Hi guys,<br>I found something interesting.<br>The option under investigation is <span style="color: rgb(255, 0, 0);">subme</span>.<br>Executing the simple call x264 c:\mobile_cif.y4m --bitrate 100 -subme 8 -o mobile.mkv<br>
x264 (r999) code crashes. <br>It appens even when subme equals to 9.<br><br>For lower values it works fine.<br>
<br>I read the changelog of Tue Sep 30 18:34:56 2008 (r996)<br>and I saw that darkshikari reworked subme system.<br>Any suggestions?<br><br>Anyway I'll test r995 tomorrow.<br><br>Txn<br>David<br><br><div class="gmail_quote">
2008/10/21 Gabriel Bouvigne <span dir="ltr"><<a href="mailto:gabriel.bouvigne@joost.com" target="_blank">gabriel.bouvigne@joost.com</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
David Sullivan a écrit :<div><br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
It crashes with the patch applied with the following parameters.<br>
c:\mobile_cif.y4m --pass 1 --bitrate 100 --stats "mobile.stats" --ref 8 --mixed-refs --bframes 3<br>
--b-adapt 2 --b-pyramid --weightb --direct auto --filter -1:-1 --subme 8 --trellis 2 --partitions p8x8,b8x8,i4x4,i8x8 --8x8dct<br>
--me umh --threads auto --thread-input --progress --no-psnr --no-ssim -o c:\mob.mkv<br>
</blockquote>
<br></div>
It might be related to the reworked weighted predictions from this commit:<br>
<a href="http://git.videolan.org/?p=x264.git;a=commitdiff;h=e21b11f4ae99ef8ec0e37c1605f72ef18adfa7c3" target="_blank">http://git.videolan.org/?p=x264.git;a=commitdiff;h=e21b11f4ae99ef8ec0e37c1605f72ef18adfa7c3</a><br>
<br>
In this case, either removing --weightb or applying this new patch would solve it. (you should apply both this wpred patch and the already posted hadamard patch)<br>
<br>
If that does not work, then the most efficient thing to do would be to check from which revision x264 started to crash. (which I can't do easily on my side as my computer is a Core and not a Core2)<br>
<br>
--<br><font color="#888888">
Gabriel<br>
</font><br>diff --git a/common/x86/mc-c.c b/common/x86/mc-c.c<br>
index 1af512b..ca4a127 100644<br>
--- a/common/x86/mc-c.c<br>
+++ b/common/x86/mc-c.c<br>
@@ -310,14 +310,17 @@ void x264_mc_init_mmx( int cpu, x264_mc_functions_t *pf )<br>
if( !(cpu&X264_CPU_SSSE3) )<br>
return;<br>
<br>
- pf->avg[PIXEL_16x16] = x264_pixel_avg_16x16_ssse3;<br>
- pf->avg[PIXEL_16x8] = x264_pixel_avg_16x8_ssse3;<br>
- pf->avg[PIXEL_8x16] = x264_pixel_avg_8x16_ssse3;<br>
- pf->avg[PIXEL_8x8] = x264_pixel_avg_8x8_ssse3;<br>
- pf->avg[PIXEL_8x4] = x264_pixel_avg_8x4_ssse3;<br>
- pf->avg[PIXEL_4x8] = x264_pixel_avg_4x8_ssse3;<br>
- pf->avg[PIXEL_4x4] = x264_pixel_avg_4x4_ssse3;<br>
- pf->avg[PIXEL_4x2] = x264_pixel_avg_4x2_ssse3;<br>
+ if (!(cpu&X264_CPU_STACK_MOD4))<br>
+ {<br>
+ pf->avg[PIXEL_16x16] = x264_pixel_avg_16x16_ssse3;<br>
+ pf->avg[PIXEL_16x8] = x264_pixel_avg_16x8_ssse3;<br>
+ pf->avg[PIXEL_8x16] = x264_pixel_avg_8x16_ssse3;<br>
+ pf->avg[PIXEL_8x8] = x264_pixel_avg_8x8_ssse3;<br>
+ pf->avg[PIXEL_8x4] = x264_pixel_avg_8x4_ssse3;<br>
+ pf->avg[PIXEL_4x8] = x264_pixel_avg_4x8_ssse3;<br>
+ pf->avg[PIXEL_4x4] = x264_pixel_avg_4x4_ssse3;<br>
+ pf->avg[PIXEL_4x2] = x264_pixel_avg_4x2_ssse3;<br>
+ }<br>
<br>
pf->hpel_filter = x264_hpel_filter_ssse3;<br>
pf->frame_init_lowres_core = x264_frame_init_lowres_core_ssse3;<br>
<br>_______________________________________________<br>
x264-devel mailing list<br>
<a href="mailto:x264-devel@videolan.org" target="_blank">x264-devel@videolan.org</a><br>
<a href="http://mailman.videolan.org/listinfo/x264-devel" target="_blank">http://mailman.videolan.org/listinfo/x264-devel</a><br>
<br></blockquote></div><br>