<html>
  <head>
    <meta content="text/html; charset=windows-1252"
      http-equiv="Content-Type">
  </head>
  <body text="#000000" bgcolor="#FFFFFF">
    <div class="moz-cite-prefix">On 05/19/2015 04:40 PM, chen wrote:<br>
    </div>
    <blockquote
      cite="mid:4828973e.10b5.14d6e8c7530.Coremail.chenm003@163.com"
      type="cite">
      <div
        style="line-height:1.7;color:#000000;font-size:14px;font-family:arial">
        <div style="color: rgb(0, 0, 0); line-height: 1.7; font-family:
          arial; font-size: 14px;">
          <div>Why x64 only? r4 always free in both x86 and x64</div>
        </div>
      </div>
    </blockquote>
    It hurts performance in the benchtest for x32 but I can make it
    cover both if that is what you want.<br>
    <blockquote
      cite="mid:4828973e.10b5.14d6e8c7530.Coremail.chenm003@163.com"
      type="cite">
      <div
        style="line-height:1.7;color:#000000;font-size:14px;font-family:arial">
        <div style="color: rgb(0, 0, 0); line-height: 1.7; font-family:
          arial; font-size: 14px;">
          <pre>
At 2015-05-20 07:33:14,<a moz-do-not-send="true" href="mailto:dtyx265@gmail.com">dtyx265@gmail.com</a> wrote:
># HG changeset patch
># User David T Yuen <<a moz-do-not-send="true" href="mailto:dtyx265@gmail.com">dtyx265@gmail.com</a>>
># Date 1432078001 25200
># Node ID 509f7cbf8e09d6ddec4aa58040cfd206879d59e7
># Parent  3e07cba4b2034db2b819b2e11e98ee4b851d52b5
>asm: interp_4tap_vert_pX_4xN sse2
>
>Improved register usage for addressing of output.  This improvement helps 64-bit .7% to 2.5% but hurt 32-bit performance so I made it 64-bit only.
>Also added interp_4tap_vert_ps_4x32 in primitives setup.
>
>diff -r 3e07cba4b203 -r 509f7cbf8e09 source/common/x86/asm-primitives.cpp
>--- a/source/common/x86/asm-primitives.cpp  Tue May 19 14:27:04 2015 -0700
>+++ b/source/common/x86/asm-primitives.cpp  Tue May 19 16:26:41 2015 -0700
>@@ -1482,6 +1482,7 @@
>         p.chroma[X265_CSP_I422].pu[CHROMA_422_4x4].filter_vps = x265_interp_4tap_vert_ps_4x4_sse2;
>         p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].filter_vps = x265_interp_4tap_vert_ps_4x8_sse2;
>         p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].filter_vps = x265_interp_4tap_vert_ps_4x16_sse2;
>+        p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].filter_vps = x265_interp_4tap_vert_ps_4x32_sse2;
>         p.chroma[X265_CSP_I444].pu[LUMA_4x4].filter_vps = x265_interp_4tap_vert_ps_4x4_sse2;
>         p.chroma[X265_CSP_I444].pu[LUMA_4x8].filter_vps = x265_interp_4tap_vert_ps_4x8_sse2;
>         p.chroma[X265_CSP_I444].pu[LUMA_4x16].filter_vps = x265_interp_4tap_vert_ps_4x16_sse2;
>diff -r 3e07cba4b203 -r 509f7cbf8e09 source/common/x86/ipfilter8.asm
>--- a/source/common/x86/ipfilter8.asm       Tue May 19 14:27:04 2015 -0700
>+++ b/source/common/x86/ipfilter8.asm       Tue May 19 16:26:41 2015 -0700
>@@ -1138,6 +1138,9 @@
>     mova        m1,        [pw_2000]
> %endif

>+%if ARCH_X86_64
>+    lea         r4,        [3 * r3]
>+%endif
>     lea         r5,        [3 * r1]
>     punpcklqdq  m0,        m0

>@@ -1243,11 +1246,23 @@
>     movd        [r2],      m2
>     psrldq      m2,        4
>     movd        [r2 + r3], m2
>+    psrldq      m2,        4
>+%if ARCH_X86_64
>+    movd        [r2 + 2 * r3],      m2
>+    psrldq      m2,        4
>+    movd        [r2 + r4], m2
>+%if x < %2/4
>+    lea         r2,        [r2 + 4 * r3]
>+%endif
>+%else
>     lea         r2,        [r2 + 2 * r3]
>-    psrldq      m2,        4
>     movd        [r2],      m2
>     psrldq      m2,        4
>     movd        [r2 + r3], m2
>+%if x < %2/4
>+    lea         r2,        [r2 + 2 * r3]
>+%endif
>+%endif
> %elifidn %1,ps
>     psrldq      m4,        2
>     psrldq      m5,        2
>@@ -1255,14 +1270,25 @@
>     pshufd      m5,        m5, q3120
>     punpcklqdq  m4,        m5
>     psubw       m4,        m1
>+%if ARCH_X86_64
>+    movh        [r2 + 2 * r3],      m4
>+    movhps      [r2 + r4], m4
>+%if x < %2/4
>+    lea         r2,        [r2 + 4 * r3]
>+%endif
>+%else
>     lea         r2,        [r2 + 2 * r3]
>     movh        [r2],      m4
>     movhps      [r2 + r3], m4
>-%endif
>-
> %if x < %2/4
>     lea         r2,        [r2 + 2 * r3]
> %endif
>+%endif
>+%endif
>+
>+; %if x < %2/4
>+;     lea         r2,        [r2 + 4 * r3]
>+; %endif

> %assign x x+1
> %endrep
>_______________________________________________
>x265-devel mailing list
><a moz-do-not-send="true" href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a>
><a class="moz-txt-link-freetext" href="https://mailman.videolan.org/listinfo/x265-devel">https://mailman.videolan.org/listinfo/x265-devel</a>
</pre>
        </div>
      </div>
      <br>
      <fieldset class="mimeAttachmentHeader"></fieldset>
      <br>
      <pre wrap="">_______________________________________________
x265-devel mailing list
<a class="moz-txt-link-abbreviated" href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a>
<a class="moz-txt-link-freetext" href="https://mailman.videolan.org/listinfo/x265-devel">https://mailman.videolan.org/listinfo/x265-devel</a>
</pre>
    </blockquote>
    <br>
  </body>
</html>