[mythtv-users] HD-3000 recording corruption

Brian Wood beww at beww.org
Wed Mar 29 18:37:27 UTC 2006

On Mar 29, 2006, at 10:07 AM, Craig Partin wrote:

>> Take a look at PCI Latency on the WiKi.
>> http://www.mythtv.org/wiki/index.php/PCI_Latency
>> In particular look at this link in the "credits" area.
>> http://www-128.ibm.com/developerworks/library/l-hw2.html
>> This sounds like a classic case of feeding data to the PCI bus  
>> faster than
>> the system can write it to the drives. You'll probably want to  
>> increase the
>> latency of your IDE subsystem if you can.
> I think the motherboard has nvidia chips for all the bus and IO stuff.
>  It's an nforce chipset.  I'll check for certain though.
> I have changed the PCI latency like the Robbins article suggests with
> no effect on the corruption issue.  I haven't tried to up the latency
> of the entire PCI controller but read the big thread and noticed that
> mine is one of the boards where it is set to 0.

Both of my nForce motherboards (an Asus and a Gigabyte) have a BIOS  
adjustment for latency, you might try playing with that in  
conjunction with setting values from within the OS (setpci).

Sure sounds like you have all the bases covered though. You mentioned  
you changed PSUs, if you simply tried another of the same type that  
might not have proven anything, as you might have an inherent problem  
and not a "failure". I saw a power supply advertised the other day as  
"600-watt power supply for $10.99", I don't think I would risk such a  
thing in any machine of mine :-)

Other than that I don't know, let us know when you do get it sorted  
out, even if it does require a new MB.

More information about the mythtv-users mailing list