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

Reduce num of buffer swap calls in idct8_1d_sse2

Browse Source 
This commit merges the initial buffer swap operations in idct8_1d_sse2
into the array transpose step, hence reducing number of instructions
therein.

Change-Id: I219f6f50813390d2ec3ee37eecf2a4a2b44ae479
This commit is contained in:
Jingning Han 2014-01-03 12:09:34 -08:00
parent 1bb11781e2
commit 0b1a27135a
1 changed files with 2 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
vp9/common/x86/vp9_idct_intrin_sse2.c
View File

@ -692,18 +692,9 @@ static void idct8_1d_sse2(__m128i *in) {
__m128i stp2_0, stp2_1, stp2_2, stp2_3, stp2_4, stp2_5, stp2_6, stp2_7;
__m128i tmp0, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7;
in0 = in[0];
in1 = in[1];
in2 = in[2];
in3 = in[3];
in4 = in[4];
in5 = in[5];
in6 = in[6];
in7 = in[7];
// 8x8 Transpose is copied from vp9_fdct8x8_sse2()
TRANSPOSE_8X8(in0, in1, in2, in3, in4, in5, in6, in7, in0, in1, in2, in3,
in4, in5, in6, in7);
TRANSPOSE_8X8(in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7],
in0, in1, in2, in3, in4, in5, in6, in7);
// 4-stage 1D idct8x8
IDCT8_1D