Destroy wrote:
This is a copy and paste of an email I sent to ATI (which has gotten
no response so far, big surprise there), NOT!!

Anyone have any ideas or suggestions? thanks
================================================== ====

Hi,
Last week or so I was having to turn the computer off then back on
again 2 times for the machine to get into bios, post and boot. Now
this
morning I go to turn my computer on and it beeps a cpu error
everytime I try to to turn it on. Can't get it to post no matter what
I do. Disconnected everything but vid and memory but still my
computer just sits there fans going, harddrives going(when connected)
but no video, no post and beep msg saying there is a cpu problem.

I'm posting this with my old 9800pro which I had to stick back in and
works fine so obviously there is no cpu problem. I can swap my X800XT
and 9800pro all day and 9800pro allows my machine to post and boot
properly with no problems while my X800XT hangs my system before bios.

I'm wondering if its a power supply problem? I have a decent 400watt
PS though. And I don't have another laying around to test with, arg.

Bios reads: 1.53,3.26,4.94,11.85,12.36,2.91,5.56 for the various
voltages. SandraSoft reads:
1.53,3.25,4.95,11.92,-12.53,-5.15,5.02,2.91 voltages. (These are with
my 9800pro however. Can't read my X800 voltages cause I cant even get
to bios.)

Well I went a bought a quality Antec power supply to try - still
doesn't work, can't post, same cpu beep code.

Put 9800pro back in and going to drop vid card off at friends house
for them to put in their system to test card.

Seems I may have a card gone bad but won't know for sure.
The odd thing is, my computer was working perfectly fine for a week
with my new X800XT. After a week my machine started acting odd: on
cold boot I would for some reason have to turn my machine on then off
then on again to get it to boot into bios and continue normally. If I
didn't do this my computer would sit there beeping the cpu error
never getting
into bios.

Well just got a call from my friend who's testing out my X800XT for me
in his(2 month old) athlon64 3200+ rig and it booted and worked fine
in his computer. I'm having him do some a few cold boots before
giving me the card back. (It booted and worked fine each time.)

It's almost as though my computer slowly degraded over a week and a
half from perfect to funky cold booting error to can't boot at all. I
changed nothing on my system during this time besides perhaps a few
demo or program installs within winXP. I just don't get it.

Its just so odd that all works great with my 9800pro and I simply put
in the X800XT and blam, computer won't post.

Any suggestions would be helpful, thank you.
Btw, I consider myself VERY computer hardware and software savy so any
tech speak is fine.
Regards, Derek
================================================== ======

Edit: Since this email, I've borrowed my card to another different
friend to test the video card in his system (very new Prescott system)
and it works perfectly fine for him also.

I'm guessing I have a motherboard problem. It's as though some
component on my motherboard slowly burned up with the X800XT in my
system and then degraded far enough that it prevents my system from
booting up. However, at the same time something different about the
9800pro design doesn't require as much 'draw' from this degraded MB
component and allows my system to work seemingly normally. This sound
possible?


Albatron K8T800ProII, ADATA PC4000 1 gig, athlon64 3200+, nothing
overclocked

I've read your post and the replies in a.c.p.v.ati; I've recently had a
1 year old Gigabyte P4 motherboard do something similar to me. It
finally got so it wouldn't post at all. Even switching video cards
didn't help. Each of your busses have power regulation. On my GA8-iHXP
there was one that 'cooked out', and became unreliable. There are
actually 3 of these in parallel, so that enough current is provided to
the PCI bus and to the AGP bus (slot). One finally failed completely,
it was a 5V- regulator and supplies the 3.3v and 1.5v regulators. You
can lose part of your onboard voltage regulation and the system will be
non functional, reduce the load enough on the affected bus and the
system can appear reliable. But it will be just a matter of time and
that failing component will give out completely, as did mine.
The conditions made it look like it could be the power supply at first,
but that proved fine on another machine. Same for the CPU, video card,
sound card, ram DIMMs. I tried two different CPU's and both behaved the
same, a different PS and vid card, same. It was the motherboard.
You've proved the X800XT to be fine on two other systems. Your process
of elimination point firmly to the mobos onboard voltage regulation.

So now the mobo is shot, why? Well, the newer video cards DO require a
seperate power cable. BUT are the cards designed to really use that
external (to the AGP bus itself) power so that the load on the AGP bus
is actually alleviated? It doesn't look like it. It appears the fan
only is run from this external power cable. That's not helpful when the
bus' voltage regulator is designed for 1 amp max and the card itself
pulls almost an amp at 'idle'. It sounds to me that either the mobo
designers are skimping a bit on design criteria for the AGP version 3
bus (calls for 2 amps) or the vid card makers are allowing their boards
needs to push the 3.0 design specs over the boundaries. Or some of
each. Either way, I am seeing a *lot* of this kind of thing happening
with the widespread use of AGP3.0 specs. It's stacking up to be
something close to a 8% failure rate of motherboards because of this.
Hm. Now take note that the particular kind of failure is due only to
use of video cards with need for external power cable and at a guess,
failure rate is upwards of 20%. This is just me talking aloud with what
I've personally had to look at locally. No matter which way I paint it
now, it's something to definitely take notice of. And that we consumers
have seen and discussed it in public, you can bet the concerned
interests have noted it as well.

McG.