David Goodenough said:
Any ideas what might cause this problem?
I have an A7V8X with a Seperon 2500+ in it, and every time I
completely disconnect the power, on next boot it claims it's not sure
about the CPU speed, and stops in the BIOS setup. I dutifully set it
back to 1750, and all is well till it powers down again.
I thougth it was the CMOS battery, but a brand new one hasn't helped.
Any ideas what else I can do?
Thanks in advance.
How bad do you want to fix this
???
It could be a sign of a cold boot problem. The next time you
power off the computer, try starting it immediately, and see
if the settings hold. (In other words, don't give the processor
time to cool off.)
When a computer starts, it will likely start twice. What I
mean by that, is when power is first applied, the motherboard
applies the "default" values, as they come from the pins of
the processor. Then, using BIOS code, and special hardware on
the board (like the Attansic ATXP3 overclock controller), it
is possible for a new set of conditions to be applied to the
processor. The processor does a mini-reset, and the new
conditions are applied to it (the "users settings").
Somewhere during this process, it could be that your processor
is crashing, and the BIOS ends up applying "safe" conditions,
thus erasing your clock setting.
I have no idea how much of the functionality in here,
that Asus is using. This just hints at the capabilities.
(The clockgen is similarly programmable, via the SMBUS. Your
CPU FSB speed is changed via the clockgen chip.) I'm not
even sure of the algorithm above - it is just a hunch.
This is the Attansic overclock controller, with watchdog
timer for detecting crashes.
http://www.attansic.com.tw/english/products/pdf/ATXP3_2.2.pdf
How to fix it ? A good question. It likely isn't crashing
during the application of the user's desired settings, as
otherwise it would be crashing on a warm restart too. It
really does seem like the defaults at reset are not right
- try backing off on the memory timing, and see if the symptoms
change. I.e. If the memory has 2-2-2-5 timings, try 3-3-3-8,
save, power down. Let it cool off, and try it, and see if there
is any difference in the symptoms.
You could also experiment with the Over_Volt1 jumper. My
theory here is, the Over_Volt1 jumper applies a voltage
boost, even right at the moment that power is first applied
to the board. If that is true, it could be that enabling
Over_Volt1 will cure it.
I think this is your processor. See PDF page 33 (table 1).
http://www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/31693.pdf
Your processor has a nominal Vcore of 1.6V. Now, when the
processor is running normally, you only want as much Vcore
as is needed to keep the processor stable. If, in fact, the
Over_Volt1 jumper was applying 1.8V, that would make the
processor 27% hotter. Ideally, you would want to go into
the BIOS, and using the manual setting, try to set it
back to close to 1.6V, its normal value.
I'm sure I'm not being clear enough here. There is the time
the "default" VID is being applied, and then there is the
time when the "users BIOS setting" VID is being applied.
The "default" VID _should_ be 1.6V (as set by the processor
VID pins) plus the boost of 0.2V from the Over_Volt1 jumper.
Once the BIOS loads the user's desired VID value, then that
value could be turned down a bit (to get it closer to the
1.6V that was being used before). You'll know what to set
the BIOS Vcore value to, by observing what range of values
is listed for that parameter (whether the BIOS value reflects
the setting of the Over_Volt1 jumper or not).
You could always attempt some bridge hacking or wire modding,
to fix this, but is it really worth it ? That stuff is
getting kinda old... And I'd have to write a few more
paragraphs for that
I've had similar problems with my A7N8X-E, and while I remember
few of the details now, I was able to fix it by tweaking the
BIOS settings. In that case, the only symptoms were some
annoying Vocal POST messages at startup.
HTH,
Paul
