summaryrefslogtreecommitdiff
path: root/libavcodec/i386/fdct_mmx.c
diff options
context:
space:
mode:
Diffstat (limited to 'libavcodec/i386/fdct_mmx.c')
-rw-r--r--libavcodec/i386/fdct_mmx.c174
1 files changed, 87 insertions, 87 deletions
diff --git a/libavcodec/i386/fdct_mmx.c b/libavcodec/i386/fdct_mmx.c
index 6a13090a13..f3023549a2 100644
--- a/libavcodec/i386/fdct_mmx.c
+++ b/libavcodec/i386/fdct_mmx.c
@@ -5,7 +5,7 @@
* SSE2 optimization is Copyright (c) 2004 Denes Balatoni.
*
* from fdctam32.c - AP922 MMX(3D-Now) forward-DCT
- *
+ *
* Intel Application Note AP-922 - fast, precise implementation of DCT
* http://developer.intel.com/vtune/cbts/appnotes.htm
*
@@ -51,7 +51,7 @@ static const int64_t fdct_one_corr ATTR_ALIGN(8) = 0x0001000100010001LL;
static const int32_t fdct_r_row[2] ATTR_ALIGN(8) = {RND_FRW_ROW, RND_FRW_ROW };
-struct
+struct
{
const int32_t fdct_r_row_sse2[4] ATTR_ALIGN(16);
} fdct_r_row_sse2 ATTR_ALIGN(16)=
@@ -61,90 +61,90 @@ struct
//static const long fdct_r_row_sse2[4] ATTR_ALIGN(16) = {RND_FRW_ROW, RND_FRW_ROW, RND_FRW_ROW, RND_FRW_ROW};
static const int16_t tab_frw_01234567[] ATTR_ALIGN(8) = { // forward_dct coeff table
- 16384, 16384, 22725, 19266,
- 16384, 16384, 12873, 4520,
- 21407, 8867, 19266, -4520,
- -8867, -21407, -22725, -12873,
- 16384, -16384, 12873, -22725,
- -16384, 16384, 4520, 19266,
- 8867, -21407, 4520, -12873,
- 21407, -8867, 19266, -22725,
-
- 22725, 22725, 31521, 26722,
- 22725, 22725, 17855, 6270,
- 29692, 12299, 26722, -6270,
- -12299, -29692, -31521, -17855,
- 22725, -22725, 17855, -31521,
- -22725, 22725, 6270, 26722,
- 12299, -29692, 6270, -17855,
- 29692, -12299, 26722, -31521,
-
- 21407, 21407, 29692, 25172,
- 21407, 21407, 16819, 5906,
- 27969, 11585, 25172, -5906,
- -11585, -27969, -29692, -16819,
- 21407, -21407, 16819, -29692,
- -21407, 21407, 5906, 25172,
- 11585, -27969, 5906, -16819,
- 27969, -11585, 25172, -29692,
-
- 19266, 19266, 26722, 22654,
- 19266, 19266, 15137, 5315,
- 25172, 10426, 22654, -5315,
- -10426, -25172, -26722, -15137,
- 19266, -19266, 15137, -26722,
- -19266, 19266, 5315, 22654,
- 10426, -25172, 5315, -15137,
- 25172, -10426, 22654, -26722,
-
- 16384, 16384, 22725, 19266,
- 16384, 16384, 12873, 4520,
- 21407, 8867, 19266, -4520,
- -8867, -21407, -22725, -12873,
- 16384, -16384, 12873, -22725,
- -16384, 16384, 4520, 19266,
- 8867, -21407, 4520, -12873,
- 21407, -8867, 19266, -22725,
-
- 19266, 19266, 26722, 22654,
- 19266, 19266, 15137, 5315,
- 25172, 10426, 22654, -5315,
- -10426, -25172, -26722, -15137,
- 19266, -19266, 15137, -26722,
- -19266, 19266, 5315, 22654,
- 10426, -25172, 5315, -15137,
- 25172, -10426, 22654, -26722,
-
- 21407, 21407, 29692, 25172,
- 21407, 21407, 16819, 5906,
- 27969, 11585, 25172, -5906,
- -11585, -27969, -29692, -16819,
- 21407, -21407, 16819, -29692,
- -21407, 21407, 5906, 25172,
- 11585, -27969, 5906, -16819,
- 27969, -11585, 25172, -29692,
-
- 22725, 22725, 31521, 26722,
- 22725, 22725, 17855, 6270,
- 29692, 12299, 26722, -6270,
- -12299, -29692, -31521, -17855,
- 22725, -22725, 17855, -31521,
- -22725, 22725, 6270, 26722,
- 12299, -29692, 6270, -17855,
- 29692, -12299, 26722, -31521,
+ 16384, 16384, 22725, 19266,
+ 16384, 16384, 12873, 4520,
+ 21407, 8867, 19266, -4520,
+ -8867, -21407, -22725, -12873,
+ 16384, -16384, 12873, -22725,
+ -16384, 16384, 4520, 19266,
+ 8867, -21407, 4520, -12873,
+ 21407, -8867, 19266, -22725,
+
+ 22725, 22725, 31521, 26722,
+ 22725, 22725, 17855, 6270,
+ 29692, 12299, 26722, -6270,
+ -12299, -29692, -31521, -17855,
+ 22725, -22725, 17855, -31521,
+ -22725, 22725, 6270, 26722,
+ 12299, -29692, 6270, -17855,
+ 29692, -12299, 26722, -31521,
+
+ 21407, 21407, 29692, 25172,
+ 21407, 21407, 16819, 5906,
+ 27969, 11585, 25172, -5906,
+ -11585, -27969, -29692, -16819,
+ 21407, -21407, 16819, -29692,
+ -21407, 21407, 5906, 25172,
+ 11585, -27969, 5906, -16819,
+ 27969, -11585, 25172, -29692,
+
+ 19266, 19266, 26722, 22654,
+ 19266, 19266, 15137, 5315,
+ 25172, 10426, 22654, -5315,
+ -10426, -25172, -26722, -15137,
+ 19266, -19266, 15137, -26722,
+ -19266, 19266, 5315, 22654,
+ 10426, -25172, 5315, -15137,
+ 25172, -10426, 22654, -26722,
+
+ 16384, 16384, 22725, 19266,
+ 16384, 16384, 12873, 4520,
+ 21407, 8867, 19266, -4520,
+ -8867, -21407, -22725, -12873,
+ 16384, -16384, 12873, -22725,
+ -16384, 16384, 4520, 19266,
+ 8867, -21407, 4520, -12873,
+ 21407, -8867, 19266, -22725,
+
+ 19266, 19266, 26722, 22654,
+ 19266, 19266, 15137, 5315,
+ 25172, 10426, 22654, -5315,
+ -10426, -25172, -26722, -15137,
+ 19266, -19266, 15137, -26722,
+ -19266, 19266, 5315, 22654,
+ 10426, -25172, 5315, -15137,
+ 25172, -10426, 22654, -26722,
+
+ 21407, 21407, 29692, 25172,
+ 21407, 21407, 16819, 5906,
+ 27969, 11585, 25172, -5906,
+ -11585, -27969, -29692, -16819,
+ 21407, -21407, 16819, -29692,
+ -21407, 21407, 5906, 25172,
+ 11585, -27969, 5906, -16819,
+ 27969, -11585, 25172, -29692,
+
+ 22725, 22725, 31521, 26722,
+ 22725, 22725, 17855, 6270,
+ 29692, 12299, 26722, -6270,
+ -12299, -29692, -31521, -17855,
+ 22725, -22725, 17855, -31521,
+ -22725, 22725, 6270, 26722,
+ 12299, -29692, 6270, -17855,
+ 29692, -12299, 26722, -31521,
};
-struct
+struct
{
const int16_t tab_frw_01234567_sse2[256] ATTR_ALIGN(16);
} tab_frw_01234567_sse2 ATTR_ALIGN(16) =
{{
-//static const int16_t tab_frw_01234567_sse2[] ATTR_ALIGN(16) = { // forward_dct coeff table
+//static const int16_t tab_frw_01234567_sse2[] ATTR_ALIGN(16) = { // forward_dct coeff table
#define TABLE_SSE2 C4, C4, C1, C3, -C6, -C2, -C1, -C5, \
C4, C4, C5, C7, C2, C6, C3, -C7, \
-C4, C4, C7, C3, C6, -C2, C7, -C5, \
- C4, -C4, C5, -C1, C2, -C6, C3, -C1,
-// c1..c7 * cos(pi/4) * 2^15
+ C4, -C4, C5, -C1, C2, -C6, C3, -C1,
+// c1..c7 * cos(pi/4) * 2^15
#define C1 22725
#define C2 21407
#define C3 19266
@@ -355,17 +355,17 @@ static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
"movq \\i(%0), %%xmm2 \n\t"
"movq \\i+8(%0), %%xmm0 \n\t"
"movdqa \\t+32(%1), %%xmm3 \n\t"
- "movdqa \\t+48(%1), %%xmm7 \n\t"
+ "movdqa \\t+48(%1), %%xmm7 \n\t"
"movdqa \\t(%1), %%xmm4 \n\t"
- "movdqa \\t+16(%1), %%xmm5 \n\t"
+ "movdqa \\t+16(%1), %%xmm5 \n\t"
".endm \n\t"
".macro FDCT_ROW_SSE2_H2 i t \n\t"
"movq \\i(%0), %%xmm2 \n\t"
"movq \\i+8(%0), %%xmm0 \n\t"
"movdqa \\t+32(%1), %%xmm3 \n\t"
- "movdqa \\t+48(%1), %%xmm7 \n\t"
+ "movdqa \\t+48(%1), %%xmm7 \n\t"
".endm \n\t"
- ".macro FDCT_ROW_SSE2 i \n\t"
+ ".macro FDCT_ROW_SSE2 i \n\t"
"movq %%xmm2, %%xmm1 \n\t"
"pshuflw $27, %%xmm0, %%xmm0 \n\t"
"paddsw %%xmm0, %%xmm1 \n\t"
@@ -376,7 +376,7 @@ static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
"pmaddwd %%xmm1, %%xmm7 \n\t"
"pmaddwd %%xmm5, %%xmm2 \n\t"
"pmaddwd %%xmm4, %%xmm1 \n\t"
- "paddd %%xmm7, %%xmm3 \n\t"
+ "paddd %%xmm7, %%xmm3 \n\t"
"paddd %%xmm2, %%xmm1 \n\t"
"paddd %%xmm6, %%xmm3 \n\t"
"paddd %%xmm6, %%xmm1 \n\t"
@@ -384,8 +384,8 @@ static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
"psrad %3, %%xmm1 \n\t"
"packssdw %%xmm3, %%xmm1 \n\t"
"movdqa %%xmm1, \\i(%4) \n\t"
- ".endm \n\t"
- "movdqa (%2), %%xmm6 \n\t"
+ ".endm \n\t"
+ "movdqa (%2), %%xmm6 \n\t"
"FDCT_ROW_SSE2_H1 0 0 \n\t"
"FDCT_ROW_SSE2 0 \n\t"
"FDCT_ROW_SSE2_H2 64 0 \n\t"
@@ -411,7 +411,7 @@ static always_inline void fdct_row_sse2(const int16_t *in, int16_t *out)
}
static always_inline void fdct_row_mmx2(const int16_t *in, int16_t *out, const int16_t *table)
-{
+{
pshufw_m2r(*(in + 4), mm5, 0x1B);
movq_m2r(*(in + 0), mm0);
movq_r2r(mm0, mm1);
@@ -454,7 +454,7 @@ static always_inline void fdct_row_mmx2(const int16_t *in, int16_t *out, const i
}
static always_inline void fdct_row_mmx(const int16_t *in, int16_t *out, const int16_t *table)
-{
+{
//FIXME reorder (i dont have a old mmx only cpu here to benchmark ...)
movd_m2r(*(in + 6), mm1);
punpcklwd_m2r(*(in + 4), mm1);
@@ -547,7 +547,7 @@ void ff_fdct_mmx2(int16_t *block)
}
}
-void ff_fdct_sse2(int16_t *block)
+void ff_fdct_sse2(int16_t *block)
{
int64_t align_tmp[16] ATTR_ALIGN(8);
int16_t * const block_tmp= (int16_t*)align_tmp;
generated by cgit v1.2.3 (git 2.39.1) at 2024-06-01 02:33:19 +0000