h264_loopfilter: port x86 simd to cpuflags.

libavcodec/x86/h264_deblock.asm

@@ -282,8 +282,8 @@ cextern pb_A1
 ;-----------------------------------------------------------------------------
 ; void deblock_v_luma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-%macro DEBLOCK_LUMA 1
-cglobal deblock_v_luma_8_%1, 5,5,10
+%macro DEBLOCK_LUMA 0
+cglobal deblock_v_luma_8, 5,5,10
     movd    m8, [r4] ; tc0
     lea     r4, [r1*3]
     dec     r2d        ; alpha-1
@@ -327,8 +327,8 @@ cglobal deblock_v_luma_8_%1, 5,5,10
 ;-----------------------------------------------------------------------------
 ; void deblock_h_luma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-INIT_MMX
-cglobal deblock_h_luma_8_%1, 5,9
+INIT_MMX cpuname
+cglobal deblock_h_luma_8, 5,9
     movsxd r7,  r1d
     lea    r8,  [r7+r7*2]
     lea    r6,  [r0-4]
@@ -355,7 +355,7 @@ cglobal deblock_h_luma_8_%1, 5,9
 %if WIN64
     mov    [rsp+0x20], r4
 %endif
-    call   deblock_v_luma_8_%1
+    call   deblock_v_luma_8
 
     ; transpose 16x4 -> original space  (only the middle 4 rows were changed by the filter)
     add    r6, 2
@@ -384,24 +384,24 @@ cglobal deblock_h_luma_8_%1, 5,9
     RET
 %endmacro
 
-INIT_XMM
-DEBLOCK_LUMA sse2
-INIT_AVX
-DEBLOCK_LUMA avx
+INIT_XMM sse2
+DEBLOCK_LUMA
+INIT_XMM avx
+DEBLOCK_LUMA
 
 %else
 
-%macro DEBLOCK_LUMA 3
+%macro DEBLOCK_LUMA 2
 ;-----------------------------------------------------------------------------
 ; void deblock_v8_luma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal deblock_%2_luma_8_%1, 5,5
+cglobal deblock_%1_luma_8, 5,5
     lea     r4, [r1*3]
     dec     r2     ; alpha-1
     neg     r4
     dec     r3     ; beta-1
     add     r4, r0 ; pix-3*stride
-    %assign pad 2*%3+12-(stack_offset&15)
+    %assign pad 2*%2+12-(stack_offset&15)
     SUB     esp, pad
 
     mova    m0, [r4+r1]   ; p1
@@ -415,7 +415,7 @@ cglobal deblock_%2_luma_8_%1, 5,5
     movd    m4, [r3] ; tc0
     punpcklbw m4, m4
     punpcklbw m4, m4 ; tc = 4x tc0[3], 4x tc0[2], 4x tc0[1], 4x tc0[0]
-    mova   [esp+%3], m4 ; tc
+    mova   [esp+%2], m4 ; tc
     pcmpgtb m4, m3
     mova    m3, [r4] ; p2
     pand    m4, m7
@@ -423,7 +423,7 @@ cglobal deblock_%2_luma_8_%1, 5,5
 
     DIFF_GT2 m1, m3, m5, m6, m7 ; |p2-p0| > beta-1
     pand    m6, m4
-    pand    m4, [esp+%3] ; tc
+    pand    m4, [esp+%2] ; tc
     psubb   m7, m4, m6
     pand    m6, m4
     LUMA_Q1 m0, m3, [r4], [r4+r1], m6, m4
@@ -431,7 +431,7 @@ cglobal deblock_%2_luma_8_%1, 5,5
     mova    m4, [r0+2*r1] ; q2
     DIFF_GT2 m2, m4, m5, m6, m3 ; |q2-q0| > beta-1
     pand    m6, [esp] ; mask
-    mova    m5, [esp+%3] ; tc
+    mova    m5, [esp+%2] ; tc
     psubb   m7, m6
     pand    m5, m6
     mova    m3, [r0+r1]
@@ -446,8 +446,8 @@ cglobal deblock_%2_luma_8_%1, 5,5
 ;-----------------------------------------------------------------------------
 ; void deblock_h_luma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-INIT_MMX
-cglobal deblock_h_luma_8_%1, 0,5
+INIT_MMX cpuname
+cglobal deblock_h_luma_8, 0,5
     mov    r0, r0mp
     mov    r3, r1m
     lea    r4, [r3*3]
@@ -470,11 +470,11 @@ cglobal deblock_h_luma_8_%1, 0,5
     PUSH   dword r2m
     PUSH   dword 16
     PUSH   dword r0
-    call   deblock_%2_luma_8_%1
-%ifidn %2, v8
+    call   deblock_%1_luma_8
+%ifidn %1, v8
     add    dword [esp   ], 8 ; pix_tmp+0x38
     add    dword [esp+16], 2 ; tc0+2
-    call   deblock_%2_luma_8_%1
+    call   deblock_%1_luma_8
 %endif
     ADD    esp, 20
 
@@ -501,12 +501,12 @@ cglobal deblock_h_luma_8_%1, 0,5
     RET
 %endmacro ; DEBLOCK_LUMA
 
-INIT_MMX
-DEBLOCK_LUMA mmxext, v8, 8
-INIT_XMM
-DEBLOCK_LUMA sse2, v, 16
-INIT_AVX
-DEBLOCK_LUMA avx, v, 16
+INIT_MMX mmx2
+DEBLOCK_LUMA v8, 8
+INIT_XMM sse2
+DEBLOCK_LUMA v, 16
+INIT_XMM avx
+DEBLOCK_LUMA v, 16
 
 %endif ; ARCH
 
@@ -608,7 +608,7 @@ DEBLOCK_LUMA avx, v, 16
     %define mask1p mask1q
 %endmacro
 
-%macro DEBLOCK_LUMA_INTRA 2
+%macro DEBLOCK_LUMA_INTRA 1
     %define p1 m0
     %define p0 m1
     %define q0 m2
@@ -643,7 +643,7 @@ DEBLOCK_LUMA avx, v, 16
 ;-----------------------------------------------------------------------------
 ; void deblock_v_luma_intra( uint8_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_%2_luma_intra_8_%1, 4,6,16
+cglobal deblock_%1_luma_intra_8, 4,6,16
 %if ARCH_X86_64 == 0
     sub     esp, 0x60
 %endif
@@ -700,12 +700,12 @@ cglobal deblock_%2_luma_intra_8_%1, 4,6,16
 %endif
     RET
 
-INIT_MMX
+INIT_MMX cpuname
 %if ARCH_X86_64
 ;-----------------------------------------------------------------------------
 ; void deblock_h_luma_intra( uint8_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_h_luma_intra_8_%1, 4,9
+cglobal deblock_h_luma_intra_8, 4,9
     movsxd r7,  r1d
     lea    r8,  [r7*3]
     lea    r6,  [r0-4]
@@ -721,7 +721,7 @@ cglobal deblock_h_luma_intra_8_%1, 4,9
 
     lea    r0,  [pix_tmp+0x40]
     mov    r1,  0x10
-    call   deblock_v_luma_intra_8_%1
+    call   deblock_v_luma_intra_8
 
     ; transpose 16x6 -> original space (but we can't write only 6 pixels, so really 16x8)
     lea    r5, [r6+r8]
@@ -734,7 +734,7 @@ cglobal deblock_h_luma_intra_8_%1, 4,9
     add    rsp, 0x88
     RET
 %else
-cglobal deblock_h_luma_intra_8_%1, 2,4
+cglobal deblock_h_luma_intra_8, 2,4
     lea    r3,  [r1*3]
     sub    r0,  4
     lea    r2,  [r0+r3]
@@ -753,10 +753,10 @@ cglobal deblock_h_luma_intra_8_%1, 2,4
     PUSH   dword r2m
     PUSH   dword 16
     PUSH   r0
-    call   deblock_%2_luma_intra_8_%1
-%ifidn %2, v8
+    call   deblock_%1_luma_intra_8
+%ifidn %1, v8
     add    dword [rsp], 8 ; pix_tmp+8
-    call   deblock_%2_luma_intra_8_%1
+    call   deblock_%1_luma_intra_8
 %endif
     ADD    esp, 16
 
@@ -775,16 +775,16 @@ cglobal deblock_h_luma_intra_8_%1, 2,4
 %endif ; ARCH_X86_64
 %endmacro ; DEBLOCK_LUMA_INTRA
 
-INIT_XMM
-DEBLOCK_LUMA_INTRA sse2, v
-INIT_AVX
-DEBLOCK_LUMA_INTRA avx , v
+INIT_XMM sse2
+DEBLOCK_LUMA_INTRA v
+INIT_XMM avx
+DEBLOCK_LUMA_INTRA v
 %if ARCH_X86_64 == 0
-INIT_MMX
-DEBLOCK_LUMA_INTRA mmxext, v8
+INIT_MMX mmx2
+DEBLOCK_LUMA_INTRA v8
 %endif
 
-INIT_MMX
+INIT_MMX mmx2
 
 %macro CHROMA_V_START 0
     dec    r2d      ; alpha-1
@@ -809,13 +809,13 @@ INIT_MMX
 ;-----------------------------------------------------------------------------
 ; void ff_deblock_v_chroma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_chroma_8_mmxext, 5,6
+cglobal deblock_v_chroma_8, 5,6
     CHROMA_V_START
     movq  m0, [t5]
     movq  m1, [t5+r1]
     movq  m2, [r0]
     movq  m3, [r0+r1]
-    call ff_chroma_inter_body_mmxext
+    call ff_chroma_inter_body_mmx2
     movq  [t5+r1], m1
     movq  [r0], m2
     RET
@@ -823,7 +823,7 @@ cglobal deblock_v_chroma_8_mmxext, 5,6
 ;-----------------------------------------------------------------------------
 ; void ff_deblock_h_chroma( uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal deblock_h_chroma_8_mmxext, 5,7
+cglobal deblock_h_chroma_8, 5,7
 %if UNIX64
     %define buf0 [rsp-24]
     %define buf1 [rsp-16]
@@ -839,7 +839,7 @@ cglobal deblock_h_chroma_8_mmxext, 5,7
     TRANSPOSE4x8_LOAD  bw, wd, dq, PASS8ROWS(t5, r0, r1, t6)
     movq  buf0, m0
     movq  buf1, m3
-    call ff_chroma_inter_body_mmxext
+    call ff_chroma_inter_body_mmx2
     movq  m0, buf0
     movq  m3, buf1
     TRANSPOSE8x4B_STORE PASS8ROWS(t5, r0, r1, t6)
@@ -849,7 +849,7 @@ cglobal deblock_h_chroma_8_mmxext, 5,7
     RET
 
 ALIGN 16
-ff_chroma_inter_body_mmxext:
+ff_chroma_inter_body_mmx2:
     LOAD_MASK  r2d, r3d
     movd       m6, [r4] ; tc0
     punpcklbw  m6, m6
@@ -876,13 +876,13 @@ ff_chroma_inter_body_mmxext:
 ;-----------------------------------------------------------------------------
 ; void ff_deblock_v_chroma_intra( uint8_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_chroma_intra_8_mmxext, 4,5
+cglobal deblock_v_chroma_intra_8, 4,5
     CHROMA_V_START
     movq  m0, [t5]
     movq  m1, [t5+r1]
     movq  m2, [r0]
     movq  m3, [r0+r1]
-    call ff_chroma_intra_body_mmxext
+    call ff_chroma_intra_body_mmx2
     movq  [t5+r1], m1
     movq  [r0], m2
     RET
@@ -890,15 +890,15 @@ cglobal deblock_v_chroma_intra_8_mmxext, 4,5
 ;-----------------------------------------------------------------------------
 ; void ff_deblock_h_chroma_intra( uint8_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_h_chroma_intra_8_mmxext, 4,6
+cglobal deblock_h_chroma_intra_8, 4,6
     CHROMA_H_START
     TRANSPOSE4x8_LOAD  bw, wd, dq, PASS8ROWS(t5, r0, r1, t6)
-    call ff_chroma_intra_body_mmxext
+    call ff_chroma_intra_body_mmx2
     TRANSPOSE8x4B_STORE PASS8ROWS(t5, r0, r1, t6)
     RET
 
 ALIGN 16
-ff_chroma_intra_body_mmxext:
+ff_chroma_intra_body_mmx2:
     LOAD_MASK r2d, r3d
     movq   m5, m1
     movq   m6, m2

libavcodec/x86/h264_deblock_10bit.asm

@@ -151,11 +151,11 @@ cextern pw_4
 %endif
 %endmacro
 
-%macro DEBLOCK_LUMA 1
+%macro DEBLOCK_LUMA 0
 ;-----------------------------------------------------------------------------
 ; void deblock_v_luma( uint16_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_luma_10_%1, 5,5,8*(mmsize/16)
+cglobal deblock_v_luma_10, 5,5,8*(mmsize/16)
     %assign pad 5*mmsize+12-(stack_offset&15)
     %define tcm [rsp]
     %define ms1 [rsp+mmsize]
@@ -210,7 +210,7 @@ cglobal deblock_v_luma_10_%1, 5,5,8*(mmsize/16)
     ADD         rsp, pad
     RET
 
-cglobal deblock_h_luma_10_%1, 5,6,8*(mmsize/16)
+cglobal deblock_h_luma_10, 5,6,8*(mmsize/16)
     %assign pad 7*mmsize+12-(stack_offset&15)
     %define tcm [rsp]
     %define ms1 [rsp+mmsize]
@@ -301,7 +301,6 @@ cglobal deblock_h_luma_10_%1, 5,6,8*(mmsize/16)
     RET
 %endmacro
 
-INIT_XMM
 %if ARCH_X86_64
 ; in:  m0=p1, m1=p0, m2=q0, m3=q1, m8=p2, m9=q2
 ;      m12=alpha, m13=beta
@@ -339,8 +338,8 @@ INIT_XMM
     SWAP         3, 9
 %endmacro
 
-%macro DEBLOCK_LUMA_64 1
-cglobal deblock_v_luma_10_%1, 5,5,15
+%macro DEBLOCK_LUMA_64 0
+cglobal deblock_v_luma_10, 5,5,15
     %define p2 m8
     %define p1 m0
     %define p0 m1
@@ -377,7 +376,7 @@ cglobal deblock_v_luma_10_%1, 5,5,15
     jg .loop
     REP_RET
 
-cglobal deblock_h_luma_10_%1, 5,7,15
+cglobal deblock_h_luma_10, 5,7,15
     shl        r2d, 2
     shl        r3d, 2
     LOAD_AB    m12, m13, r2, r3
@@ -417,10 +416,10 @@ cglobal deblock_h_luma_10_%1, 5,7,15
     REP_RET
 %endmacro
 
-INIT_XMM
-DEBLOCK_LUMA_64 sse2
-INIT_AVX
-DEBLOCK_LUMA_64 avx
+INIT_XMM sse2
+DEBLOCK_LUMA_64
+INIT_XMM avx
+DEBLOCK_LUMA_64
 %endif
 
 %macro SWAPMOVA 2
@@ -602,8 +601,8 @@ DEBLOCK_LUMA_64 avx
 ;-----------------------------------------------------------------------------
 ; void deblock_v_luma_intra( uint16_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-%macro DEBLOCK_LUMA_INTRA_64 1
-cglobal deblock_v_luma_intra_10_%1, 4,7,16
+%macro DEBLOCK_LUMA_INTRA_64 0
+cglobal deblock_v_luma_intra_10, 4,7,16
     %define t0 m1
     %define t1 m2
     %define t2 m4
@@ -653,7 +652,7 @@ cglobal deblock_v_luma_intra_10_%1, 4,7,16
 ;-----------------------------------------------------------------------------
 ; void deblock_h_luma_intra( uint16_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_h_luma_intra_10_%1, 4,7,16
+cglobal deblock_h_luma_intra_10, 4,7,16
     %define t0 m15
     %define t1 m14
     %define t2 m2
@@ -712,18 +711,18 @@ cglobal deblock_h_luma_intra_10_%1, 4,7,16
     RET
 %endmacro
 
-INIT_XMM
-DEBLOCK_LUMA_INTRA_64 sse2
-INIT_AVX
-DEBLOCK_LUMA_INTRA_64 avx
+INIT_XMM sse2
+DEBLOCK_LUMA_INTRA_64
+INIT_XMM avx
+DEBLOCK_LUMA_INTRA_64
 
 %endif
 
-%macro DEBLOCK_LUMA_INTRA 1
+%macro DEBLOCK_LUMA_INTRA 0
 ;-----------------------------------------------------------------------------
 ; void deblock_v_luma_intra( uint16_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_luma_intra_10_%1, 4,7,8*(mmsize/16)
+cglobal deblock_v_luma_intra_10, 4,7,8*(mmsize/16)
     LUMA_INTRA_INIT 3
     lea     r4, [r1*4]
     lea     r5, [r1*3]
@@ -751,7 +750,7 @@ cglobal deblock_v_luma_intra_10_%1, 4,7,8*(mmsize/16)
 ;-----------------------------------------------------------------------------
 ; void deblock_h_luma_intra( uint16_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_h_luma_intra_10_%1, 4,7,8*(mmsize/16)
+cglobal deblock_h_luma_intra_10, 4,7,8*(mmsize/16)
     LUMA_INTRA_INIT 8
 %if mmsize == 8
     lea     r4, [r1*3]
@@ -793,15 +792,15 @@ cglobal deblock_h_luma_intra_10_%1, 4,7,8*(mmsize/16)
 %endmacro
 
 %if ARCH_X86_64 == 0
-INIT_MMX
-DEBLOCK_LUMA mmxext
-DEBLOCK_LUMA_INTRA mmxext
-INIT_XMM
-DEBLOCK_LUMA sse2
-DEBLOCK_LUMA_INTRA sse2
-INIT_AVX
-DEBLOCK_LUMA avx
-DEBLOCK_LUMA_INTRA avx
+INIT_MMX mmx2
+DEBLOCK_LUMA
+DEBLOCK_LUMA_INTRA
+INIT_XMM sse2
+DEBLOCK_LUMA
+DEBLOCK_LUMA_INTRA
+INIT_XMM avx
+DEBLOCK_LUMA
+DEBLOCK_LUMA_INTRA
 %endif
 
 ; in: %1=p0, %2=q0, %3=p1, %4=q1, %5=mask, %6=tmp, %7=tmp
@@ -843,11 +842,11 @@ DEBLOCK_LUMA_INTRA avx
     psraw       %1, 6
 %endmacro
 
-%macro DEBLOCK_CHROMA 1
+%macro DEBLOCK_CHROMA 0
 ;-----------------------------------------------------------------------------
 ; void deblock_v_chroma( uint16_t *pix, int stride, int alpha, int beta, int8_t *tc0 )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_chroma_10_%1, 5,7-(mmsize/16),8*(mmsize/16)
+cglobal deblock_v_chroma_10, 5,7-(mmsize/16),8*(mmsize/16)
     mov         r5, r0
     sub         r0, r1
     sub         r0, r1
@@ -881,7 +880,7 @@ cglobal deblock_v_chroma_10_%1, 5,7-(mmsize/16),8*(mmsize/16)
 ;-----------------------------------------------------------------------------
 ; void deblock_v_chroma_intra( uint16_t *pix, int stride, int alpha, int beta )
 ;-----------------------------------------------------------------------------
-cglobal deblock_v_chroma_intra_10_%1, 4,6-(mmsize/16),8*(mmsize/16)
+cglobal deblock_v_chroma_intra_10, 4,6-(mmsize/16),8*(mmsize/16)
     mov         r4, r0
     sub         r0, r1
     sub         r0, r1
@@ -908,10 +907,10 @@ cglobal deblock_v_chroma_intra_10_%1, 4,6-(mmsize/16),8*(mmsize/16)
 %endmacro
 
 %if ARCH_X86_64 == 0
-INIT_MMX
-DEBLOCK_CHROMA mmxext
+INIT_MMX mmx2
+DEBLOCK_CHROMA
 %endif
-INIT_XMM
-DEBLOCK_CHROMA sse2
-INIT_AVX
-DEBLOCK_CHROMA avx
+INIT_XMM sse2
+DEBLOCK_CHROMA
+INIT_XMM avx
+DEBLOCK_CHROMA

libavcodec/x86/h264dsp_mmx.c

@@ -249,12 +249,12 @@ void ff_deblock_ ## DIR ## _ ## TYPE ## _ ## DEPTH ## _ ## OPT (uint8_t *pix, in
                                                                 int alpha, int beta);
 
 #define LF_FUNCS(type, depth)\
-LF_FUNC (h,  chroma,       depth, mmxext)\
-LF_IFUNC(h,  chroma_intra, depth, mmxext)\
-LF_FUNC (v,  chroma,       depth, mmxext)\
-LF_IFUNC(v,  chroma_intra, depth, mmxext)\
-LF_FUNC (h,  luma,         depth, mmxext)\
-LF_IFUNC(h,  luma_intra,   depth, mmxext)\
+LF_FUNC (h,  chroma,       depth, mmx2)\
+LF_IFUNC(h,  chroma_intra, depth, mmx2)\
+LF_FUNC (v,  chroma,       depth, mmx2)\
+LF_IFUNC(v,  chroma_intra, depth, mmx2)\
+LF_FUNC (h,  luma,         depth, mmx2)\
+LF_IFUNC(h,  luma_intra,   depth, mmx2)\
 LF_FUNC (h,  luma,         depth, sse2)\
 LF_IFUNC(h,  luma_intra,   depth, sse2)\
 LF_FUNC (v,  luma,         depth, sse2)\
@@ -276,24 +276,24 @@ LF_FUNCS( uint8_t,  8)
 LF_FUNCS(uint16_t, 10)
 
 #if ARCH_X86_32
-LF_FUNC (v8, luma,             8, mmxext)
-static void ff_deblock_v_luma_8_mmxext(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
+LF_FUNC (v8, luma,             8, mmx2)
+static void ff_deblock_v_luma_8_mmx2(uint8_t *pix, int stride, int alpha, int beta, int8_t *tc0)
 {
     if((tc0[0] & tc0[1]) >= 0)
-        ff_deblock_v8_luma_8_mmxext(pix+0, stride, alpha, beta, tc0);
+        ff_deblock_v8_luma_8_mmx2(pix+0, stride, alpha, beta, tc0);
     if((tc0[2] & tc0[3]) >= 0)
-        ff_deblock_v8_luma_8_mmxext(pix+8, stride, alpha, beta, tc0+2);
+        ff_deblock_v8_luma_8_mmx2(pix+8, stride, alpha, beta, tc0+2);
 }
-LF_IFUNC(v8, luma_intra,        8, mmxext)
-static void ff_deblock_v_luma_intra_8_mmxext(uint8_t *pix, int stride, int alpha, int beta)
+LF_IFUNC(v8, luma_intra,        8, mmx2)
+static void ff_deblock_v_luma_intra_8_mmx2(uint8_t *pix, int stride, int alpha, int beta)
 {
-    ff_deblock_v8_luma_intra_8_mmxext(pix+0, stride, alpha, beta);
-    ff_deblock_v8_luma_intra_8_mmxext(pix+8, stride, alpha, beta);
+    ff_deblock_v8_luma_intra_8_mmx2(pix+0, stride, alpha, beta);
+    ff_deblock_v8_luma_intra_8_mmx2(pix+8, stride, alpha, beta);
 }
 #endif /* ARCH_X86_32 */
 
-LF_FUNC (v,  luma,            10, mmxext)
-LF_IFUNC(v,  luma_intra,      10, mmxext)
+LF_FUNC (v,  luma,            10, mmx2)
+LF_IFUNC(v,  luma_intra,      10, mmx2)
 
 /***********************************/
 /* weighted prediction */
@@ -373,17 +373,17 @@ void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth, const int chrom
                 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_v_loop_filter_chroma_intra= ff_deblock_v_chroma_intra_8_mmxext;
+            c->h264_v_loop_filter_chroma= ff_deblock_v_chroma_8_mmx2;
+            c->h264_v_loop_filter_chroma_intra= ff_deblock_v_chroma_intra_8_mmx2;
             if (chroma_format_idc == 1) {
-                c->h264_h_loop_filter_chroma= ff_deblock_h_chroma_8_mmxext;
-                c->h264_h_loop_filter_chroma_intra= ff_deblock_h_chroma_intra_8_mmxext;
+                c->h264_h_loop_filter_chroma= ff_deblock_h_chroma_8_mmx2;
+                c->h264_h_loop_filter_chroma_intra= ff_deblock_h_chroma_intra_8_mmx2;
             }
 #if ARCH_X86_32
-            c->h264_v_loop_filter_luma= ff_deblock_v_luma_8_mmxext;
-            c->h264_h_loop_filter_luma= ff_deblock_h_luma_8_mmxext;
-            c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_mmxext;
-            c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmxext;
+            c->h264_v_loop_filter_luma= ff_deblock_v_luma_8_mmx2;
+            c->h264_h_loop_filter_luma= ff_deblock_h_luma_8_mmx2;
+            c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_8_mmx2;
+            c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_8_mmx2;
 #endif
             c->weight_h264_pixels_tab[0]= ff_h264_weight_16_mmx2;
             c->weight_h264_pixels_tab[1]= ff_h264_weight_8_mmx2;
@@ -436,12 +436,12 @@ void ff_h264dsp_init_x86(H264DSPContext *c, const int bit_depth, const int chrom
     if (mm_flags & AV_CPU_FLAG_MMX) {
         if (mm_flags & AV_CPU_FLAG_MMX2) {
 #if ARCH_X86_32
-            c->h264_v_loop_filter_chroma= ff_deblock_v_chroma_10_mmxext;
-            c->h264_v_loop_filter_chroma_intra= ff_deblock_v_chroma_intra_10_mmxext;
-            c->h264_v_loop_filter_luma= ff_deblock_v_luma_10_mmxext;
-            c->h264_h_loop_filter_luma= ff_deblock_h_luma_10_mmxext;
-            c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_mmxext;
-            c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_mmxext;
+            c->h264_v_loop_filter_chroma= ff_deblock_v_chroma_10_mmx2;
+            c->h264_v_loop_filter_chroma_intra= ff_deblock_v_chroma_intra_10_mmx2;
+            c->h264_v_loop_filter_luma= ff_deblock_v_luma_10_mmx2;
+            c->h264_h_loop_filter_luma= ff_deblock_h_luma_10_mmx2;
+            c->h264_v_loop_filter_luma_intra = ff_deblock_v_luma_intra_10_mmx2;
+            c->h264_h_loop_filter_luma_intra = ff_deblock_h_luma_intra_10_mmx2;
 #endif
             c->h264_idct_dc_add= ff_h264_idct_dc_add_10_mmx2;
             if (mm_flags&AV_CPU_FLAG_SSE2) {