8KNS high CPU temperatures

I have a 8KNS and have no overclocking. It is a lightweight system with an
old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at DDR 400
dual channel. The temperatures are stable (38 to 42 C) until I do anything
graphical. Then the temperature will go slowly up to the limit, which I set
to 65 C. This happens with any kind of game, even the old Hasbro kid games.
I have replaced the heatsink. I have used Arctic Silver regular and
Ceramique. I have lowered the Vcore in stages down to its present 1.32. I
have turned ECC on and off. The power supply is an Antec 350, plugged in to
a Belkin 1200 AVR UPS. I can run for days and days, doing spreadsheets,
surfing the web, etc - anything except a game of any kind (not including
Solitaire, which runs trouble free). I think that the problem is due to one
of three things: (1) flakey CPU, (2) faulty mainboard, (3) some kind of
issue with the old Diamond TNT32 Ultra. If anybody has ever experienced
anything similar and found a workaround, I would be most grateful in
learning your solution.

Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...
I have a 8KNS and have no overclocking. It is a lightweight system with an
old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at DDR
400 dual channel. The temperatures are stable (38 to 42 C) until I do
anything graphical. Then the temperature will go slowly up to the limit,
which I set to 65 C. This happens with any kind of game, even the old
Hasbro kid games. I have replaced the heatsink. I have used Arctic Silver
regular and Ceramique. I have lowered the Vcore in stages down to its
present 1.32. I have turned ECC on and off. The power supply is an Antec
350, plugged in to a Belkin 1200 AVR UPS. I can run for days and days,
doing spreadsheets, surfing the web, etc - anything except a game of any
kind (not including Solitaire, which runs trouble free). I think that the
problem is due to one of three things: (1) flakey CPU, (2) faulty
mainboard, (3) some kind of issue with the old Diamond TNT32 Ultra. If
anybody has ever experienced anything similar and found a workaround, I
would be most grateful in learning your solution.

On Fri, 25 Nov 2005 23:46:45 GMT, "John McDonnell"
wrote:

I have a 8KNS and have no overclocking. It is a lightweight system with an
old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at DDR 400
dual channel. The temperatures are stable (38 to 42 C) until I do anything
graphical. Then the temperature will go slowly up to the limit, which I set
to 65 C. This happens with any kind of game, even the old Hasbro kid games.
I have replaced the heatsink. I have used Arctic Silver regular and
Ceramique. I have lowered the Vcore in stages down to its present 1.32. I
have turned ECC on and off. The power supply is an Antec 350, plugged in to
a Belkin 1200 AVR UPS. I can run for days and days, doing spreadsheets,
surfing the web, etc - anything except a game of any kind (not including
Solitaire, which runs trouble free). I think that the problem is due to one
of three things: (1) flakey CPU, (2) faulty mainboard, (3) some kind of
issue with the old Diamond TNT32 Ultra. If anybody has ever experienced
anything similar and found a workaround, I would be most grateful in
learning your solution.

Sounds like it could be DirectX or an AGP x2/x4 compatibility problem.
I have a 7400S-L which will only work with x4 or x8.I am sure your TNT
is x2 or x1. x2 runs at 3.3v, x4 and up at 1.6v. Check that your board
is dual standard. Either way, I am sure you will find the card is
not compatible, most likely with DX9. You can find a DX9 AGP x8 card
cheap, but if you want to keep your TNT, try an older version of
DirectX

"John McDonnell" wrote in message
news%Nhf.30$xQ3.12@trndny04...
Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...
I have a 8KNS and have no overclocking. It is a lightweight system with
an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C) until I do
anything graphical. Then the temperature will go slowly up to the limit,
which I set to 65 C. This happens with any kind of game, even the old
Hasbro kid games. I have replaced the heatsink. I have used Arctic Silver
regular and Ceramique. I have lowered the Vcore in stages down to its
present 1.32. I have turned ECC on and off. The power supply is an Antec
350, plugged in to a Belkin 1200 AVR UPS. I can run for days and days,
doing spreadsheets, surfing the web, etc - anything except a game of any
kind (not including Solitaire, which runs trouble free). I think that the
problem is due to one of three things: (1) flakey CPU, (2) faulty
mainboard, (3) some kind of issue with the old Diamond TNT32 Ultra. If
anybody has ever experienced anything similar and found a workaround, I
would be most grateful in learning your solution.



I think actually bumping the core voltage UP a notch is better than lowering
it for keeping it running cooler. Is your CPU fan running? It just sounds
to me like the cooling solution isn't enough for what you're doing with that
cpu. you mentioned replacing the heatsink, what about the CPU fan? And,
are you SURE that the heat sink is mounted *flat* on top of the CPU?
Something seems to not be doing it's job. I don't think it's a flakey A64
3000+ or TNT2 Ultra. When the CPU is stressed under load, that heatsink
and fan have to take the heat the cpu gives off away. That's not happening,
apparently. Poor contact of hs to cpu, poor air circulation in the case,
fan and/or hs inadequate or faulty fan.
McG.

Thanks for the reply. Yes, standard thinking says to raise the CPU voltage
and not lower it. The normal CPU voltage for this motherboard is 1.40. I
had first tried a setting of 1.45. After restart, I went into the Bios
setup and brought up the PC Health screen and watched it for a while. The
temperature went up quickly 1 degree at a time from 38 to 65, at which time
I shut down and set it back to 1.32. The PC Health screen then stayed
between 35 and 40 for quite a while. If I use the default of 1.40, the CPU
temperature will still rise to the danger level, but much more slowly.

The heatsink is definitely seated OK, the Arctic Silver was applied
properly, and the fan runs steady about 3000 rpm. I'm pretty experienced in
that area. The circulation in the case is not the best, but removing the
side panel makes very little difference. Running non-graphical apps causes
no problems. While writing this email I have watched the temperature
monitor slowly drop from its initial 38 to its present 34. But if I should
start up any kind of game, it will slowly creep up to past 60, at which time
I will shut down.

I am still suspicious of the TNT2 board. It is an older model - Diamond
Viper TNT32 Ultra (one of the hottest boards around when I purchased it in
early 1999) with a jumper for setting AGP 2X / 4X. I say this because the
system operates trouble-free except when running graphics, even kiddie
games. I built this box about six months ago and used it regularly
(internet browsing, VPN, mail, worksheets, etc.) with no problem. A couple
of weeks ago I loaded some Hasbro kiddie games for my grandson and that's
when things began to happen.

"McGrandpa" wrote in message
...

"John McDonnell" wrote in message
news%Nhf.30$xQ3.12@trndny04...
Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...
I have a 8KNS and have no overclocking. It is a lightweight system with
an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C) until I
do anything graphical. Then the temperature will go slowly up to the
limit, which I set to 65 C. This happens with any kind of game, even the
old Hasbro kid games. I have replaced the heatsink. I have used Arctic
Silver regular and Ceramique. I have lowered the Vcore in stages down to
its present 1.32. I have turned ECC on and off. The power supply is an
Antec 350, plugged in to a Belkin 1200 AVR UPS. I can run for days and
days, doing spreadsheets, surfing the web, etc - anything except a game
of any kind (not including Solitaire, which runs trouble free). I think
that the problem is due to one of three things: (1) flakey CPU, (2)
faulty mainboard, (3) some kind of issue with the old Diamond TNT32
Ultra. If anybody has ever experienced anything similar and found a
workaround, I would be most grateful in learning your solution.



I think actually bumping the core voltage UP a notch is better than
lowering it for keeping it running cooler. Is your CPU fan running? It
just sounds to me like the cooling solution isn't enough for what you're
doing with that cpu. you mentioned replacing the heatsink, what about the
CPU fan? And, are you SURE that the heat sink is mounted *flat* on top of
the CPU? Something seems to not be doing it's job. I don't think it's a
flakey A64 3000+ or TNT2 Ultra. When the CPU is stressed under load,
that heatsink and fan have to take the heat the cpu gives off away.
That's not happening, apparently. Poor contact of hs to cpu, poor air
circulation in the case, fan and/or hs inadequate or faulty fan.
McG.

"John McDonnell" wrote in message
news:vPgkf.2357$1y.2302@trnddc07...

"McGrandpa" wrote in message
...

"John McDonnell" wrote in message
news%Nhf.30$xQ3.12@trndny04...
Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...

John said:
I have a 8KNS and have no overclocking. It is a lightweight system with
an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C)
until I do anything graphical. Then the temperature will go slowly up
to the limit, which I set to 65 C. This happens with any kind of game,
even the old Hasbro kid games. I have replaced the heatsink. I have
used Arctic Silver regular and Ceramique. I have lowered the Vcore in
stages down to its present 1.32. I have turned ECC on and off. The
power supply is an Antec 350, plugged in to a Belkin 1200 AVR UPS. I
can run for days and days, doing spreadsheets, surfing the web, etc -
anything except a game of any kind (not including Solitaire, which runs
trouble free). I think that the problem is due to one of three things:
(1) flakey CPU, (2) faulty mainboard, (3) some kind of issue with the
old Diamond TNT32 Ultra. If anybody has ever experienced anything
similar and found a workaround, I would be most grateful in learning
your solution.



McG. said:
I think actually bumping the core voltage UP a notch is better than
lowering it for keeping it running cooler. Is your CPU fan running? It
just sounds to me like the cooling solution isn't enough for what you're
doing with that cpu. you mentioned replacing the heatsink, what about
the CPU fan? And, are you SURE that the heat sink is mounted *flat* on
top of the CPU? Something seems to not be doing it's job. I don't think
it's a flakey A64 3000+ or TNT2 Ultra. When the CPU is stressed under
load, that heatsink and fan have to take the heat the cpu gives off away.
That's not happening, apparently. Poor contact of hs to cpu, poor air
circulation in the case, fan and/or hs inadequate or faulty fan.
McG.



John said:
Thanks for the reply. Yes, standard thinking says to raise the CPU
voltage and not lower it. The normal CPU voltage for this motherboard is
1.40. I had first tried a setting of 1.45. After restart, I went into
the Bios setup and brought up the PC Health screen and watched it for a
while. The temperature went up quickly 1 degree at a time from 38 to 65,
at which time I shut down and set it back to 1.32. The PC Health screen
then stayed between 35 and 40 for quite a while. If I use the default of
1.40, the CPU temperature will still rise to the danger level, but much
more slowly.

The heatsink is definitely seated OK, the Arctic Silver was applied
properly, and the fan runs steady about 3000 rpm. I'm pretty experienced
in that area. The circulation in the case is not the best, but removing
the side panel makes very little difference. Running non-graphical apps
causes no problems. While writing this email I have watched the
temperature monitor slowly drop from its initial 38 to its present 34.
But if I should start up any kind of game, it will slowly creep up to past
60, at which time I will shut down.

I am still suspicious of the TNT2 board. It is an older model - Diamond
Viper TNT32 Ultra (one of the hottest boards around when I purchased it in
early 1999) with a jumper for setting AGP 2X / 4X. I say this because the
system operates trouble-free except when running graphics, even kiddie
games. I built this box about six months ago and used it regularly
(internet browsing, VPN, mail, worksheets, etc.) with no problem. A
couple of weeks ago I loaded some Hasbro kiddie games for my grandson and
that's when things began to happen.

McG. says:
Heh! You DID say "Viper TnT-32 Ultra" That was the immediate
predecessor to the Diamond Viper V770 Ultra, an AGP 2x/4x TnT2 Ultra 32 meg
card. Which I bought on July 1, 2001 for myself. My 47th birthday
That card is still running fine in an ABit BF6 (440BX) motherboard with a
P3-450 and 768 megs ram, which I gave to my youngest daughter about four
years ago. eh Hem... I digress

You're right John, your card is an oldie, but should be a goodie. With
Diamonds reputation being worse than ATI's old one with drivers and
stability, I'm astounded at the utter reliability of these two ancient
cards.
Neither have a temp sensor. But you can use the auxilliary thermal diode
that comes with many mobos to give you some indication of what's going on
with the GPU. You have clear indication of what's going on with your CPU
temp wise. I think I can see what's driving your thinking to the GPU.
Sitting down here in Houston with a different perspective, I can't see the
video card causing a hardware proplem to so affect the CPU in such a direct
manner.

Is this your mobo? The GA-K8NS?
http://www.giga-byte.com/MotherBoard...ec_GA-K8NS.htm
Socket 754 with an s754 A64-3000+?

Ok, I don't see or recall reading anything at all problematic with this mobo
or the CPU, even with standard retail boxed hs/fan. Makes for a nice
little rig, decent gamer too. Some folks doing some oc'ing with it.
Ok, I see a couple possibilities to look at. 1) My V770Ultra has a single
jumper on it to force the 1x/2x OR 4x for the AGP. Modern motherboards
like ours might be able to run that card at AGP 2X BUT that voltage is 3.3v
and your mobo specs say 1.5v only. It would have to be AGP4X period. Hm.
Ok. So you have an old vid card that any of the 3D games are beating up
on. Ok, that still should not drive the CPU temp up. What could drive it
up would be the CPU doing most of the graphics intensive work that a newer
vid card would normally do itself under DirectX 7 or higher.
The CPU is heavy loaded and undercooled. That's what I think is going on.
Perhaps a more modern video card would help a little bit, but where things
get real CPU intensive the temp is still going to go up. Only ways to test
this would be another newer type video card and then a better hs/fan than
you have on the CPU now.
It's not a 'little' problem, not when your CPU temp goes all the way to core
shutdown.
Hope something here helps!
McG.

Once again I really appreciate the advice. The mainboard is the
K8NS-939-Ultra (nForce3 ultra). The video card is the Diamond 770 Ultra
TNT32. The CPU is a 3000+ Winchester. I have ordered a Rosewill nVidia
6200 256mb AGP card from Newegg. I will certainly report back the results
to the group.

As long as I run no 3D apps and set the Vcore to 1.325, the system is fine.
But my grandson likes his Hasbro et al games. Most recently he was playing
Monopoly (not the most fast-moving game around, but 3D nonetheless) and the
system shut down and would not come back up. I got the loud single beep
every so many seconds. According to the manual, this indicates a dram
issue. As to whether this could be a video dram issue and not a mainboard
dram issue, the book didn't say either way. And it's no good to leave the
box shut down for a short time and start it up again. It must be off for at
least an hour. Very strange.

I have to eliminate components one at a time - improper seating of boards
and/or chips, video incompatibility, ram failure, insufficient cooling,
defective CPU, motherboard failure. And I'm starting with the video card.

Thanks, John
"McGrandpa" wrote in message
...
"John McDonnell" wrote in message
news:vPgkf.2357$1y.2302@trnddc07...

"McGrandpa" wrote in message
...

"John McDonnell" wrote in message
news%Nhf.30$xQ3.12@trndny04...
Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...

John said:
I have a 8KNS and have no overclocking. It is a lightweight system
with an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC
running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C)
until I do anything graphical. Then the temperature will go slowly up
to the limit, which I set to 65 C. This happens with any kind of game,
even the old Hasbro kid games. I have replaced the heatsink. I have
used Arctic Silver regular and Ceramique. I have lowered the Vcore in
stages down to its present 1.32. I have turned ECC on and off. The
power supply is an Antec 350, plugged in to a Belkin 1200 AVR UPS. I
can run for days and days, doing spreadsheets, surfing the web, etc -
anything except a game of any kind (not including Solitaire, which runs
trouble free). I think that the problem is due to one of three things:
(1) flakey CPU, (2) faulty mainboard, (3) some kind of issue with the
old Diamond TNT32 Ultra. If anybody has ever experienced anything
similar and found a workaround, I would be most grateful in learning
your solution.



McG. said:
I think actually bumping the core voltage UP a notch is better than
lowering it for keeping it running cooler. Is your CPU fan running?
It just sounds to me like the cooling solution isn't enough for what
you're doing with that cpu. you mentioned replacing the heatsink, what
about the CPU fan? And, are you SURE that the heat sink is mounted
*flat* on top of the CPU? Something seems to not be doing it's job. I
don't think it's a flakey A64 3000+ or TNT2 Ultra. When the CPU is
stressed under load, that heatsink and fan have to take the heat the cpu
gives off away. That's not happening, apparently. Poor contact of hs to
cpu, poor air circulation in the case, fan and/or hs inadequate or
faulty fan.
McG.



John said:
Thanks for the reply. Yes, standard thinking says to raise the CPU
voltage and not lower it. The normal CPU voltage for this motherboard is
1.40. I had first tried a setting of 1.45. After restart, I went into
the Bios setup and brought up the PC Health screen and watched it for a
while. The temperature went up quickly 1 degree at a time from 38 to 65,
at which time I shut down and set it back to 1.32. The PC Health screen
then stayed between 35 and 40 for quite a while. If I use the default of
1.40, the CPU temperature will still rise to the danger level, but much
more slowly.

The heatsink is definitely seated OK, the Arctic Silver was applied
properly, and the fan runs steady about 3000 rpm. I'm pretty experienced
in that area. The circulation in the case is not the best, but removing
the side panel makes very little difference. Running non-graphical apps
causes no problems. While writing this email I have watched the
temperature monitor slowly drop from its initial 38 to its present 34.
But if I should start up any kind of game, it will slowly creep up to
past 60, at which time I will shut down.

I am still suspicious of the TNT2 board. It is an older model - Diamond
Viper TNT32 Ultra (one of the hottest boards around when I purchased it
in early 1999) with a jumper for setting AGP 2X / 4X. I say this because
the system operates trouble-free except when running graphics, even
kiddie games. I built this box about six months ago and used it
regularly (internet browsing, VPN, mail, worksheets, etc.) with no
problem. A couple of weeks ago I loaded some Hasbro kiddie games for my
grandson and that's when things began to happen.

McG. says:
Heh! You DID say "Viper TnT-32 Ultra" That was the immediate
predecessor to the Diamond Viper V770 Ultra, an AGP 2x/4x TnT2 Ultra 32
meg card. Which I bought on July 1, 2001 for myself. My 47th birthday
That card is still running fine in an ABit BF6 (440BX) motherboard with
a P3-450 and 768 megs ram, which I gave to my youngest daughter about four
years ago. eh Hem... I digress

You're right John, your card is an oldie, but should be a goodie. With
Diamonds reputation being worse than ATI's old one with drivers and
stability, I'm astounded at the utter reliability of these two ancient
cards.
Neither have a temp sensor. But you can use the auxilliary thermal diode
that comes with many mobos to give you some indication of what's going on
with the GPU. You have clear indication of what's going on with your CPU
temp wise. I think I can see what's driving your thinking to the GPU.
Sitting down here in Houston with a different perspective, I can't see the
video card causing a hardware proplem to so affect the CPU in such a
direct manner.

Is this your mobo? The GA-K8NS?
http://www.giga-byte.com/MotherBoard...ec_GA-K8NS.htm
Socket 754 with an s754 A64-3000+?

Ok, I don't see or recall reading anything at all problematic with this
mobo or the CPU, even with standard retail boxed hs/fan. Makes for a
nice little rig, decent gamer too. Some folks doing some oc'ing with it.
Ok, I see a couple possibilities to look at. 1) My V770Ultra has a single
jumper on it to force the 1x/2x OR 4x for the AGP. Modern motherboards
like ours might be able to run that card at AGP 2X BUT that voltage is
3.3v and your mobo specs say 1.5v only. It would have to be AGP4X period.
Hm. Ok. So you have an old vid card that any of the 3D games are beating
up on. Ok, that still should not drive the CPU temp up. What could drive
it up would be the CPU doing most of the graphics intensive work that a
newer vid card would normally do itself under DirectX 7 or higher.
The CPU is heavy loaded and undercooled. That's what I think is going
on. Perhaps a more modern video card would help a little bit, but where
things get real CPU intensive the temp is still going to go up. Only
ways to test this would be another newer type video card and then a better
hs/fan than you have on the CPU now.
It's not a 'little' problem, not when your CPU temp goes all the way to
core shutdown.
Hope something here helps!
McG.

Yes, a series of long slow beeps is a problem with system memory. That
beeping is coming from your motherboard, and concerns the SDRam. Check
something; have the side off the case, so when it does it again, shut the
power switch off on the back of the computer (power supply rocker switch on
back), then touch the back side of one of the modules. Hot? Like in
noticeably hot? They *should* be barely warm. Almost cool. I've been
playing aroud with 3 other older systems getting them running from old junk
parts. I was checking that very thing on one of them. Seems I actually had
a tiny fleck of paper down inside the dram slot. Cleaned that out and the
modules are fine.

Now. After reading a few things I gotta ask this one; what video driver
version number are you running? If it's the 8x.xx forceware series, then
you've run across one of the many unusual problems that have shown up with
this brand new series. If you're using them, I suggest uninstalling them
and then installing a driver set of older series (like 77.77) and seeing how
the system behaves with that set.
McG.

"John McDonnell" wrote in message
news:Wo4lf.1282$Oq3.688@trnddc05...
Once again I really appreciate the advice. The mainboard is the
K8NS-939-Ultra (nForce3 ultra). The video card is the Diamond 770 Ultra
TNT32. The CPU is a 3000+ Winchester. I have ordered a Rosewill nVidia
6200 256mb AGP card from Newegg. I will certainly report back the results
to the group.

As long as I run no 3D apps and set the Vcore to 1.325, the system is
fine. But my grandson likes his Hasbro et al games. Most recently he was
playing Monopoly (not the most fast-moving game around, but 3D
nonetheless) and the system shut down and would not come back up. I got
the loud single beep every so many seconds. According to the manual, this
indicates a dram issue. As to whether this could be a video dram issue
and not a mainboard dram issue, the book didn't say either way. And it's
no good to leave the box shut down for a short time and start it up again.
It must be off for at least an hour. Very strange.

I have to eliminate components one at a time - improper seating of boards
and/or chips, video incompatibility, ram failure, insufficient cooling,
defective CPU, motherboard failure. And I'm starting with the video card.

Thanks, John
"McGrandpa" wrote in message
...
"John McDonnell" wrote in message
news:vPgkf.2357$1y.2302@trnddc07...

"McGrandpa" wrote in message
...

"John McDonnell" wrote in message
news%Nhf.30$xQ3.12@trndny04...
Forgot to mention: A64 3000+ Winchester. Bios F9 (Oct 2005).
"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...

John said:
I have a 8KNS and have no overclocking. It is a lightweight system
with an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC
running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C)
until I do anything graphical. Then the temperature will go slowly up
to the limit, which I set to 65 C. This happens with any kind of
game, even the old Hasbro kid games. I have replaced the heatsink. I
have used Arctic Silver regular and Ceramique. I have lowered the
Vcore in stages down to its present 1.32. I have turned ECC on and
off. The power supply is an Antec 350, plugged in to a Belkin 1200
AVR UPS. I can run for days and days, doing spreadsheets, surfing the
web, etc - anything except a game of any kind (not including
Solitaire, which runs trouble free). I think that the problem is due
to one of three things: (1) flakey CPU, (2) faulty mainboard, (3) some
kind of issue with the old Diamond TNT32 Ultra. If anybody has ever
experienced anything similar and found a workaround, I would be most
grateful in learning your solution.



McG. said:
I think actually bumping the core voltage UP a notch is better than
lowering it for keeping it running cooler. Is your CPU fan running?
It just sounds to me like the cooling solution isn't enough for what
you're doing with that cpu. you mentioned replacing the heatsink, what
about the CPU fan? And, are you SURE that the heat sink is mounted
*flat* on top of the CPU? Something seems to not be doing it's job. I
don't think it's a flakey A64 3000+ or TNT2 Ultra. When the CPU is
stressed under load, that heatsink and fan have to take the heat the
cpu gives off away. That's not happening, apparently. Poor contact of
hs to cpu, poor air circulation in the case, fan and/or hs inadequate
or faulty fan.
McG.



John said:
Thanks for the reply. Yes, standard thinking says to raise the CPU
voltage and not lower it. The normal CPU voltage for this motherboard
is 1.40. I had first tried a setting of 1.45. After restart, I went
into the Bios setup and brought up the PC Health screen and watched it
for a while. The temperature went up quickly 1 degree at a time from 38
to 65, at which time I shut down and set it back to 1.32. The PC Health
screen then stayed between 35 and 40 for quite a while. If I use the
default of 1.40, the CPU temperature will still rise to the danger
level, but much more slowly.

The heatsink is definitely seated OK, the Arctic Silver was applied
properly, and the fan runs steady about 3000 rpm. I'm pretty
experienced in that area. The circulation in the case is not the best,
but removing the side panel makes very little difference. Running
non-graphical apps causes no problems. While writing this email I have
watched the temperature monitor slowly drop from its initial 38 to its
present 34. But if I should start up any kind of game, it will slowly
creep up to past 60, at which time I will shut down.

I am still suspicious of the TNT2 board. It is an older model - Diamond
Viper TNT32 Ultra (one of the hottest boards around when I purchased it
in early 1999) with a jumper for setting AGP 2X / 4X. I say this
because the system operates trouble-free except when running graphics,
even kiddie games. I built this box about six months ago and used it
regularly (internet browsing, VPN, mail, worksheets, etc.) with no
problem. A couple of weeks ago I loaded some Hasbro kiddie games for my
grandson and that's when things began to happen.

McG. says:
Heh! You DID say "Viper TnT-32 Ultra" That was the immediate
predecessor to the Diamond Viper V770 Ultra, an AGP 2x/4x TnT2 Ultra 32
meg card. Which I bought on July 1, 2001 for myself. My 47th birthday
That card is still running fine in an ABit BF6 (440BX) motherboard
with a P3-450 and 768 megs ram, which I gave to my youngest daughter
about four years ago. eh Hem... I digress

You're right John, your card is an oldie, but should be a goodie. With
Diamonds reputation being worse than ATI's old one with drivers and
stability, I'm astounded at the utter reliability of these two ancient
cards.
Neither have a temp sensor. But you can use the auxilliary thermal
diode that comes with many mobos to give you some indication of what's
going on with the GPU. You have clear indication of what's going on
with your CPU temp wise. I think I can see what's driving your thinking
to the GPU. Sitting down here in Houston with a different perspective, I
can't see the video card causing a hardware proplem to so affect the CPU
in such a direct manner.

Is this your mobo? The GA-K8NS?
http://www.giga-byte.com/MotherBoard...ec_GA-K8NS.htm
Socket 754 with an s754 A64-3000+?

Ok, I don't see or recall reading anything at all problematic with this
mobo or the CPU, even with standard retail boxed hs/fan. Makes for a
nice little rig, decent gamer too. Some folks doing some oc'ing with it.
Ok, I see a couple possibilities to look at. 1) My V770Ultra has a
single jumper on it to force the 1x/2x OR 4x for the AGP. Modern
motherboards like ours might be able to run that card at AGP 2X BUT that
voltage is 3.3v and your mobo specs say 1.5v only. It would have to be
AGP4X period. Hm. Ok. So you have an old vid card that any of the 3D
games are beating up on. Ok, that still should not drive the CPU temp
up. What could drive it up would be the CPU doing most of the graphics
intensive work that a newer vid card would normally do itself under
DirectX 7 or higher.
The CPU is heavy loaded and undercooled. That's what I think is going
on. Perhaps a more modern video card would help a little bit, but where
things get real CPU intensive the temp is still going to go up. Only
ways to test this would be another newer type video card and then a
better hs/fan than you have on the CPU now.
It's not a 'little' problem, not when your CPU temp goes all the way to
core shutdown.
Hope something here helps!
McG.

I'm glad to hear about that concerning the CPU heating up being solved! The
temps you're listing sound very normal. I'll remember that glitch with the
older vid cards in a new-ish mobo.
McG.

"John McDonnell" wrote in message
news:IbJmf.11411$OK6.9832@trnddc03...
Replacing the old Diamond Viper TNT32 Ultra with a new Rosewill GeForce
6200 has solved the heat problem. As for the dram beeps, I switched the
two modules in the slots and (maybe because I seated them carefully) so
far so good.

The CPU temp (I set the vcore back to 1.40) stays from 28 to 32 in normal
windows operation (email, internet, worksheets, word proc) and from 30 to
38 running a 3D game. Not bad.

It looks like the old TNT32 made the CPU work extra hard during any 3D
app. There was one other funny thing - Everest reported that "AGP was not
enabled". With the new video board, that warning has disappeared.

"John McDonnell" wrote in message
news:FnNhf.975$4r.179@trndny01...
I have a 8KNS and have no overclocking. It is a lightweight system with
an old TNT32 Ultra AGP card. The ram is Crucual PC 3200 ECC running at
DDR 400 dual channel. The temperatures are stable (38 to 42 C) until I do
anything graphical. Then the temperature will go slowly up to the limit,
which I set to 65 C. This happens with any kind of game, even the old
Hasbro kid games. I have replaced the heatsink. I have used Arctic Silver
regular and Ceramique. I have lowered the Vcore in stages down to its
present 1.32. I have turned ECC on and off. The power supply is an Antec
350, plugged in to a Belkin 1200 AVR UPS. I can run for days and days,
doing spreadsheets, surfing the web, etc - anything except a game of any
kind (not including Solitaire, which runs trouble free). I think that the
problem is due to one of three things: (1) flakey CPU, (2) faulty
mainboard, (3) some kind of issue with the old Diamond TNT32 Ultra. If
anybody has ever experienced anything similar and found a workaround, I
would be most grateful in learning your solution.