INIT_MMX
; ff_h264_idct_add_mmx(uint8_t *dst, int16_t *block, int stride)
-cglobal h264_idct_add_mmx, 3, 3, 0
+cglobal h264_idct_add_8_mmx, 3, 3, 0
IDCT4_ADD r0, r1, r2
RET
SUMSUB_BA w, 0, 4
SUMSUB_BA w, 3, 2
SUMSUB_BA w, 1, 5
- SWAP 7, 6, 4, 5, 2, 3, 1, 0 ; 70315246 -> 01234567
+ SWAP 7, 6, 4, 5, 2, 3, 1, 0 ; 70315246 -> 01234567
%endmacro
%macro IDCT8_1D_FULL 1
INIT_MMX
; ff_h264_idct8_add_mmx(uint8_t *dst, int16_t *block, int stride)
-cglobal h264_idct8_add_mmx, 3, 4, 0
+cglobal h264_idct8_add_8_mmx, 3, 4, 0
%assign pad 128+4-(stack_offset&7)
SUB rsp, pad
INIT_XMM
; ff_h264_idct8_add_sse2(uint8_t *dst, int16_t *block, int stride)
-cglobal h264_idct8_add_sse2, 3, 4, 10
+cglobal h264_idct8_add_8_sse2, 3, 4, 10
IDCT8_ADD_SSE r0, r1, r2, r3
RET
packuswb m1, m1
%endmacro
-%macro DC_ADD_MMX2_OP 3-4
+%macro DC_ADD_MMX2_OP 4
%1 m2, [%2 ]
%1 m3, [%2+%3 ]
%1 m4, [%2+%3*2]
INIT_MMX
; ff_h264_idct_dc_add_mmx2(uint8_t *dst, int16_t *block, int stride)
-cglobal h264_idct_dc_add_mmx2, 3, 3, 0
+cglobal h264_idct_dc_add_8_mmx2, 3, 3, 0
DC_ADD_MMX2_INIT r1, r2
DC_ADD_MMX2_OP movh, r0, r2, r1
RET
; ff_h264_idct8_dc_add_mmx2(uint8_t *dst, int16_t *block, int stride)
-cglobal h264_idct8_dc_add_mmx2, 3, 3, 0
+cglobal h264_idct8_dc_add_8_mmx2, 3, 3, 0
DC_ADD_MMX2_INIT r1, r2
DC_ADD_MMX2_OP mova, r0, r2, r1
lea r0, [r0+r2*4]
; ff_h264_idct_add16_mmx(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16_mmx, 5, 7, 0
+cglobal h264_idct_add16_8_mmx, 5, 7, 0
xor r5, r5
%ifdef PIC
lea r11, [scan8_mem]
; ff_h264_idct8_add4_mmx(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct8_add4_mmx, 5, 7, 0
+cglobal h264_idct8_add4_8_mmx, 5, 7, 0
%assign pad 128+4-(stack_offset&7)
SUB rsp, pad
; ff_h264_idct_add16_mmx2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16_mmx2, 5, 7, 0
+cglobal h264_idct_add16_8_mmx2, 5, 7, 0
xor r5, r5
%ifdef PIC
lea r11, [scan8_mem]
; ff_h264_idct_add16intra_mmx(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16intra_mmx, 5, 7, 0
+cglobal h264_idct_add16intra_8_mmx, 5, 7, 0
xor r5, r5
%ifdef PIC
lea r11, [scan8_mem]
; ff_h264_idct_add16intra_mmx2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16intra_mmx2, 5, 7, 0
+cglobal h264_idct_add16intra_8_mmx2, 5, 7, 0
xor r5, r5
%ifdef PIC
lea r11, [scan8_mem]
; ff_h264_idct8_add4_mmx2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct8_add4_mmx2, 5, 7, 0
+cglobal h264_idct8_add4_8_mmx2, 5, 7, 0
%assign pad 128+4-(stack_offset&7)
SUB rsp, pad
INIT_XMM
; ff_h264_idct8_add4_sse2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct8_add4_sse2, 5, 7, 10
+cglobal h264_idct8_add4_8_sse2, 5, 7, 10
xor r5, r5
%ifdef PIC
lea r11, [scan8_mem]
; ff_h264_idct_add8_mmx(uint8_t **dest, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add8_mmx, 5, 7, 0
+cglobal h264_idct_add8_8_mmx, 5, 7, 0
mov r5, 16
add r2, 512
%ifdef PIC
; ff_h264_idct_add8_mmx2(uint8_t **dest, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add8_mmx2, 5, 7, 0
+cglobal h264_idct_add8_8_mmx2, 5, 7, 0
mov r5, 16
add r2, 512
%ifdef ARCH_X86_64
; ff_h264_idct_add16_sse2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16_sse2, 5, 5, 8
+cglobal h264_idct_add16_8_sse2, 5, 5, 8
%ifdef ARCH_X86_64
mov r10, r0
%endif
; ff_h264_idct_add16intra_sse2(uint8_t *dst, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add16intra_sse2, 5, 7, 8
+cglobal h264_idct_add16intra_8_sse2, 5, 7, 8
%ifdef ARCH_X86_64
mov r10, r0
%endif
; ff_h264_idct_add8_sse2(uint8_t **dest, const int *block_offset,
; DCTELEM *block, int stride, const uint8_t nnzc[6*8])
-cglobal h264_idct_add8_sse2, 5, 7, 8
+cglobal h264_idct_add8_8_sse2, 5, 7, 8
add r2, 512
%ifdef ARCH_X86_64
mov r10, r0
#define IDCT_ADD_FUNC(NUM, DEPTH, OPT) \
void ff_h264_idct ## NUM ## _add_ ## DEPTH ## _ ## OPT (uint8_t *dst, int16_t *block, int stride);
+IDCT_ADD_FUNC(, 8, mmx)
IDCT_ADD_FUNC(, 10, sse2)
+IDCT_ADD_FUNC(_dc, 8, mmx2)
IDCT_ADD_FUNC(_dc, 10, mmx2)
+IDCT_ADD_FUNC(8_dc, 8, mmx2)
IDCT_ADD_FUNC(8_dc, 10, sse2)
+IDCT_ADD_FUNC(8, 8, mmx)
+IDCT_ADD_FUNC(8, 8, sse2)
IDCT_ADD_FUNC(8, 10, sse2)
#if HAVE_AVX
IDCT_ADD_FUNC(, 10, avx)
(uint8_t *dst, const int *block_offset, \
DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
+IDCT_ADD_REP_FUNC(8, 4, 8, mmx)
+IDCT_ADD_REP_FUNC(8, 4, 8, mmx2)
+IDCT_ADD_REP_FUNC(8, 4, 8, sse2)
IDCT_ADD_REP_FUNC(8, 4, 10, sse2)
IDCT_ADD_REP_FUNC(8, 4, 10, avx)
+IDCT_ADD_REP_FUNC(, 16, 8, mmx)
+IDCT_ADD_REP_FUNC(, 16, 8, mmx2)
+IDCT_ADD_REP_FUNC(, 16, 8, sse2)
IDCT_ADD_REP_FUNC(, 16, 10, sse2)
+IDCT_ADD_REP_FUNC(, 16intra, 8, mmx)
+IDCT_ADD_REP_FUNC(, 16intra, 8, mmx2)
+IDCT_ADD_REP_FUNC(, 16intra, 8, sse2)
IDCT_ADD_REP_FUNC(, 16intra, 10, sse2)
#if HAVE_AVX
IDCT_ADD_REP_FUNC(, 16, 10, avx)
void ff_h264_idct ## NUM ## _add ## REP ## _ ## DEPTH ## _ ## OPT \
(uint8_t **dst, const int *block_offset, \
DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
+IDCT_ADD_REP_FUNC2(, 8, 8, mmx)
+IDCT_ADD_REP_FUNC2(, 8, 8, mmx2)
+IDCT_ADD_REP_FUNC2(, 8, 8, sse2)
IDCT_ADD_REP_FUNC2(, 8, 10, sse2)
#if HAVE_AVX
IDCT_ADD_REP_FUNC2(, 8, 10, avx)
#endif
-void ff_h264_idct_add_mmx (uint8_t *dst, int16_t *block, int stride);
-void ff_h264_idct8_add_mmx (uint8_t *dst, int16_t *block, int stride);
-void ff_h264_idct8_add_sse2 (uint8_t *dst, int16_t *block, int stride);
-void ff_h264_idct_dc_add_mmx2 (uint8_t *dst, int16_t *block, int stride);
-void ff_h264_idct8_dc_add_mmx2(uint8_t *dst, int16_t *block, int stride);
-
-void ff_h264_idct_add16_mmx (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct8_add4_mmx (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add16_mmx2 (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add16intra_mmx (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add16intra_mmx2(uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct8_add4_mmx2 (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct8_add4_sse2 (uint8_t *dst, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add8_mmx (uint8_t **dest, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add8_mmx2 (uint8_t **dest, const int *block_offset,
- DCTELEM *block, int stride, const uint8_t nnzc[6*8]);
-
-void ff_h264_idct_add16_sse2 (uint8_t *dst, const int *block_offset, DCTELEM *block,
- int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add16intra_sse2(uint8_t *dst, const int *block_offset, DCTELEM *block,
- int stride, const uint8_t nnzc[6*8]);
-void ff_h264_idct_add8_sse2 (uint8_t **dest, const int *block_offset, DCTELEM *block,
- int stride, const uint8_t nnzc[6*8]);
void ff_h264_luma_dc_dequant_idct_mmx (DCTELEM *output, DCTELEM *input, int qmul);
void ff_h264_luma_dc_dequant_idct_sse2(DCTELEM *output, DCTELEM *input, int qmul);
}
#if HAVE_YASM
if (mm_flags & AV_CPU_FLAG_MMX) {
- c->h264_idct_dc_add=
- c->h264_idct_add= ff_h264_idct_add_mmx;
- c->h264_idct8_dc_add=
- c->h264_idct8_add= ff_h264_idct8_add_mmx;
-
- c->h264_idct_add16 = ff_h264_idct_add16_mmx;
- c->h264_idct8_add4 = ff_h264_idct8_add4_mmx;
- c->h264_idct_add8 = ff_h264_idct_add8_mmx;
- c->h264_idct_add16intra= ff_h264_idct_add16intra_mmx;
+ c->h264_idct_dc_add =
+ c->h264_idct_add = ff_h264_idct_add_8_mmx;
+ c->h264_idct8_dc_add =
+ c->h264_idct8_add = ff_h264_idct8_add_8_mmx;
+
+ c->h264_idct_add16 = ff_h264_idct_add16_8_mmx;
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx;
+ c->h264_idct_add8 = ff_h264_idct_add8_8_mmx;
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_8_mmx;
c->h264_luma_dc_dequant_idct= ff_h264_luma_dc_dequant_idct_mmx;
if (mm_flags & AV_CPU_FLAG_MMX2) {
- c->h264_idct_dc_add= ff_h264_idct_dc_add_mmx2;
- c->h264_idct8_dc_add= ff_h264_idct8_dc_add_mmx2;
- c->h264_idct_add16 = ff_h264_idct_add16_mmx2;
- c->h264_idct8_add4 = ff_h264_idct8_add4_mmx2;
- c->h264_idct_add8 = ff_h264_idct_add8_mmx2;
- c->h264_idct_add16intra= ff_h264_idct_add16intra_mmx2;
+ c->h264_idct_dc_add = ff_h264_idct_dc_add_8_mmx2;
+ c->h264_idct8_dc_add = ff_h264_idct8_dc_add_8_mmx2;
+ c->h264_idct_add16 = ff_h264_idct_add16_8_mmx2;
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_mmx2;
+ c->h264_idct_add8 = ff_h264_idct_add8_8_mmx2;
+ c->h264_idct_add16intra= ff_h264_idct_add16intra_8_mmx2;
c->h264_v_loop_filter_chroma= ff_deblock_v_chroma_8_mmxext;
c->h264_h_loop_filter_chroma= ff_deblock_h_chroma_8_mmxext;
c->biweight_h264_pixels_tab[7]= ff_h264_biweight_4x2_mmx2;
if (mm_flags&AV_CPU_FLAG_SSE2) {
- c->h264_idct8_add = ff_h264_idct8_add_sse2;
- c->h264_idct8_add4= ff_h264_idct8_add4_sse2;
+ c->h264_idct8_add = ff_h264_idct8_add_8_sse2;
+
+ c->h264_idct_add16 = ff_h264_idct_add16_8_sse2;
+ c->h264_idct8_add4 = ff_h264_idct8_add4_8_sse2;
+ c->h264_idct_add8 = ff_h264_idct_add8_8_sse2;
+ c->h264_idct_add16intra = ff_h264_idct_add16intra_8_sse2;
c->h264_luma_dc_dequant_idct= ff_h264_luma_dc_dequant_idct_sse2;
c->weight_h264_pixels_tab[0]= ff_h264_weight_16x16_sse2;
c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_sse2;
c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_sse2;
#endif
-
- c->h264_idct_add16 = ff_h264_idct_add16_sse2;
- c->h264_idct_add8 = ff_h264_idct_add8_sse2;
- c->h264_idct_add16intra = ff_h264_idct_add16intra_sse2;
}
if (mm_flags&AV_CPU_FLAG_SSSE3) {
c->biweight_h264_pixels_tab[0]= ff_h264_biweight_16x16_ssse3;