[mythtv-users] does setpci latency_timer fail on some mobos?

Mark A. Miller mamillerpa at yahoo.com
Thu Jun 7 21:37:04 UTC 2007

Thanks for the feedback, Craig (and other responders.)  I had wanted to
keep my original post short, so I didn't mention that I have built
several MythDora systems, and all but one refused to change the SATA or
IDE latency values.  The ones that didn't work included Asus and ECS
nForce mobos.  I may return the board I have now and try a MSI board
with an ATI chipset.


--- Craig Huff <huffcs at yahoo.com> wrote:

> I had seen notes on the list about latency and Mark's question
> finally got me curious enough to look into it, so I found the wiki
> page about latency.
> Now I suspect unadjusted latency may be the reason I've been getting
> IOBOUND errors, seeing buffer overflow messages in /var/log/messages,
> etc.
> Unfortunately, it seems I'm in the same boat as Mark!
> I have an Asus A8N-SLI Premium mobo with three built-in ATA
> controllers -- 1 PATA, 2 SATA -- and all three (and another device
> that escapes me right now) won't change latencies.  They are all
> stuck at zero (0) !!!
> I did verify that I was doing the commands right by changing the
> latency on one of my PVR cards (from 64 to 0x64=100) and back.  In
> the process I learned that the value setpci takes is assumed hex and
> the value displayed by lspci is decimal ;-).
> Seems that Asus mobos have a common thread.  It is as though they put
> the PCI parameters for built-in devices in ROM and we're stuck with
> whatever they chose.  In my case, they appear to be the worst
> possible values for disk controllers -- at least that is what it
> seems like to me.

More information about the mythtv-users mailing list