | vp9/decoder/vp9_dequantize.c | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/vp9/decoder/vp9_dequantize.c b/vp9/decoder/vp9_dequantize.c index 2b8c61f1e..d38f9165c 100644 --- a/vp9/decoder/vp9_dequantize.c +++ b/vp9/decoder/vp9_dequantize.c @@ -273,7 +273,7 @@ void vp9_ht_dequant_idct_add_16x16_c(TX_TYPE tx_type, int16_t *input, vp9_short_iht16x16(input, output, 16, tx_type); // the idct halves ( >> 1) the pitch - // vp9_short_idct16x16_c(input, output, 32); + // vp9_short_idct16x16(input, output, 32); vpx_memset(input, 0, 512); @@ -296,7 +296,7 @@ void vp9_dequant_idct_add_16x16_c(int16_t *input, const int16_t *dq, int16_t in = input[0] * dq[0]; int16_t out; /* Note: the idct1 will need to be modified accordingly whenever - * vp9_short_idct16x16_c() is modified. */ + * vp9_short_idct16x16() is modified. */ vp9_short_idct1_16x16_c(&in, &out); input[0] = 0; @@ -333,7 +333,7 @@ void vp9_dequant_idct_add_16x16_c(int16_t *input, const int16_t *dq, input[i] *= dq[1]; // the idct halves ( >> 1) the pitch - vp9_short_idct16x16_c(input, output, 16 << 1); + vp9_short_idct16x16(input, output, 16 << 1); vpx_memset(input, 0, 512); |