generic-library/vpx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Re-design quantization process

Browse Source 
This commit re-designs the quantization process for transform
coefficient blocks of size 4x4 to 16x16. It improves compression
performance for speed 7 by 3.85%. The SSSE3 version for the
new quantization process is included.

The average runtime of the 8x8 block quantization is reduced
from 285 cycles -> 255 cycles, i.e., over 10% faster.

Change-Id: I61278aa02efc70599b962d3314671db5b0446a50
This commit is contained in:
Jingning Han
2014-07-01 16:10:44 -07:00
parent 6643b8868d
commit 9ac2f66320
10 changed files with 310 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vp9/encoder/vp9_quantize.h
View File

@@ -28,6 +28,8 @@ typedef struct {
// if we want to deprecate the current use of y_quant.
DECLARE_ALIGNED(16, int16_t, y_quant_fp[QINDEX_RANGE][8]);
DECLARE_ALIGNED(16, int16_t, uv_quant_fp[QINDEX_RANGE][8]);
DECLARE_ALIGNED(16, int16_t, y_round_fp[QINDEX_RANGE][8]);
DECLARE_ALIGNED(16, int16_t, uv_round_fp[QINDEX_RANGE][8]);
DECLARE_ALIGNED(16, int16_t, uv_quant[QINDEX_RANGE][8]);
DECLARE_ALIGNED(16, int16_t, uv_quant_shift[QINDEX_RANGE][8]);