[MEncoder-users] Encoding stops with '00 motion_type at 17 6' after 990 sec

Kasper Bonne kbonne at gmail.com
Wed Mar 31 16:10:21 CEST 2010


On Wed, Mar 31, 2010 at 02:13, Christian. <cmf.ml at gmx.de> wrote:
> I hope this is what you asked for. After installation and compilation of the
> debian sources I get,
[snip]
> (gdb) backtrace
> #0  0x0863c8a2 in hScale_MMX2 (dst=Cannot access memory at address 0xfffffa68
> ) at swscale_template.c:2053
> Cannot access memory at address 0xfffffa64

Mmm, hard to say for sure what the problem is but it looks to me like
a hardware acceleration problem.

Try adding -zoom to your command line (enables software scaling). If
that alone doesn't work, you can select a different software scaling
algorithm using -sws. The function where things go wrong seems to be
related to bilinear / bicubic scaling so using another method might
bypass the problem.

If I'm right and the problem is related to hardware acceleration, it
could even be that updating the drivers for you graphics card would do
the trick.

/Kasper


More information about the MEncoder-users mailing list