Write SSSE3 sub-pixel filter function

1. Process 16 pixels at one time instead of 8.
2. Add check for both xoffset =0 and yoffset=0, which happens
   during motion search.
This change gave encoder 1%~3% performance gain.

Change-Id: Idaa39506b48f4f8b2fbbeb45aae8226fa32afb3e

5 files changed, 496 insertions, 0 deletions

diff --git a/vp8/encoder/x86/variance_impl_ssse3.asm b/vp8/encoder/x86/variance_impl_ssse3.asm
new file mode 100644
index 000000000..b1976328d
--- /dev/null
+++ b/vp8/encoder/x86/variance_impl_ssse3.asm
@@ -0,0 +1,348 @@
+;
+;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+;
+;  Use of this source code is governed by a BSD-style license
+;  that can be found in the LICENSE file in the root of the source
+;  tree. An additional intellectual property rights grant can be found
+;  in the file PATENTS.  All contributing project authors may
+;  be found in the AUTHORS file in the root of the source tree.
+;
+
+
+%include "vpx_ports/x86_abi_support.asm"
+
+%define xmm_filter_shift            7
+
+
+;void vp8_filter_block2d_bil_var_ssse3
+;(
+;    unsigned char *ref_ptr,
+;    int ref_pixels_per_line,
+;    unsigned char *src_ptr,
+;    int src_pixels_per_line,
+;    unsigned int Height,
+;    int  xoffset,
+;    int  yoffset,
+;    int *sum,
+;    unsigned int *sumsquared;;
+;
+;)
+;Note: The filter coefficient at offset=0 is 128. Since the second register
+;for Pmaddubsw is signed bytes, we must calculate zero offset seperately.
+global sym(vp8_filter_block2d_bil_var_ssse3)
+sym(vp8_filter_block2d_bil_var_ssse3):
+    push        rbp
+    mov         rbp, rsp
+    SHADOW_ARGS_TO_STACK 9
+    SAVE_XMM
+    GET_GOT     rbx
+    push rsi
+    push rdi
+    push rbx
+    ; end prolog
+
+        pxor            xmm6,           xmm6
+        pxor            xmm7,           xmm7
+
+        lea             rcx,            [GLOBAL(vp8_bilinear_filters_ssse3)]
+        movsxd          rax,            dword ptr arg(5)     ; xoffset
+
+        cmp             rax,            0                    ; skip first_pass filter if xoffset=0
+        je              filter_block2d_bil_var_ssse3_sp_only
+
+        shl             rax,            4                    ; point to filter coeff with xoffset
+        lea             rax,            [rax + rcx]          ; HFilter
+
+        movsxd          rdx,            dword ptr arg(6)     ; yoffset
+
+        cmp             rdx,            0                    ; skip second_pass filter if yoffset=0
+        je              filter_block2d_bil_var_ssse3_fp_only
+
+        shl             rdx,            4
+        lea             rdx,            [rdx + rcx]          ; VFilter
+
+        mov             rsi,            arg(0)               ;ref_ptr
+        mov             rdi,            arg(2)               ;src_ptr
+        movsxd          rcx,            dword ptr arg(4)     ;Height
+
+        movdqu          xmm0,           XMMWORD PTR [rsi]
+        movdqu          xmm1,           XMMWORD PTR [rsi+1]
+        movdqa          xmm2,           xmm0
+
+        punpcklbw       xmm0,           xmm1
+        punpckhbw       xmm2,           xmm1
+        pmaddubsw       xmm0,           [rax]
+        pmaddubsw       xmm2,           [rax]
+
+        paddw           xmm0,           [GLOBAL(xmm_bi_rd)]
+        paddw           xmm2,           [GLOBAL(xmm_bi_rd)]
+        psraw           xmm0,           xmm_filter_shift
+        psraw           xmm2,           xmm_filter_shift
+
+        packuswb        xmm0,           xmm2
+
+        movsxd          rbx,            dword ptr arg(1) ;ref_pixels_per_line
+        lea             rsi,            [rsi + rbx]
+%if ABI_IS_32BIT=0
+        movsxd          r9,             dword ptr arg(3) ;src_pixels_per_line
+%endif
+
+filter_block2d_bil_var_ssse3_loop:
+        movdqu          xmm1,           XMMWORD PTR [rsi]
+        movdqu          xmm2,           XMMWORD PTR [rsi+1]
+        movdqa          xmm3,           xmm1
+
+        punpcklbw       xmm1,           xmm2
+        punpckhbw       xmm3,           xmm2
+        pmaddubsw       xmm1,           [rax]
+        pmaddubsw       xmm3,           [rax]
+
+        paddw           xmm1,           [GLOBAL(xmm_bi_rd)]
+        paddw           xmm3,           [GLOBAL(xmm_bi_rd)]
+        psraw           xmm1,           xmm_filter_shift
+        psraw           xmm3,           xmm_filter_shift
+        packuswb        xmm1,           xmm3
+
+        movdqa          xmm2,           xmm0
+        movdqa          xmm0,           xmm1
+        movdqa          xmm3,           xmm2
+
+        punpcklbw       xmm2,           xmm1
+        punpckhbw       xmm3,           xmm1
+        pmaddubsw       xmm2,           [rdx]
+        pmaddubsw       xmm3,           [rdx]
+
+        paddw           xmm2,           [GLOBAL(xmm_bi_rd)]
+        paddw           xmm3,           [GLOBAL(xmm_bi_rd)]
+        psraw           xmm2,           xmm_filter_shift
+        psraw           xmm3,           xmm_filter_shift
+
+        movq            xmm1,           QWORD PTR [rdi]
+        pxor            xmm4,           xmm4
+        punpcklbw       xmm1,           xmm4
+        movq            xmm5,           QWORD PTR [rdi+8]
+        punpcklbw       xmm5,           xmm4
+
+        psubw           xmm2,           xmm1
+        psubw           xmm3,           xmm5
+        paddw           xmm6,           xmm2
+        paddw           xmm6,           xmm3
+        pmaddwd         xmm2,           xmm2
+        pmaddwd         xmm3,           xmm3
+        paddd           xmm7,           xmm2
+        paddd           xmm7,           xmm3
+
+        lea             rsi,            [rsi + rbx]          ;ref_pixels_per_line
+%if ABI_IS_32BIT
+        add             rdi,            dword ptr arg(3)     ;src_pixels_per_line
+%else
+        lea             rdi,            [rdi + r9]
+%endif
+
+        sub             rcx,            1
+        jnz             filter_block2d_bil_var_ssse3_loop
+
+        jmp             filter_block2d_bil_variance
+
+filter_block2d_bil_var_ssse3_sp_only:
+        movsxd          rdx,            dword ptr arg(6)     ; yoffset
+
+        cmp             rdx,            0                    ; Both xoffset =0 and yoffset=0
+        je              filter_block2d_bil_var_ssse3_full_pixel
+
+        shl             rdx,            4
+        lea             rdx,            [rdx + rcx]          ; VFilter
+
+        mov             rsi,            arg(0)               ;ref_ptr
+        mov             rdi,            arg(2)               ;src_ptr
+        movsxd          rcx,            dword ptr arg(4)     ;Height
+        movsxd          rax,            dword ptr arg(1)     ;ref_pixels_per_line
+
+        movdqu          xmm1,           XMMWORD PTR [rsi]
+        movdqa          xmm0,           xmm1
+
+        movsxd          rbx,            dword ptr arg(3)     ;src_pixels_per_line
+        lea             rsi,            [rsi + rax]
+
+filter_block2d_bil_sp_only_loop:
+        movdqu          xmm3,           XMMWORD PTR [rsi]
+        movdqa          xmm2,           xmm1
+        movdqa          xmm0,           xmm3
+
+        punpcklbw       xmm1,           xmm3
+        punpckhbw       xmm2,           xmm3
+        pmaddubsw       xmm1,           [rdx]
+        pmaddubsw       xmm2,           [rdx]
+
+        paddw           xmm1,           [GLOBAL(xmm_bi_rd)]
+        paddw           xmm2,           [GLOBAL(xmm_bi_rd)]
+        psraw           xmm1,           xmm_filter_shift
+        psraw           xmm2,           xmm_filter_shift
+
+        movq            xmm3,           QWORD PTR [rdi]
+        pxor            xmm4,           xmm4
+        punpcklbw       xmm3,           xmm4
+        movq            xmm5,           QWORD PTR [rdi+8]
+        punpcklbw       xmm5,           xmm4
+
+        psubw           xmm1,           xmm3
+        psubw           xmm2,           xmm5
+        paddw           xmm6,           xmm1
+        paddw           xmm6,           xmm2
+        pmaddwd         xmm1,           xmm1
+        pmaddwd         xmm2,           xmm2
+        paddd           xmm7,           xmm1
+        paddd           xmm7,           xmm2
+
+        movdqa          xmm1,           xmm0
+        lea             rsi,            [rsi + rax]          ;ref_pixels_per_line
+        lea             rdi,            [rdi + rbx]          ;src_pixels_per_line
+
+        sub             rcx,            1
+        jnz             filter_block2d_bil_sp_only_loop
+
+        jmp             filter_block2d_bil_variance
+
+filter_block2d_bil_var_ssse3_full_pixel:
+        mov             rsi,            arg(0)               ;ref_ptr
+        mov             rdi,            arg(2)               ;src_ptr
+        movsxd          rcx,            dword ptr arg(4)     ;Height
+        movsxd          rax,            dword ptr arg(1)     ;ref_pixels_per_line
+        movsxd          rbx,            dword ptr arg(3)     ;src_pixels_per_line
+        pxor            xmm0,           xmm0
+
+filter_block2d_bil_full_pixel_loop:
+        movq            xmm1,           QWORD PTR [rsi]
+        punpcklbw       xmm1,           xmm0
+        movq            xmm2,           QWORD PTR [rsi+8]
+        punpcklbw       xmm2,           xmm0
+
+        movq            xmm3,           QWORD PTR [rdi]
+        punpcklbw       xmm3,           xmm0
+        movq            xmm4,           QWORD PTR [rdi+8]
+        punpcklbw       xmm4,           xmm0
+
+        psubw           xmm1,           xmm3
+        psubw           xmm2,           xmm4
+        paddw           xmm6,           xmm1
+        paddw           xmm6,           xmm2
+        pmaddwd         xmm1,           xmm1
+        pmaddwd         xmm2,           xmm2
+        paddd           xmm7,           xmm1
+        paddd           xmm7,           xmm2
+
+        lea             rsi,            [rsi + rax]          ;ref_pixels_per_line
+        lea             rdi,            [rdi + rbx]          ;src_pixels_per_line
+        sub             rcx,            1
+        jnz             filter_block2d_bil_full_pixel_loop
+
+        jmp             filter_block2d_bil_variance
+
+filter_block2d_bil_var_ssse3_fp_only:
+        mov             rsi,            arg(0)               ;ref_ptr
+        mov             rdi,            arg(2)               ;src_ptr
+        movsxd          rcx,            dword ptr arg(4)     ;Height
+        movsxd          rdx,            dword ptr arg(1)     ;ref_pixels_per_line
+
+        pxor            xmm0,           xmm0
+        movsxd          rbx,            dword ptr arg(3)     ;src_pixels_per_line
+
+filter_block2d_bil_fp_only_loop:
+        movdqu          xmm1,           XMMWORD PTR [rsi]
+        movdqu          xmm2,           XMMWORD PTR [rsi+1]
+        movdqa          xmm3,           xmm1
+
+        punpcklbw       xmm1,           xmm2
+        punpckhbw       xmm3,           xmm2
+        pmaddubsw       xmm1,           [rax]
+        pmaddubsw       xmm3,           [rax]
+
+        paddw           xmm1,           [GLOBAL(xmm_bi_rd)]
+        paddw           xmm3,           [GLOBAL(xmm_bi_rd)]
+        psraw           xmm1,           xmm_filter_shift
+        psraw           xmm3,           xmm_filter_shift
+
+        movq            xmm2,           XMMWORD PTR [rdi]
+        pxor            xmm4,           xmm4
+        punpcklbw       xmm2,           xmm4
+        movq            xmm5,           QWORD PTR [rdi+8]
+        punpcklbw       xmm5,           xmm4
+
+        psubw           xmm1,           xmm2
+        psubw           xmm3,           xmm5
+        paddw           xmm6,           xmm1
+        paddw           xmm6,           xmm3
+        pmaddwd         xmm1,           xmm1
+        pmaddwd         xmm3,           xmm3
+        paddd           xmm7,           xmm1
+        paddd           xmm7,           xmm3
+
+        lea             rsi,            [rsi + rdx]
+        lea             rdi,            [rdi + rbx]          ;src_pixels_per_line
+
+        sub             rcx,            1
+        jnz             filter_block2d_bil_fp_only_loop
+
+        jmp             filter_block2d_bil_variance
+
+filter_block2d_bil_variance:
+        pxor        xmm0,           xmm0
+        pxor        xmm1,           xmm1
+        pxor        xmm5,           xmm5
+
+        punpcklwd   xmm0,           xmm6
+        punpckhwd   xmm1,           xmm6
+        psrad       xmm0,           16
+        psrad       xmm1,           16
+        paddd       xmm0,           xmm1
+        movdqa      xmm1,           xmm0
+
+        movdqa      xmm6,           xmm7
+        punpckldq   xmm6,           xmm5
+        punpckhdq   xmm7,           xmm5
+        paddd       xmm6,           xmm7
+
+        punpckldq   xmm0,           xmm5
+        punpckhdq   xmm1,           xmm5
+        paddd       xmm0,           xmm1
+
+        movdqa      xmm7,           xmm6
+        movdqa      xmm1,           xmm0
+
+        psrldq      xmm7,           8
+        psrldq      xmm1,           8
+
+        paddd       xmm6,           xmm7
+        paddd       xmm0,           xmm1
+
+        mov         rsi,            arg(7) ;[Sum]
+        mov         rdi,            arg(8) ;[SSE]
+
+        movd        [rsi],       xmm0
+        movd        [rdi],       xmm6
+
+    ; begin epilog
+    pop rbx
+    pop rdi
+    pop rsi
+    RESTORE_GOT
+    RESTORE_XMM
+    UNSHADOW_ARGS
+    pop         rbp
+    ret
+
+
+SECTION_RODATA
+align 16
+xmm_bi_rd:
+    times 8 dw 64
+align 16
+vp8_bilinear_filters_ssse3:
+    times 8 db 128, 0
+    times 8 db 112, 16
+    times 8 db 96,  32
+    times 8 db 80,  48
+    times 8 db 64,  64
+    times 8 db 48,  80
+    times 8 db 32,  96
+    times 8 db 16,  112
diff --git a/vp8/encoder/x86/variance_ssse3.c b/vp8/encoder/x86/variance_ssse3.c
new file mode 100644
index 000000000..750ae8b86
--- /dev/null
+++ b/vp8/encoder/x86/variance_ssse3.c
@@ -0,0 +1,140 @@
+/*
+ *  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ *
+ *  Use of this source code is governed by a BSD-style license
+ *  that can be found in the LICENSE file in the root of the source
+ *  tree. An additional intellectual property rights grant can be found
+ *  in the file PATENTS.  All contributing project authors may
+ *  be found in the AUTHORS file in the root of the source tree.
+ */
+
+
+#include "vp8/encoder/variance.h"
+#include "vp8/common/pragmas.h"
+#include "vpx_ports/mem.h"
+
+extern unsigned int vp8_get16x16var_sse2
+(
+    const unsigned char *src_ptr,
+    int source_stride,
+    const unsigned char *ref_ptr,
+    int recon_stride,
+    unsigned int *SSE,
+    int *Sum
+);
+extern void vp8_half_horiz_vert_variance16x_h_sse2
+(
+    const unsigned char *ref_ptr,
+    int ref_pixels_per_line,
+    const unsigned char *src_ptr,
+    int src_pixels_per_line,
+    unsigned int Height,
+    int *sum,
+    unsigned int *sumsquared
+);
+extern void vp8_half_horiz_variance16x_h_sse2
+(
+    const unsigned char *ref_ptr,
+    int ref_pixels_per_line,
+    const unsigned char *src_ptr,
+    int src_pixels_per_line,
+    unsigned int Height,
+    int *sum,
+    unsigned int *sumsquared
+);
+extern void vp8_half_vert_variance16x_h_sse2
+(
+    const unsigned char *ref_ptr,
+    int ref_pixels_per_line,
+    const unsigned char *src_ptr,
+    int src_pixels_per_line,
+    unsigned int Height,
+    int *sum,
+    unsigned int *sumsquared
+);
+extern void vp8_filter_block2d_bil_var_ssse3
+(
+    const unsigned char *ref_ptr,
+    int ref_pixels_per_line,
+    const unsigned char *src_ptr,
+    int src_pixels_per_line,
+    unsigned int Height,
+    int  xoffset,
+    int  yoffset,
+    int *sum,
+    unsigned int *sumsquared
+);
+
+unsigned int vp8_sub_pixel_variance16x16_ssse3
+(
+    const unsigned char  *src_ptr,
+    int  src_pixels_per_line,
+    int  xoffset,
+    int  yoffset,
+    const unsigned char *dst_ptr,
+    int dst_pixels_per_line,
+    unsigned int *sse
+)
+{
+    int xsum0, xsum1;
+    unsigned int xxsum0, xxsum1;
+
+    // note we could avoid these if statements if the calling function
+    // just called the appropriate functions inside.
+    if (xoffset == 4 && yoffset == 0)
+    {
+        vp8_half_horiz_variance16x_h_sse2(
+            src_ptr, src_pixels_per_line,
+            dst_ptr, dst_pixels_per_line, 16,
+            &xsum0, &xxsum0);
+
+        vp8_half_horiz_variance16x_h_sse2(
+            src_ptr + 8, src_pixels_per_line,
+            dst_ptr + 8, dst_pixels_per_line, 16,
+            &xsum1, &xxsum1);
+
+        xsum0 += xsum1;
+        xxsum0 += xxsum1;
+    }
+    else if (xoffset == 0 && yoffset == 4)
+    {
+        vp8_half_vert_variance16x_h_sse2(
+            src_ptr, src_pixels_per_line,
+            dst_ptr, dst_pixels_per_line, 16,
+            &xsum0, &xxsum0);
+
+        vp8_half_vert_variance16x_h_sse2(
+            src_ptr + 8, src_pixels_per_line,
+            dst_ptr + 8, dst_pixels_per_line, 16,
+            &xsum1, &xxsum1);
+
+        xsum0 += xsum1;
+        xxsum0 += xxsum1;
+    }
+    else if (xoffset == 4 && yoffset == 4)
+    {
+        vp8_half_horiz_vert_variance16x_h_sse2(
+            src_ptr, src_pixels_per_line,
+            dst_ptr, dst_pixels_per_line, 16,
+            &xsum0, &xxsum0);
+
+        vp8_half_horiz_vert_variance16x_h_sse2(
+            src_ptr + 8, src_pixels_per_line,
+            dst_ptr + 8, dst_pixels_per_line, 16,
+            &xsum1, &xxsum1);
+
+        xsum0 += xsum1;
+        xxsum0 += xxsum1;
+    }
+    else
+    {
+      vp8_filter_block2d_bil_var_ssse3(
+          src_ptr, src_pixels_per_line,
+          dst_ptr, dst_pixels_per_line, 16,
+          xoffset, yoffset,
+          &xsum0, &xxsum0);
+    }
+
+    *sse = xxsum0;
+    return (xxsum0 - ((xsum0 * xsum0) >> 8));
+}
diff --git a/vp8/encoder/x86/variance_x86.h b/vp8/encoder/x86/variance_x86.h
index 6bea15ebc..1e2fb3490 100644
--- a/vp8/encoder/x86/variance_x86.h
+++ b/vp8/encoder/x86/variance_x86.h
@@ -286,6 +286,7 @@ extern prototype_sad_multi_dif_address(vp8_sad4x4x4d_sse3);
 #if HAVE_SSSE3
 extern prototype_sad_multi_same_address(vp8_sad16x16x3_ssse3);
 extern prototype_sad_multi_same_address(vp8_sad16x8x3_ssse3);
+extern prototype_subpixvariance(vp8_sub_pixel_variance16x16_ssse3);
 
 #if !CONFIG_RUNTIME_CPU_DETECT
 #undef  vp8_variance_sad16x16x3
@@ -294,6 +295,9 @@ extern prototype_sad_multi_same_address(vp8_sad16x8x3_ssse3);
 #undef  vp8_variance_sad16x8x3
 #define vp8_variance_sad16x8x3 vp8_sad16x8x3_ssse3
 
+#undef  vp8_variance_subpixvar16x16
+#define vp8_variance_subpixvar16x16 vp8_sub_pixel_variance16x16_ssse3
+
 #endif
 #endif
 
diff --git a/vp8/encoder/x86/x86_csystemdependent.c b/vp8/encoder/x86/x86_csystemdependent.c
index 61c603229..c7639a7e4 100644
--- a/vp8/encoder/x86/x86_csystemdependent.c
+++ b/vp8/encoder/x86/x86_csystemdependent.c
@@ -334,6 +334,8 @@ void vp8_arch_x86_encoder_init(VP8_COMP *cpi)
         cpi->rtcd.variance.sad16x16x3            = vp8_sad16x16x3_ssse3;
         cpi->rtcd.variance.sad16x8x3             = vp8_sad16x8x3_ssse3;
 
+        cpi->rtcd.variance.subpixvar16x16        = vp8_sub_pixel_variance16x16_ssse3;
+
         cpi->rtcd.quantize.fastquantb            = vp8_fast_quantize_b_ssse3;
 
     }
diff --git a/vp8/vp8cx.mk b/vp8/vp8cx.mk
index ed2feddae..c0ae250f5 100644
--- a/vp8/vp8cx.mk
+++ b/vp8/vp8cx.mk
@@ -110,6 +110,8 @@ VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/subtract_sse2.asm
 VP8_CX_SRCS-$(HAVE_SSE2) += encoder/x86/temporal_filter_apply_sse2.asm
 VP8_CX_SRCS-$(HAVE_SSE3) += encoder/x86/sad_sse3.asm
 VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/sad_ssse3.asm
+VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/variance_ssse3.c
+VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/variance_impl_ssse3.asm
 VP8_CX_SRCS-$(HAVE_SSSE3) += encoder/x86/quantize_ssse3.asm
 VP8_CX_SRCS-$(HAVE_SSE4_1) += encoder/x86/sad_sse4.asm
 VP8_CX_SRCS-$(ARCH_X86)$(ARCH_X86_64) += encoder/x86/quantize_mmx.asm