pc reboots when acceleration set to full

  • Thread starter Thread starter los
  • Start date Start date
L

los

I recently upgraded some parts of my computer.

I have a leadtek Winfast a350 based on Nvidia 5900 XT 128 meg card,
AMD 2.6 Barton
Asus a7v8x-x MB
1 gig Corsair 300 mhx SDRAM
Antec 350 watt power supply
Windows XP

When I don't have Nvidia's drivers installed or I have Hardware acceleration
set to 0 the pc runs fine. With the newest nvidia drivers installed and hw
acceleration set past 0, the computer reboots when I scroll down on a web
page in Internet explorer, (or sometimes when I move a window around on the
desktop).
I have just recently formated the HD and re installed XP. The mem, and the
video card run fine on another MB. Any ideas?
 
"los" said:
I recently upgraded some parts of my computer.

I have a leadtek Winfast a350 based on Nvidia 5900 XT 128 meg card,
AMD 2.6 Barton
Asus a7v8x-x MB
1 gig Corsair 300 mhx SDRAM
Antec 350 watt power supply
Windows XP

When I don't have Nvidia's drivers installed or I have Hardware acceleration
set to 0 the pc runs fine. With the newest nvidia drivers installed and hw
acceleration set past 0, the computer reboots when I scroll down on a web
page in Internet explorer, (or sometimes when I move a window around on the
desktop).
I have just recently formated the HD and re installed XP. The mem, and the
video card run fine on another MB. Any ideas?

It could be a power problem, or it could be an AGP slot problem.

This sounds like a Herolchi 350W supply, done under contract to
Antec. 3.3@20A, 5V@33A (max 180W for both), 12V@15A. The reason
I know the numbers, is I have one too (only I use a R8500, which
doesn't need nearly as much power as the 5900).

On an AMD motherboard, there generally isn't a 2x2 connector for +12V,
so this motherboard draws processor power from +5V. A 2.6 barton
needs 53.7W @ 1.65V, and that is 32.5 amps. Converted from +5V, this
is 32.5*(1.65/5.0)=10.7 amps. With an assumed conversion efficiency of
80%, the current needed rises to 13.4 amps from the +5V supply.

The Nvidia 5900 draws 5V@10A and 12V@2A, under some 3D gaming conditions.
This is purely a guesstimate, based on some numbers from a power supply
estimator website.

The sum of those two numbers, is 23.4A from +5V. Out of 33A available,
this still sounds safe.

Now, consider the combined power limit. Of the 180W combined allowed
on +3.3 and +5.0, 117W is consumed by the +5V. This leaves 63 watts
from 3.3V, or 19 amps. From this, you could be powering DIMMs (maybe
5W a piece), and any PCI bus based chips could be drawing from +3.3V
as well.

On the surface, there should be enough power there, so maybe your
PSU is a little weak, or you have some other loads you haven't mentioned.

Get a copy of Motherboard Monitor (MBM5) from livewiredev.com .
Check the voltages using the MBM5 dashboard, while you do some
stuff. What I would expect you would see, is the +5V a little on
the low side (the most stressed supply), +12V on the high side,
and +3.3V pretty close to normal. Typically, PSUs have a 5% tolerance
on regulation, so if you are seeing larger deviations, like 10% on the
low side on any output, you might consider getting a larger supply.

The other possibility, is the AGP interface just isn't ready to
transfer data at the rate the FX5900 is capable of. You might try
reducing the AGP transfer rate setting, to see if you can get some
stability. You could try disabling fast write, if the motherboard has
such an option in the BIOS. Once you can get the motherboard to "stay up"
for a few minutes, then it is time to try 3Dmark in demo mode (3dmark2001se
if using directx8, 3dmark2003 if directx9, I think). This will do a
good job of making the video card draw current representative
of a gaming situation. Before starting 3dmark, enter the options menu
for MBM5, and enable the "logging to txt file" function. Set it to
record all readings at 10 second intervals. If the computer goes
berserk at some point, and reboots, open the txt file after the
computer reboots, and examine the recorded values near the end of
the file, for anything suspicious.

If you haven't already, you should also run memtest86 on your memory,
to make sure it is solid. Your symptoms don't suggest that memory
is an issue, but error free memory is a good thing to test for
anyway.

HTH,
Paul
 
I got motherboard monitor. And my power consumption readings are
--Sensor Chip Readings--

Temperature Readout[1] : 32
Temperature Readout[2] : 43
Voltage Readout[1] : 1.696
Voltage Readout[3] : 3.312
Voltage Readout[4] : 4.8495
Voltage Readout[5] : 12.1644
Fan Readout[2] : 3924
Fan Readout[3] : 5273

The bios has the cpu temp a little higher (50C).
I'm trying to figure of the following sensors:
Possible Temp Sensors[1] : ITE8712F-1
Possible Temp Sensors[2] : ITE8712F-1 Diode
Possible Temp Sensors[3] : ITE8712F-2
Possible Temp Sensors[4] : ITE8712F-2 Diode
Possible Temp Sensors[5] : ITE8712F-3
Possible Temp Sensors[6] : ITE8712F-3 Diode

which is associated with my CPU socket, because when I select ITE8712F-3
for the cpu socket it says the tempeture is at 86C,and when I set ITE8712F-3
Diode for the cpu socket it says the temp is 127, which I don't
think could be right.
I noticed my bios has no reading for the CPU socket, is this even an
important reading considering my cpu and case temp is ok?
Also my not familiar enough with power to figure out the total watts my psu
is using by reading the voltage readouts in MB mon, is there a good utility
that will tell me the total amount of whats that my PSU is using, or if not
is there a quick way to figure this out? Great reply by the way, thanks.
 
"los" said:
I got motherboard monitor. And my power consumption readings are
--Sensor Chip Readings--

Temperature Readout[1] : 32
Temperature Readout[2] : 43
Voltage Readout[1] : 1.696
Voltage Readout[3] : 3.312
Voltage Readout[4] : 4.8495
Voltage Readout[5] : 12.1644
Fan Readout[2] : 3924
Fan Readout[3] : 5273

The bios has the cpu temp a little higher (50C).
I'm trying to figure of the following sensors:
Possible Temp Sensors[1] : ITE8712F-1
Possible Temp Sensors[2] : ITE8712F-1 Diode
Possible Temp Sensors[3] : ITE8712F-2
Possible Temp Sensors[4] : ITE8712F-2 Diode
Possible Temp Sensors[5] : ITE8712F-3
Possible Temp Sensors[6] : ITE8712F-3 Diode

which is associated with my CPU socket, because when I select ITE8712F-3
for the cpu socket it says the tempeture is at 86C,and when I set ITE8712F-3
Diode for the cpu socket it says the temp is 127, which I don't
think could be right.
I noticed my bios has no reading for the CPU socket, is this even an
important reading considering my cpu and case temp is ok?
Also my not familiar enough with power to figure out the total watts my psu
is using by reading the voltage readouts in MB mon, is there a good utility
that will tell me the total amount of whats that my PSU is using, or if not
is there a quick way to figure this out? Great reply by the way, thanks.
<<snip>>

It doesn't look like the CPU diode is supported, at least
according to this list. Compare your board to the A7N8X for
example.

http://mbm.livewiredev.com/comp/asus.html

Something else I want to check, is that the power cable to the FX5900
is not connected to a disk drive, as well as the video card. A
high end video card, with aux power cable, should get its power
directly from the power supply, with no other loads on that
cable.

OK. Everything in MBM looks fine, so it is time to work on the
AGP slot. First try slowing the AGP slot down, on the theory that
the Northbridge cannot handle the speed of the FX5900. You might
also want to try some other version of driver. I find I have
to try three or four versions until I find one that works.
And, it isn't the latest one that will necessarily be the good
one. Some drivers get over-optimized, at the expense of stability.
It helps to check Google to see which ones other people are
satisfied with.

Here is a sample:
http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&[email protected]

HTH,
Paul
 
It doesn't look like the CPU diode is supported, at least
according to this list. Compare your board to the A7N8X for
example.

http://mbm.livewiredev.com/comp/asus.html

Something else I want to check, is that the power cable to the FX5900
is not connected to a disk drive, as well as the video card. A
high end video card, with aux power cable, should get its power
directly from the power supply, with no other loads on that
cable.

OK. Everything in MBM looks fine, so it is time to work on the
AGP slot. First try slowing the AGP slot down, on the theory that
the Northbridge cannot handle the speed of the FX5900. You might
also want to try some other version of driver. I find I have
to try three or four versions until I find one that works.
And, it isn't the latest one that will necessarily be the good
one. Some drivers get over-optimized, at the expense of stability.
It helps to check Google to see which ones other people are
satisfied with.
Well, what I've done is taken out the 5900 FX and put back in my old Geforce
3 ti200 card. The computer does not crash but the display looks terrible.
Text is choppy and flickers, lines are not smooth. After I put my old card
back in I reinstalled all drivers and same thing. So it appears my system
does not do well with two different video cards (both of which worked fine
when installed on my old motherboard).
 
"los" said:
Well, what I've done is taken out the 5900 FX and put back in my old Geforce
3 ti200 card. The computer does not crash but the display looks terrible.
Text is choppy and flickers, lines are not smooth. After I put my old card
back in I reinstalled all drivers and same thing. So it appears my system
does not do well with two different video cards (both of which worked fine
when installed on my old motherboard).

It is generally a good idea to uninstall drivers on one video card,
before installing the other. There are differences between
brands, as to how tolerant the install process is.

When the drivers are uninstalled, you should be left at 640x480
with your original card. Then, you can shut down and install
the new video card.

I think there are installer cleaners available for Nvidia, ATI,
and Matrox. These cleaners remove old drivers and entries in the
registry, in cases where some residue has been left behind. Have
a look for some of those (or if I get a chance later, I'll see
if I can dig up some links).

HTH,
Paul
 
Paul said:
I think there are installer cleaners available for Nvidia, ATI,
and Matrox. These cleaners remove old drivers and entries in the
registry, in cases where some residue has been left behind. Have
a look for some of those (or if I get a chance later, I'll see
if I can dig up some links).

HTH,
Paul

Paul, I went out and bought a 430 watt PSU and installed it and wahla! No
more crashing, and I've got accelaration at full. I was ready to send
everything back for replacement when I figured I had changed out every
single component except for the CPU, so I may as well change out the PSU as
well, and that did it. For some reason these componets together were not
getting all the power they wanted, or I guess another possibilty but less
likely was that I had a bad power connector, but I did switch a few of
those out to test them, and I think the video card will give you a message I
think if you power it on with no power connected to it. Thanks for the help.
Cheers.
 
It is generally a good idea to uninstall drivers on one video card,
before installing the other. There are differences between
brands, as to how tolerant the install process is.

When the drivers are uninstalled, you should be left at 640x480
with your original card. Then, you can shut down and install
the new video card.

I think there are installer cleaners available for Nvidia, ATI,
and Matrox. These cleaners remove old drivers and entries in the
registry, in cases where some residue has been left behind. Have
a look for some of those (or if I get a chance later, I'll see
if I can dig up some links).

HTH,
Paul

Here are some uninstallers.

Catalyst uninstallers:
http://www.ati.com/support/drivers/misc/catalystutils.html

Another ATI uninstaller - don't know anything about this one
http://www.rage3d.com/board/search.php recommends...
http://www.driverheaven.net/cleaner/

Detonator RIP (use after normal Nvidia uninstall)
More recent than Detonator Destroyer
http://www.guru3d.com/index.php?page=detonatorrip&menu=0

Detonator Destroyer (not a current util?)
http://www.guru3d.com/detonator-destroyer/

Matrox Powerdesk Uninstaller
http://www.matrox.com/mga/support/drivers/files/ftp_util2.cfm

Paul
 
"los" said:
Paul, I went out and bought a 430 watt PSU and installed it and wahla! No
more crashing, and I've got accelaration at full. I was ready to send
everything back for replacement when I figured I had changed out every
single component except for the CPU, so I may as well change out the PSU as
well, and that did it. For some reason these componets together were not
getting all the power they wanted, or I guess another possibilty but less
likely was that I had a bad power connector, but I did switch a few of
those out to test them, and I think the video card will give you a message I
think if you power it on with no power connected to it. Thanks for the help.
Cheers.

Great to hear! Maybe the FX5900 just beat the old PSU to a pulp ?

Paul
 
Well, what I've done is taken out the 5900 FX and put back in my old Geforce
3 ti200 card. The computer does not crash but the display looks terrible.
Text is choppy and flickers, lines are not smooth. After I put my old card
back in I reinstalled all drivers and same thing. So it appears my system
does not do well with two different video cards (both of which worked fine
when installed on my old motherboard).

The gf3 ti200 should not have problems like you describe, unless it is a
refresh rate issue you are refering to (display/advanced/monitor). At any
rate, sounds like you've got the new PS working now. Do you have the
chipset/agp drivers installed (via 4-in-1 correct)? If not, might be
worth a try - but be careful! This is the place I've gotten burnt the
most times by non-intel (specifically via) chipsets, requiring OS
re-install. Usually it's the UDMA drivers which don't seem to install
correctly, then eventually cause the BSOD boot unrecognized disk issue.

--
We HAVE been at war with Iraq for 13 years now, bombing their
country on at least a weekly basis.
"U.S.-led sanctions have killed over a million Iraqi citizens,
according to UN studies" - James Jennings
3,000+ innocent Iraqi civilian casualties can't be "wrong"...
 
The gf3 ti200 should not have problems like you describe, unless it is a
refresh rate issue you are refering to (display/advanced/monitor). At any
rate, sounds like you've got the new PS working now. Do you have the
chipset/agp drivers installed (via 4-in-1 correct)? If not, might be
worth a try - but be careful! This is the place I've gotten burnt the
most times by non-intel (specifically via) chipsets, requiring OS
re-install. Usually it's the UDMA drivers which don't seem to install
correctly, then eventually cause the BSOD boot unrecognized disk issue.
I do have the 4 in 1 drivers installed. They seem to be doing fine on my
system. Not sure why the ti200 was doing what it was doing, may have been
related to the power problem. I'm just thankful to have a working pc.
 
Back
Top