On Friday 06 Feb 2004 11:19 am, John Richard Smith wrote: > Hi Robert, > > Yes OK then, have done that, > Incidentally the replacement is a new straight out of the box 350W, > nothing special, as to make, > just kept as a spare for just such test purposes. > Unforetunately it makes no difference at all. > The first cpu was a quite new item in itself > and the reason I didn't suspect it. > I don't really suspect the PSU. > Always open to correction though. > > I still have instability in high load Mandrake cpu situations. > I have to underclock the FSB rate to achieve stability in M9.1, > not in W2K though, and so I get ribbed by my family, > I'm seen as a Linux geek to everyone around me. > Though I noticed one of my daughters has recently started using > xcdroast a lot , say's it works great. Oh well, great things from > small beginnings, do proceed. You know it's a confidence thing ! > But I digress. > Yesterday I reseated the heatsink for the 'n'th time. > This time I deliberately bent the retaining clip so as to exert > more retention thrust onto the cpu, in otherwords the > it took more of a push to retain the clip to the cpu. > Current results are:- > > FSB100 > ====== > startup/desktop Full CPU load > cpu 34C/41C 49C > Case25C/30C 35C > fan 3624 3668 > Vc 1.74 1.74 > Vh 1.25 1.24 > Vp 3.315 3.37 > 4.85 4.84 > 12.3 12.3 > -12.40 -12.40 > -5.07 -5.07 > These are not bad temperatures to my mind. > The full load(99%) is achieved by running a > mencoder job rather than mprime. > Seems like clip retention pressure makes a difference > But no need to exagerate it ! > > I am currently unable to attain desktop in FSB133 > I get boot messages, > ide task - not syncing > stack ( loads of number) > calltime > kernel panic, killing interupts handler-not syncing > > Food for thought there don't you think ? > CPU temp is well within tollerances now > and was withing AMD's own tollerances anyway. > PSU probably fine. > I'm open minded about memory, > though it will pass the memtest. > I haven't the faintest idea what syncing is about ? > Guess that is the next thing to find out about. > > John
Sorry John, I didn't realise that Windows was running OK... (I presume it runs OK at full CPU load.) Have you tried running a different distro (say Koppix from CD)? Is ACPI enabled? Try switching this setting on. What kernel are U using? I presume it is 2.4.x series :-) Is it stock or custom? You might try a custom kernel with ACPI statically linked and APM left out completely (APM can potentially clash - although I think this problem is fixed now in the 2.4.x kernel). Does <dmesg> have anything useful in it? Maybe you could post it here? Whats lurking in your /proc/ directories (interrupts, pci, etc.)? I am just clutching a straws here but it does sound like the Linux interrupt hander is shagged. So first place to check is ACPI settings I guess... Robert
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com