Tom Brinkman grabbed a keyboard and wrote:
>
> Look at 'dmesg' you should see a line like this,
> "Checking 'hlt' instruction... OK."
Yea, I've seen those.
> When you go into bios, those
> idle statements aren't being sent to the cpu, and neither is the cpu
> under much if any load. Plus, even if the systems only been off for a
> minute or so, the heatsink's cooled down.
I may reboot (to get to the BIOS or whatever), but I don't turn it off
unless I'm pulling chips or whatever. Of course, I've been doing a *lot*
of that, lately.... :-)
> BIOS does get it's info from the same place 'sensors' does, the
> Read /usr/share/doc/lm_sensors-2.6.2/doc/FAQ and the rest of the
> docs that come with lm_sensors. It should be as easy as runnin
> 'sensors-detect' and putting a few lines it generates into rc.local
> and modules.conf With ML8.2 I've found I also needed to put
> 'i2c-proc' in /etc/modules, eg,
That did it. :-) Running sensors-detect set up whatever was needed to get
"sensors" to show me something. I'll read the FAQ and other documents so
that I can understand just what the heck I did. :-) Gkrellm now has
Builtins | Sensors available to me, and I've turned on the voltage
monitors. I put CPU in the label field next to a temprature, but I'm not
seeing anything showing up in the gkrellm display. Hopefully, I'll be able
to figure out whatever it is that's not working here....
> > DIMMs one at a time and found that when one particular DIMM is in
> > the system, it won't start. When it's out, the machine will come
> > up fine. (And yes, I moved one of the other modules to the slot
> > that the troublemaker was in, to make sure it wasn't the slot. :)
>
> Well good for you, hopefully you've found the problem. I'm
> surprised memtest86 didn't find it,
As am I. I've had really good results with that in the past, which is why
this is so surprising for me. I'll have a better feel for if this actually
was the problem after I leave my system up for a few days. If it doesn't
hang, the 550MHz chip goes back in and I give it another week. :-)
> but I betcha cpuburn would'a ;)
It may well have. But after what I've read about the program, there's no
way in hell I was going to run it without a way to actively monitor my CPU
temp while it was running. :-) Running "sensors" now shows temp1 as 27
degrees C, and temp2 & temp3 as 32 degrees C. (Am I correct in assuming
that temp1 is the motherboard temp?) So if I can figure out how to get
those values displayed in gkrellm, I'm good to go.
Ok, I just figured out the last part. Apparently, it's case sensitive. :-)
"CPU" won't make it show up, but "cpu" will. Go figure... <grin>
> I still suggest you get cpu monitoring going tho. Besides temps, you
> should see steady voltage values very close to spec or slightly over
> spec. On good quality mobo's like your Epox, hopefully you'll see
> I/O values slightly higher than 3.3 volts. 3.4 to 3.7 is good, less
> than 3.3 is very bad. This is the power to the ram, and a little
> extra goes along way towards providing stability.
My +3.3v display is currently showing 3.55, so that sounds good. :-)
BTW, when running sensors-detect, it mentioned that you didn't need to do
certain modprobe commands if those modules were already built in to the
kernel. Is there an easy way to determine if a given module is compiled
into the kernel?
--Dave
--
David Guntner GEnie: Just say NO!
http://www.akaMail.com/pgpkey/davidg or key server
for PGP Public key
Want to buy your Pack or Services from MandrakeSoft?
Go to http://www.mandrakestore.com
