Has anyone else seen this post on the new version of the Nehemiah/M10k chipset?
http://forums.viaarena.com/messageview. ... SGDBTABLE=
Having checked my own cpu, it looks like I've got one of the newer versions - which explains a lot about the performance issues I've been having...
I've modified the myth source to ensure the correct cpu info is reported and as a result my frontend <seems> to be a little zippier. Anyone else tried it and had success?
Regards
Mark
New chipset - M10k performance
Moderator: Pablo
Markk,
For once something is working in my favour ...
processor : 0
vendor_id : CentaurHauls
cpu family : 6
model : 9
model name : VIA Nehemiah
stepping : 1 <<<
BTW was that you I responded to on the MythTV mailing list about Ivors fix to the via slice?
Have you tried it ?
Its working really well here.
Regards,
Darren
For once something is working in my favour ...
processor : 0
vendor_id : CentaurHauls
cpu family : 6
model : 9
model name : VIA Nehemiah
stepping : 1 <<<

BTW was that you I responded to on the MythTV mailing list about Ivors fix to the via slice?
Have you tried it ?
Its working really well here.
Regards,
Darren
Yes - it was me you responded to. The patch works very well for me too - once I'd spotted the effect it was extremely annoying and the improvement was obvious.
(Are you using RCA or S-Video out? - there is a definite edge effect/blur with RCA and am wondering whether S-Video is any better? - would need a new tv though...)
Regards
Mark
(Are you using RCA or S-Video out? - there is a definite edge effect/blur with RCA and am wondering whether S-Video is any better? - would need a new tv though...)
Regards
Mark
As per my other post, I've set up a new minimyth frontend this week with a new case and new M10k motherboard.
First thing I did was check the cpuinfo - and it looks like I've got another (newer?) version...
processor : 0
vendor_id : CentaurHauls
cpu family : 6
model : 9
model name : VIA Nehemiah
stepping : 5 <<<<
Not potentially very interesting, but I did a couple of comparisons between the old box and new. Watching the same livetv programs, my old motherboard runs at around 40% CPU using hardware decoding and the new one at around 25-30%, so the changes don't appear to be cosmetic.
Additionally - and this is the really nice bit - the quality of the tv out is much better. DVD playback is on a par with my player and it's difficult to tell the difference between normal tv and myth now. Still a slight fuzziness, but you'd not notice it without looking hard. This could be due to the case - my old case was a DIY aluminium effort (
First thing I did was check the cpuinfo - and it looks like I've got another (newer?) version...
processor : 0
vendor_id : CentaurHauls
cpu family : 6
model : 9
model name : VIA Nehemiah
stepping : 5 <<<<
Not potentially very interesting, but I did a couple of comparisons between the old box and new. Watching the same livetv programs, my old motherboard runs at around 40% CPU using hardware decoding and the new one at around 25-30%, so the changes don't appear to be cosmetic.
Additionally - and this is the really nice bit - the quality of the tv out is much better. DVD playback is on a par with my player and it's difficult to tell the difference between normal tv and myth now. Still a slight fuzziness, but you'd not notice it without looking hard. This could be due to the case - my old case was a DIY aluminium effort (
I have a M10k with a stepping 5 CPU. According to this post to mythtv-users, stepping 5 has the same bug:
http://www.gossamer-threads.com/archive ... e_P116652/
On my box (MiniMyth 0.4, playing MPEG4) I don't appear to have any functional problems (video plays smoothly), but it uses about 80% of the CPU to play a 480x480 MPEG4 stream.
0. Does this CPU usage seem reasonable or not?
1. Can anyone confirm if stepping 5 has the bug or not?
2. Does minimyth have the work-around already compiled into it?
thanks,
andrew
http://www.gossamer-threads.com/archive ... e_P116652/
On my box (MiniMyth 0.4, playing MPEG4) I don't appear to have any functional problems (video plays smoothly), but it uses about 80% of the CPU to play a 480x480 MPEG4 stream.
0. Does this CPU usage seem reasonable or not?
1. Can anyone confirm if stepping 5 has the bug or not?
2. Does minimyth have the work-around already compiled into it?
thanks,
andrew
Andrew
I'm not too sure whether 80% cpu is reasonable for MPEG4 as I've never used MPEG4 with Myth.
I haven't investigated fully, but my feeling is that the newest, stepping 5 version does not have the same problem. Performance definitely outstrips the stepping 3 version - but this could be due to other changes they've introduced that haven't yet been spotted. My '5' frontend is almost perfect on my main tv - occasionally losing audio sync after a few channel changes/use of OSD and use of the EPG. The '3' frontend has noticeable stutter problems playing back recordings and loses audio sync more readily.
That said, I also believe there is a bug somewhere in myth. I can seen no reason why video playback should have problems while livetv is always perfect.
I'm going to update to CVS today sometime and start some digging...
Regards
Mark
I'm not too sure whether 80% cpu is reasonable for MPEG4 as I've never used MPEG4 with Myth.
I haven't investigated fully, but my feeling is that the newest, stepping 5 version does not have the same problem. Performance definitely outstrips the stepping 3 version - but this could be due to other changes they've introduced that haven't yet been spotted. My '5' frontend is almost perfect on my main tv - occasionally losing audio sync after a few channel changes/use of OSD and use of the EPG. The '3' frontend has noticeable stutter problems playing back recordings and loses audio sync more readily.
That said, I also believe there is a bug somewhere in myth. I can seen no reason why video playback should have problems while livetv is always perfect.
I'm going to update to CVS today sometime and start some digging...
Regards
Mark