Merge remote branch 'internal/upstream' into HEAD

Conflicts:
	vp8/encoder/encodeframe.c
	vp8/encoder/rdopt.c

Change-Id: I183fd3ce9e94617ec888c9f891055b9f1f8ca6c5

7 files changed, 23 insertions, 336 deletions

diff --git a/vp8/encoder/x86/dct_x86.h b/vp8/encoder/x86/dct_x86.h
index 59a5cb1d7..19f6c1686 100644
--- a/vp8/encoder/x86/dct_x86.h
+++ b/vp8/encoder/x86/dct_x86.h
@@ -31,6 +31,12 @@ extern prototype_fdct(vp8_short_fdct8x4_mmx);
 #undef  vp8_fdct_short8x4
 #define vp8_fdct_short8x4 vp8_short_fdct8x4_mmx
 
+#undef  vp8_fdct_fast4x4
+#define vp8_fdct_fast4x4 vp8_short_fdct4x4_mmx
+
+#undef  vp8_fdct_fast8x4
+#define vp8_fdct_fast8x4 vp8_short_fdct8x4_mmx
+
 #endif
 
 #endif
diff --git a/vp8/encoder/x86/variance_impl_mmx.asm b/vp8/encoder/x86/variance_impl_mmx.asm
index 67a9b4d3e..13b76ea91 100644
--- a/vp8/encoder/x86/variance_impl_mmx.asm
+++ b/vp8/encoder/x86/variance_impl_mmx.asm
@@ -843,136 +843,6 @@ filter_block2d_bil_var_mmx_loop:
     pop         rbp
     ret
 
-;unsigned int vp8_get16x16pred_error_mmx
-;(
-;    unsigned char *src_ptr,
-;    int src_stride,
-;    unsigned char *ref_ptr,
-;    int ref_stride
-;)
-global sym(vp8_get16x16pred_error_mmx)
-sym(vp8_get16x16pred_error_mmx):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 4
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    sub         rsp, 16
-    ; end prolog
-
-        mov         rsi,            arg(0) ;DWORD PTR [src_ptr]
-        mov         rdi,            arg(2) ;DWORD PTR [ref_ptr]
-
-        movsxd      rax,            DWORD PTR arg(1) ;[src_stride]
-        movsxd      rdx,            DWORD PTR arg(3) ;[ref_stride]
-
-        pxor        mm0,            mm0                     ; clear xmm0 for unpack
-        pxor        mm7,            mm7                     ; clear xmm7 for accumulating diffs
-
-        pxor        mm6,            mm6                     ; clear xmm6 for accumulating sse
-        mov         rcx,            16
-
-var16loop:
-
-        movq        mm1,            [rsi]
-        movq        mm2,            [rdi]
-
-        movq        mm3,            mm1
-        movq        mm4,            mm2
-
-        punpcklbw   mm1,            mm0
-        punpckhbw   mm3,            mm0
-
-        punpcklbw   mm2,            mm0
-        punpckhbw   mm4,            mm0
-
-        psubw       mm1,            mm2
-        psubw       mm3,            mm4
-
-        paddw       mm7,            mm1
-        pmaddwd     mm1,            mm1
-
-        paddw       mm7,            mm3
-        pmaddwd     mm3,            mm3
-
-        paddd       mm6,            mm1
-        paddd       mm6,            mm3
-
-
-        movq        mm1,            [rsi+8]
-        movq        mm2,            [rdi+8]
-
-        movq        mm3,            mm1
-        movq        mm4,            mm2
-
-        punpcklbw   mm1,            mm0
-        punpckhbw   mm3,            mm0
-
-        punpcklbw   mm2,            mm0
-        punpckhbw   mm4,            mm0
-
-        psubw       mm1,            mm2
-        psubw       mm3,            mm4
-
-        paddw       mm7,            mm1
-        pmaddwd     mm1,            mm1
-
-        paddw       mm7,            mm3
-        pmaddwd     mm3,            mm3
-
-        paddd       mm6,            mm1
-        paddd       mm6,            mm3
-
-        add         rsi,            rax
-        add         rdi,            rdx
-
-        sub         rcx,            1
-        jnz         var16loop
-
-
-        movq        mm1,            mm6
-        pxor        mm6,            mm6
-
-        pxor        mm5,            mm5
-        punpcklwd   mm6,            mm7
-
-        punpckhwd   mm5,            mm7
-        psrad       mm5,            16
-
-        psrad       mm6,            16
-        paddd       mm6,            mm5
-
-        movq        mm2,            mm1
-        psrlq       mm1,            32
-
-        paddd       mm2,            mm1
-        movq        mm7,            mm6
-
-        psrlq       mm6,            32
-        paddd       mm6,            mm7
-
-        movd DWORD PTR [rsp],       mm6  ;Sum
-        movd DWORD PTR [rsp+4],     mm2  ;SSE
-
-        ; return (SSE-((Sum*Sum)>>8));
-        movsxd      rdx, dword ptr [rsp]
-        imul        rdx, rdx
-        sar         rdx, 8
-        movsxd      rax, dword ptr [rsp + 4]
-        sub         rax, rdx
-
-
-    ; begin epilog
-    add rsp, 16
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
-
 
 SECTION_RODATA
 ;short mmx_bi_rd[4] = { 64, 64, 64, 64};
diff --git a/vp8/encoder/x86/variance_impl_sse2.asm b/vp8/encoder/x86/variance_impl_sse2.asm
index 5becc7344..b7a6b3286 100644
--- a/vp8/encoder/x86/variance_impl_sse2.asm
+++ b/vp8/encoder/x86/variance_impl_sse2.asm
@@ -213,122 +213,6 @@ var16loop:
     ret
 
 
-;unsigned int vp8_get16x16pred_error_sse2
-;(
-;   unsigned char *src_ptr,
-;    int src_stride,
-;    unsigned char *ref_ptr,
-;    int ref_stride
-;)
-global sym(vp8_get16x16pred_error_sse2)
-sym(vp8_get16x16pred_error_sse2):
-    push        rbp
-    mov         rbp, rsp
-    SHADOW_ARGS_TO_STACK 4
-    SAVE_XMM 7
-    GET_GOT     rbx
-    push rsi
-    push rdi
-    sub         rsp, 16
-    ; end prolog
-
-        mov         rsi,            arg(0) ;[src_ptr]
-        mov         rdi,            arg(2) ;[ref_ptr]
-
-        movsxd      rax,            DWORD PTR arg(1) ;[src_stride]
-        movsxd      rdx,            DWORD PTR arg(3) ;[ref_stride]
-
-        pxor        xmm0,           xmm0                        ; clear xmm0 for unpack
-        pxor        xmm7,           xmm7                        ; clear xmm7 for accumulating diffs
-
-        pxor        xmm6,           xmm6                        ; clear xmm6 for accumulating sse
-        mov         rcx,            16
-
-var16peloop:
-        movdqu      xmm1,           XMMWORD PTR [rsi]
-        movdqu      xmm2,           XMMWORD PTR [rdi]
-
-        movdqa      xmm3,           xmm1
-        movdqa      xmm4,           xmm2
-
-        punpcklbw   xmm1,           xmm0
-        punpckhbw   xmm3,           xmm0
-
-        punpcklbw   xmm2,           xmm0
-        punpckhbw   xmm4,           xmm0
-
-        psubw       xmm1,           xmm2
-        psubw       xmm3,           xmm4
-
-        paddw       xmm7,           xmm1
-        pmaddwd     xmm1,           xmm1
-
-        paddw       xmm7,           xmm3
-        pmaddwd     xmm3,           xmm3
-
-        paddd       xmm6,           xmm1
-        paddd       xmm6,           xmm3
-
-        add         rsi,            rax
-        add         rdi,            rdx
-
-        sub         rcx,            1
-        jnz         var16peloop
-
-
-        movdqa      xmm1,           xmm6
-        pxor        xmm6,           xmm6
-
-        pxor        xmm5,           xmm5
-        punpcklwd   xmm6,           xmm7
-
-        punpckhwd   xmm5,           xmm7
-        psrad       xmm5,           16
-
-        psrad       xmm6,           16
-        paddd       xmm6,           xmm5
-
-        movdqa      xmm2,           xmm1
-        punpckldq   xmm1,           xmm0
-
-        punpckhdq   xmm2,           xmm0
-        movdqa      xmm7,           xmm6
-
-        paddd       xmm1,           xmm2
-        punpckldq   xmm6,           xmm0
-
-        punpckhdq   xmm7,           xmm0
-        paddd       xmm6,           xmm7
-
-        movdqa      xmm2,           xmm1
-        movdqa      xmm7,           xmm6
-
-        psrldq      xmm1,           8
-        psrldq      xmm6,           8
-
-        paddd       xmm7,           xmm6
-        paddd       xmm1,           xmm2
-
-        movd DWORD PTR [rsp],       xmm7  ;Sum
-        movd DWORD PTR [rsp+4],     xmm1  ;SSE
-
-        ; return (SSE-((Sum*Sum)>>8));
-        movsxd      rdx, dword ptr [rsp]
-        imul        rdx, rdx
-        sar         rdx, 8
-        movsxd      rax, dword ptr [rsp + 4]
-        sub         rax, rdx
-
-    ; begin epilog
-    add rsp, 16
-    pop rdi
-    pop rsi
-    RESTORE_GOT
-    RESTORE_XMM
-    UNSHADOW_ARGS
-    pop         rbp
-    ret
-
 
 
 ;unsigned int vp8_get8x8var_sse2
diff --git a/vp8/encoder/x86/variance_mmx.c b/vp8/encoder/x86/variance_mmx.c
index 4a89868c2..92b695f17 100644
--- a/vp8/encoder/x86/variance_mmx.c
+++ b/vp8/encoder/x86/variance_mmx.c
@@ -76,43 +76,6 @@ extern void vp8_filter_block2d_bil_var_mmx
     int *sum,
     unsigned int *sumsquared
 );
-extern unsigned int vp8_get16x16pred_error_mmx
-(
-    const unsigned char *src_ptr,
-    int src_stride,
-    const unsigned char *ref_ptr,
-    int ref_stride
-);
-
-unsigned int vp8_get16x16var_mmx(
-    const unsigned char *src_ptr,
-    int  source_stride,
-    const unsigned char *ref_ptr,
-    int  recon_stride,
-    unsigned int *SSE,
-    int *SUM
-)
-{
-    unsigned int sse0, sse1, sse2, sse3, var;
-    int sum0, sum1, sum2, sum3, avg;
-
-
-    vp8_get8x8var_mmx(src_ptr, source_stride, ref_ptr, recon_stride, &sse0, &sum0) ;
-    vp8_get8x8var_mmx(src_ptr + 8, source_stride, ref_ptr + 8, recon_stride, &sse1, &sum1);
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride, source_stride, ref_ptr + 8 * recon_stride, recon_stride, &sse2, &sum2) ;
-    vp8_get8x8var_mmx(src_ptr + 8 * source_stride + 8, source_stride, ref_ptr + 8 * recon_stride + 8, recon_stride, &sse3, &sum3);
-
-    var = sse0 + sse1 + sse2 + sse3;
-    avg = sum0 + sum1 + sum2 + sum3;
-
-    *SSE = var;
-    *SUM = avg;
-    return (var - ((avg * avg) >> 8));
-
-}
-
-
-
 
 
 unsigned int vp8_variance4x4_mmx(
diff --git a/vp8/encoder/x86/variance_sse2.c b/vp8/encoder/x86/variance_sse2.c
index dfc0915b9..24062eb9b 100644
--- a/vp8/encoder/x86/variance_sse2.c
+++ b/vp8/encoder/x86/variance_sse2.c
@@ -53,13 +53,6 @@ unsigned int vp8_get16x16var_sse2
     unsigned int *SSE,
     int *Sum
 );
-unsigned int vp8_get16x16pred_error_sse2
-(
-    const unsigned char *src_ptr,
-    int src_stride,
-    const unsigned char *ref_ptr,
-    int ref_stride
-);
 unsigned int vp8_get8x8var_sse2
 (
     const unsigned char *src_ptr,
diff --git a/vp8/encoder/x86/variance_x86.h b/vp8/encoder/x86/variance_x86.h
index 77e05e1e8..0ee8eb7e5 100644
--- a/vp8/encoder/x86/variance_x86.h
+++ b/vp8/encoder/x86/variance_x86.h
@@ -41,9 +41,7 @@ extern prototype_variance(vp8_variance_halfpixvar16x16_hv_mmx);
 extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_mmx);
 extern prototype_getmbss(vp8_get_mb_ss_mmx);
 extern prototype_variance(vp8_mse16x16_mmx);
-extern prototype_get16x16prederror(vp8_get16x16pred_error_mmx);
 extern prototype_variance2(vp8_get8x8var_mmx);
-extern prototype_variance2(vp8_get16x16var_mmx);
 extern prototype_get16x16prederror(vp8_get4x4sse_cs_mmx);
 
 #if !CONFIG_RUNTIME_CPU_DETECT
@@ -110,15 +108,6 @@ extern prototype_get16x16prederror(vp8_get4x4sse_cs_mmx);
 #undef  vp8_variance_mse16x16
 #define vp8_variance_mse16x16 vp8_mse16x16_mmx
 
-#undef  vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_mmx
-
-#undef  vp8_variance_get8x8var
-#define vp8_variance_get8x8var vp8_get8x8var_mmx
-
-#undef  vp8_variance_get16x16var
-#define vp8_variance_get16x16var vp8_get16x16var_mmx
-
 #undef  vp8_variance_get4x4sse_cs
 #define vp8_variance_get4x4sse_cs vp8_get4x4sse_cs_mmx
 
@@ -148,7 +137,6 @@ extern prototype_variance(vp8_variance_halfpixvar16x16_hv_wmt);
 extern prototype_subpixvariance(vp8_sub_pixel_mse16x16_wmt);
 extern prototype_getmbss(vp8_get_mb_ss_sse2);
 extern prototype_variance(vp8_mse16x16_wmt);
-extern prototype_get16x16prederror(vp8_get16x16pred_error_sse2);
 extern prototype_variance2(vp8_get8x8var_sse2);
 extern prototype_variance2(vp8_get16x16var_sse2);
 
@@ -216,15 +204,6 @@ extern prototype_variance2(vp8_get16x16var_sse2);
 #undef  vp8_variance_mse16x16
 #define vp8_variance_mse16x16 vp8_mse16x16_wmt
 
-#undef  vp8_variance_get16x16prederror
-#define vp8_variance_get16x16prederror vp8_get16x16pred_error_sse2
-
-#undef  vp8_variance_get8x8var
-#define vp8_variance_get8x8var vp8_get8x8var_sse2
-
-#undef  vp8_variance_get16x16var
-#define vp8_variance_get16x16var vp8_get16x16var_sse2
-
 #endif
 #endif
 
diff --git a/vp8/encoder/x86/x86_csystemdependent.c b/vp8/encoder/x86/x86_csystemdependent.c
index 378b14066..9a324ec12 100644
--- a/vp8/encoder/x86/x86_csystemdependent.c
+++ b/vp8/encoder/x86/x86_csystemdependent.c
@@ -16,7 +16,7 @@
 
 
 #if HAVE_MMX
-static void short_fdct8x4_mmx(short *input, short *output, int pitch)
+void vp8_short_fdct8x4_mmx(short *input, short *output, int pitch)
 {
     vp8_short_fdct4x4_mmx(input,   output,    pitch);
     vp8_short_fdct4x4_mmx(input + 4, output + 16, pitch);
@@ -26,7 +26,7 @@ int vp8_fast_quantize_b_impl_mmx(short *coeff_ptr, short *zbin_ptr,
                                  short *qcoeff_ptr, short *dequant_ptr,
                                  short *scan_mask, short *round_ptr,
                                  short *quant_ptr, short *dqcoeff_ptr);
-static void fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
+void vp8_fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
 {
     short *scan_mask   = vp8_default_zig_zag_mask;//d->scan_order_mask_ptr;
     short *coeff_ptr   = b->coeff;
@@ -51,7 +51,7 @@ static void fast_quantize_b_mmx(BLOCK *b, BLOCKD *d)
 }
 
 int vp8_mbblock_error_mmx_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-static int mbblock_error_mmx(MACROBLOCK *mb, int dc)
+int vp8_mbblock_error_mmx(MACROBLOCK *mb, int dc)
 {
     short *coeff_ptr =  mb->block[0].coeff;
     short *dcoef_ptr =  mb->e_mbd.block[0].dqcoeff;
@@ -59,7 +59,7 @@ static int mbblock_error_mmx(MACROBLOCK *mb, int dc)
 }
 
 int vp8_mbuverror_mmx_impl(short *s_ptr, short *d_ptr);
-static int mbuverror_mmx(MACROBLOCK *mb)
+int vp8_mbuverror_mmx(MACROBLOCK *mb)
 {
     short *s_ptr = &mb->coeff[256];
     short *d_ptr = &mb->e_mbd.dqcoeff[256];
@@ -69,7 +69,7 @@ static int mbuverror_mmx(MACROBLOCK *mb)
 void vp8_subtract_b_mmx_impl(unsigned char *z,  int src_stride,
                              short *diff, unsigned char *predictor,
                              int pitch);
-static void subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
+void vp8_subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
 {
     unsigned char *z = *(be->base_src) + be->src;
     unsigned int  src_stride = be->src_stride;
@@ -82,7 +82,7 @@ static void subtract_b_mmx(BLOCK *be, BLOCKD *bd, int pitch)
 
 #if HAVE_SSE2
 int vp8_mbblock_error_xmm_impl(short *coeff_ptr, short *dcoef_ptr, int dc);
-static int mbblock_error_xmm(MACROBLOCK *mb, int dc)
+int vp8_mbblock_error_xmm(MACROBLOCK *mb, int dc)
 {
     short *coeff_ptr =  mb->block[0].coeff;
     short *dcoef_ptr =  mb->e_mbd.block[0].dqcoeff;
@@ -90,7 +90,7 @@ static int mbblock_error_xmm(MACROBLOCK *mb, int dc)
 }
 
 int vp8_mbuverror_xmm_impl(short *s_ptr, short *d_ptr);
-static int mbuverror_xmm(MACROBLOCK *mb)
+int vp8_mbuverror_xmm(MACROBLOCK *mb)
 {
     short *s_ptr = &mb->coeff[256];
     short *d_ptr = &mb->e_mbd.dqcoeff[256];
@@ -100,7 +100,7 @@ static int mbuverror_xmm(MACROBLOCK *mb)
 void vp8_subtract_b_sse2_impl(unsigned char *z,  int src_stride,
                              short *diff, unsigned char *predictor,
                              int pitch);
-static void subtract_b_sse2(BLOCK *be, BLOCKD *bd, int pitch)
+void vp8_subtract_b_sse2(BLOCK *be, BLOCKD *bd, int pitch)
 {
     unsigned char *z = *(be->base_src) + be->src;
     unsigned int  src_stride = be->src_stride;
@@ -175,26 +175,23 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.variance.mse16x16              = vp8_mse16x16_mmx;
         cpi->rtcd.variance.getmbss               = vp8_get_mb_ss_mmx;
 
-        cpi->rtcd.variance.get16x16prederror     = vp8_get16x16pred_error_mmx;
-        cpi->rtcd.variance.get8x8var             = vp8_get8x8var_mmx;
-        cpi->rtcd.variance.get16x16var           = vp8_get16x16var_mmx;
         cpi->rtcd.variance.get4x4sse_cs          = vp8_get4x4sse_cs_mmx;
 
         cpi->rtcd.fdct.short4x4                  = vp8_short_fdct4x4_mmx;
-        cpi->rtcd.fdct.short8x4                  = short_fdct8x4_mmx;
+        cpi->rtcd.fdct.short8x4                  = vp8_short_fdct8x4_mmx;
         cpi->rtcd.fdct.fast4x4                   = vp8_short_fdct4x4_mmx;
-        cpi->rtcd.fdct.fast8x4                   = short_fdct8x4_mmx;
+        cpi->rtcd.fdct.fast8x4                   = vp8_short_fdct8x4_mmx;
 
         cpi->rtcd.fdct.walsh_short4x4            = vp8_short_walsh4x4_c;
 
         cpi->rtcd.encodemb.berr                  = vp8_block_error_mmx;
-        cpi->rtcd.encodemb.mberr                 = mbblock_error_mmx;
-        cpi->rtcd.encodemb.mbuverr               = mbuverror_mmx;
-        cpi->rtcd.encodemb.subb                  = subtract_b_mmx;
+        cpi->rtcd.encodemb.mberr                 = vp8_mbblock_error_mmx;
+        cpi->rtcd.encodemb.mbuverr               = vp8_mbuverror_mmx;
+        cpi->rtcd.encodemb.subb                  = vp8_subtract_b_mmx;
         cpi->rtcd.encodemb.submby                = vp8_subtract_mby_mmx;
         cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_mmx;
 
-        /*cpi->rtcd.quantize.fastquantb            = fast_quantize_b_mmx;*/
+        /*cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_mmx;*/
     }
 #endif
 
@@ -226,11 +223,6 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.variance.mse16x16              = vp8_mse16x16_wmt;
         cpi->rtcd.variance.getmbss               = vp8_get_mb_ss_sse2;
 
-        cpi->rtcd.variance.get16x16prederror     = vp8_get16x16pred_error_sse2;
-        cpi->rtcd.variance.get8x8var             = vp8_get8x8var_sse2;
-        cpi->rtcd.variance.get16x16var           = vp8_get16x16var_sse2;
-
-
         /* cpi->rtcd.variance.get4x4sse_cs  not implemented for wmt */;
 
         cpi->rtcd.fdct.short4x4                  = vp8_short_fdct4x4_sse2;
@@ -241,9 +233,9 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.fdct.walsh_short4x4            = vp8_short_walsh4x4_sse2 ;
 
         cpi->rtcd.encodemb.berr                  = vp8_block_error_xmm;
-        cpi->rtcd.encodemb.mberr                 = mbblock_error_xmm;
-        cpi->rtcd.encodemb.mbuverr               = mbuverror_xmm;
-        cpi->rtcd.encodemb.subb                  = subtract_b_sse2;
+        cpi->rtcd.encodemb.mberr                 = vp8_mbblock_error_xmm;
+        cpi->rtcd.encodemb.mbuverr               = vp8_mbuverror_xmm;
+        cpi->rtcd.encodemb.subb                  = vp8_subtract_b_sse2;
         cpi->rtcd.encodemb.submby                = vp8_subtract_mby_sse2;
         cpi->rtcd.encodemb.submbuv               = vp8_subtract_mbuv_sse2;