--- Steven Ellis <[EMAIL PROTECTED]> wrote:
> There is an article on PCI Latency at the MythTV > wiki - > http://www.mythtv.org/wiki/index.php/PCI_Latency - > that recommends using > setpci to increase the latency value of IDE devices > to be greater than > the value allocated to the ivtv devices. Has anyone > here tried this as a > possible fix for Trac issues 48 & 49. > > One reason I'm asking is I can't change these > latency values on my > A8N-VM CSM motherboard. I'm running a 2.6.15 kernel > with the very latest > release of pciutils. > > lspci -vd 10de:0265 > 00:0d.0 IDE interface: nVidia Corporation MCP51 IDE > (rev a1) (prog-if 8a > [Master SecP PriP]) > Subsystem: ASUSTeK Computer Inc. Unknown > device 81bc > Flags: bus master, 66MHz, fast devsel, > latency 0 > I/O ports at ffa0 [size=16] > Capabilities: [44] Power Management version > 2 > > setpci -v -G -d *:0265 latency_timer=b0 > Trying method 1......using /sys/bus/pci...OK > Decided to use Linux-sysfs > 00:0d.0:0d b0 > > lspci -vd 10de:0265 > 00:0d.0 IDE interface: nVidia Corporation MCP51 IDE > (rev a1) (prog-if 8a > [Master SecP PriP]) > Subsystem: ASUSTeK Computer Inc. Unknown > device 81bc > Flags: bus master, 66MHz, fast devsel, > latency 0 > I/O ports at ffa0 [size=16] > Capabilities: [44] Power Management version > 2 > > So no change but the system thinks it worked. > > I've tried the same command on other boxes with > different chipsets and > it appears to work, just fails on my Asus > motherboard. > > My hope is that by raising the latency value I can > avoid the > artefact/corruption issues I keep getting with the > card. > > Steve > > > _______________________________________________ > ivtv-devel mailing list > [email protected] > http://ivtvdriver.org/mailman/listinfo/ivtv-devel > I personally can't say that setting the latencies has helped me resolve any problems, but... I was running my MythTV backend on an Intel P4 2.4GHz with an Abit motherboard (865PE chipset) that didn't allow for setting the PCI latencies on the IDE interfaces. I found that board was giving me IVTV DMA errors when I was capturing on my PVR-500. Most of my hard drives are actually connected to PCI cards - I have a Silicon Image 3114 for a RAID-5 and a Promise TX-4 for a backup drive and the MythTV drive. Hence, I could almost get away without using the IDE ports, and set the latencies on the controller cards in their slots. (I can't get away without using the IDE though, since one of the drives in the RAID-5 is also my boot drive, and it's IDE.) Since then, I dragged an Athlon XP 2400+ out of the closet running a SiS 748 chipset (it's a Gigabyte board). This board allows for my IDE interfaces to have their latency set by setpci (which I have done). Overall, it seems to perform much better, I have had no IVTV DMA errors at all. So, while I'm not sure that the two issues are related, I would suspect that perhaps the boards that allow for the PCI latencies to be set are better designed. Now that I think about it, it's funny - both my frontend and backend are both Sis chipsets (the backend, a SiS748, the frontend a SiS755). My random two cents worth, Joe __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com _______________________________________________ ivtv-devel mailing list [email protected] http://ivtvdriver.org/mailman/listinfo/ivtv-devel
