<div dir="ltr"><br><div class="gmail_extra"><br><br><div class="gmail_quote">On Fri, Mar 21, 2014 at 8:01 PM, chen <span dir="ltr"><<a href="mailto:chenm003@163.com" target="_blank">chenm003@163.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">At 2014-03-21 13:35:31,<a href="mailto:murugan@multicorewareinc.com">murugan@multicorewareinc.com</a> wrote:<br>
<div class="">># HG changeset patch<br>
># User Murugan Vairavel <<a href="mailto:murugan@multicorewareinc.com">murugan@multicorewareinc.com</a>><br>
># Date 1395379456 -19800<br>
># Fri Mar 21 10:54:16 2014 +0530<br>
># Node ID 29728f7728591116192575d411ef2db2dff49c18<br>
># Parent 435e50b2b92c83e10fdb2bd86bc8e8df91b7338b<br>
>asm: code for input pixel upShift/downShift<br>
><br>
</div><div class="">>+; Input 10bpp, Output 8bpp, width is multiple of 16<br>
>+;------------------------------------------------------------------------------------------------------------------------<br>
>+;void planecopy_sp(uint16_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height, int shift, uint16_t mask)<br>
>+;------------------------------------------------------------------------------------------------------------------------<br>
>+INIT_XMM sse2<br>
>+cglobal downShift_10, 7,7,3<br>
>+ movd m0, r6d ; m0 = shift<br>
>+ add r1, r1<br>
>+ dec r5d>+.loopH:><br>
>+ xor r6, r6<br>
</div>tip: r6 is a offset, when you do prepare 'r1=r1-r4', you may direct operator on r0<br></blockquote><div>But the pixels processed in each row is not equal to the width(r4), in case the width is not a multiple of 16. If i do it as above then the output mismatch will occur. </div>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class=""><br>
>+.loopW:<br>
>+ movu m1, [r0 + r6 * 2]<br>
>+ movu m2, [r0 + r6 * 2 + 16]<br>
>+ psrlw m1, m0<br>
>+ psrlw m2, m0<br>
>+ packuswb m1, m2<br>
>+ movu [r2 + r6], m1<br>
>+<br>
>+ add r6, 16<br>
>+ cmp r6d, r4d<br>
>+ jl .loopW<br>
>+<br>
>+ ; move to next row<br>
>+ lea r0, [r0 + r1]<br>
>+ lea r2, [r2 + r3]<br>
</div>add r0,r1<br>
add r2,r3<br></blockquote><div>I will modify that. </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class=""><br>
>+ dec r5d<br>
>+ jnz .loopH<br>
>+<br>
>+;processing last row of every frame [To handle width which not a multiple of 16]<br>
>+<br>
>+.loop16:<br>
>+ movu m1, [r0]<br>
>+ movu m2, [r0 + 16]<br>
>+ psrlw m1, m0<br>
>+ psrlw m2, m0<br>
>+ packuswb m1, m2<br>
>+ movu [r2], m1<br>
>+<br>
>+ add r0, 2 * mmsize<br>
>+ add r2, mmsize<br>
>+ sub r4d, 16<br>
>+ jz .end<br>
>+ cmp r4d, 15<br>
>+ jg .loop16<br>
<br>
</div>--> (X > 16) && (X >15) ??<br></blockquote><div>means?? </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div><div class="h5"><br>
>+ cmp r4d, 8<br>
>+ jl .process4<br>
>+ movu m1, [r0]<br>
>+ psrlw m1, m0<br>
>+ packuswb m1, m1<br>
>+ movh [r2], m1<br>
>+<br>
>+ add r0, mmsize<br>
>+ add r2, 8<br>
>+ sub r4d, 8<br>
>+ jz .end<br>
>+<br>
>+.process4:<br>
>+ cmp r4d, 4<br>
>+ jl .process2<br>
>+ movh m1,[r0]<br>
>+ psrlw m1, m0<br>
>+ packuswb m1, m1<br>
>+ movd [r2], m1<br>
>+<br>
>+ add r0, 8<br>
>+ add r2, 4<br>
>+ sub r4d, 4<br>
>+ jz .end<br>
>+<br>
>+.process2:<br>
>+ cmp r4d, 2<br>
>+ jl .process1<br>
>+ movd m1, [r0]<br>
>+ psrlw m1, m0<br>
>+ packuswb m1, m1<br>
>+ movd r6, m1<br>
>+ mov [r2], r6w<br>
>+<br>
>+ add r0, 4<br>
>+ add r2, 2<br>
>+ sub r4d, 2<br>
>+ jz .end<br>
>+<br>
>+.process1:<br>
>+ movd m1, [r0]<br>
>+ psrlw m1, m0<br>
>+ packuswb m1, m1<br>
>+ movd r6, m1<br>
>+ mov [r2], r6b<br>
>+.end:<br>
>+ RET<br>
<br>
</div></div>(4, 2, 1) pixels path may share calculate code<br></blockquote><div>Do you mean defining a macro for that?? </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class=""><br>
<br>
>+; Input 8bpp, Output 16bpp, width is multiple of 16<br>
>+;-----------------------------------------------------------------------------------------------------<br>
>+;void planecopy_cp(uint8_t *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int width, int height)<br>
>+;-----------------------------------------------------------------------------------------------------<br>
>+INIT_XMM sse4<br>
</div>>+cglobal upShift_8, 6,7,2upShift_8to10<br>
<div><div class="h5">>+<br>
>+ add r3, r3<br>
>+ dec r5d<br>
>+<br>
>+.loopH:<br>
>+ xor r6, r6<br>
>+.loopW:<br>
>+ pmovzxbw m0,[r0 + r6]<br>
>+ pmovzxbw m1,[r0 + r6 + 8]<br>
>+ psllw m0, 2<br>
>+ psllw m1, 2<br>
>+ movu [r2 + r6 * 2], m0<br>
>+ movu [r2 + r6 * 2 + 16], m1<br>
>+<br>
>+ add r6, 16<br>
>+ cmp r6d, r4d<br>
>+ jl .loopW<br>
>+<br>
>+ ; move to next row<br>
>+ lea r0, [r0 + r1]<br>
>+ lea r2, [r2 + r3]<br>
>+ dec r5d<br>
>+ jnz .loopH<br>
>+<br>
>+;processing last row of every frame [To handle width which not a multiple of 16]<br>
<br>
</div></div>same comment as previous module<br>
<div><div class="h5"><br>
>+.loop16:<br>
>+ pmovzxbw m0,[r0]<br>
>+ pmovzxbw m1,[r0 + 8]<br>
>+ psllw m0, 2<br>
>+ psllw m1, 2<br>
>+ movu [r2], m0<br>
>+ movu [r2 + 16], m1<br>
>+<br>
>+ add r0, mmsize<br>
>+ add r2, 2 * mmsize<br>
>+ sub r4d, 16<br>
>+ jz .end<br>
>+ cmp r4d, 15<br>
>+ jg .loop16<br>
>+<br>
>+ cmp r4d, 8<br>
>+ jl .process4<br>
>+ pmovzxbw m0,[r0]<br>
>+ psllw m0, 2<br>
>+ movu [r2], m0<br>
>+<br>
>+ add r0, 8<br>
>+ add r2, mmsize<br>
>+ sub r4d, 8<br>
>+ jz .end<br>
>+<br>
>+.process4:<br>
>+ cmp r4d, 4<br>
>+ jl .process2<br>
>+ movd m0,[r0]<br>
>+ pmovzxbw m0,m0<br>
>+ psllw m0, 2<br>
>+ movh [r2], m0<br>
>+<br>
>+ add r0, 4<br>
>+ add r2, 8<br>
>+ sub r4d, 4<br>
>+ jz .end<br>
>+<br>
>+.process2:<br>
>+ cmp r4d, 2<br>
>+ jl .process1<br>
>+ movzx r6d, byte [r0]<br>
>+ shl r6d, 2<br>
>+ mov [r2], r6w<br>
>+ movzx r6d, byte [r0 + 1]<br>
>+ shl r6d, 2<br>
>+ mov [r2 + 2], r6w<br>
>+<br>
>+ add r0, 2<br>
>+ add r2, 4<br>
>+ sub r4d, 2<br>
>+ jz .end<br>
>+<br>
>+.process1:<br>
>+ movzx r6d, byte [r0]<br>
>+ shl r6d, 2<br>
>+ mov [r2], r6w<br>
>+.end:<br>
>+ RET<br>
</div></div>_______________________________________________<br>
x265-devel mailing list<br>
<a href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a><br>
<a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank">https://mailman.videolan.org/listinfo/x265-devel</a><br>
</blockquote></div><br><br clear="all"><div><br></div>-- <br><div dir="ltr">With Regards,<div><br></div><div>Murugan. V</div><div>+919659287478</div></div>
</div></div>