• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Unstable HD2900XT - Heat - PSU - Drivers?

Soldato
Joined
11 Sep 2003
Posts
14,716
Location
London
Hi Guys,

just trying to work out whats wrong with my Windows XP system after upgrading the GPU from a X1950XT to HD2900XT.

It seems to work fine most of the time however now and then when I run 3D-Mark (03/05/06) there is a good chance it will hang during the benchmark?

3DMark wouldn't work at all until I read there was an issue and followed the fix (I deleted C:\WINDOWS\system32\Futuremark\MSC\Direcpll.dll).

Managed to get some results as it doesn't always crash . . .

c2d28ghz2900xt512mb1280ai8.jpg

3DMark 06 - *Stock* HD2900XT - C2D E6300 @ 2800MHz

When it goes wrong the screen just goes blank and enter Standby Mode and I am not able to access the PC anymore (even via the network shares) until I reboot the machine. Doom3 seems to run fine as does BIO-Shock (very impressed by how good the game is on this card at 1920x1200!!!!)

I have been troubleshooting this slowly but surely, the system itself is Orthos Stable (Blend 8 Hours) and has the Catalyst™ 7.9 Display Driver for Windows XP installed. I have tried running the system at stock but the issue remains.

Below is a picture of RivaTuner taken after the machine had been running rthdribl for a few hours, does the temps and stuff look ok? I don't think I've seen the GPU temps rise over 76°C yet (rthdribl can run for hours and hours with no problems btw)

rthdriblstocktemps01wr5.jpg

RivaTuner after running rthdribl *fullscreen* 1920x1200 for 2 hours

So now If heat isn't an issue I'm wondering if my PSU is causing the problem? Its a 9 month old ANTEC TruePower TRIO 550w and I have both its 6-pin PCI-E plugged into the card. Looking at the PSU Calculator it seems a bit borderline to me?

psu2900xtev0.jpg

My Hardware

Any ideas?

What is the most likely cause for 3DMark hanging? The only things of that I can try now are:

  • Reseat the card
  • Fresh Windows Install
  • Try a New Meaty PSU?

Thanks in advance! :)
 
Hi,

I have a similar system using a corsair 620w.

The Antec 550W should be fine, I have a plug in power meter on my system and it usually draws around 350w max during 3dmark06 or heavy gaming.

It's worth checking the cooling in your PC case, the 2900 sucks a lot of air through once the fan start running at speed, it could be reducing the amount of air passing through other devices eg. the PSU and so reducing the effective cooling of it and the output.

Check the temperature of outlet air with your hand, from the case fans or psu it should feel no more than a little warm.

Try ATItool - check for artifacts, that heats my card up nicely

Orthos should really be small integer to stress the CPU as blend doesn't push the system too hard.

You could also use driver cleaner in safe mode to strip out all of the old drivers and restart, could just be a driver issue.

Try to setup monitoring of the +12v supply in your system, that should give you an idea of if the voltage is stable.

Otherwise your temps look fine so it's a difficult one to troubleshoot.

AD
 
My 2900XT does exactly what your saying in 3DMark, yet it will run everything else fine.

I'm not sure of a way, but it may be a good idea to check how hot the card is actually getting in 3DMark 06. Specifically my card is limited by heat, at higher clocks (850/950) it refuses to go above 65c which is hard to stay away from with a 2900.

All I can say is check it's not going above 75c.
 
run your kit all at stock and see if you still have the problem, work from there,

While people bang on about Orthos stability I have had situations where my systen was Orthos stable for 12+ hours but would hang while playing games.

Also don't forget the extra load in 3D that won't be present running orthos in the desktop.
 
Try to setup monitoring of the +12v supply in your system, that should give you an idea of if the voltage is stable

hd2900xt12vog1.jpg

+12v Rail Idle, then running 3DMark06, then idle again

I think it looks ok, at least from what the software is telling me?

It's worth checking the cooling in your PC case, the 2900 sucks a lot of air through once the fan start running at speed, it could be reducing the amount of air passing through other devices eg. the PSU and so reducing the effective cooling of it and the output
During intense 3D the air is indeed very warm coming out the back of the card, however the air coming out the case exhaust fans and PSU is rather cool, case temps are about 32°C
 
I have the 2900 pro, which is the same, only with slower clocking ram, I have an OCZ 600w GameXStream with 6 HDDs, a Q6600, SATA RAID card, Terratec soundcard etc, and i am not getting any issues from lack of power, so you 550W Antec should be fine. :)
 
My 2900XT does exactly what your saying in 3DMark, yet it will run everything else fine
Hmm thanks for letting me know, could well be just a 3DMark thing, however Doom3 has failed to launch on me once (just once though!).

I'm not sure of a way, but it may be a good idea to check how hot the card is actually getting in 3DMark 06. Specifically my card is limited by heat, at higher clocks (850/950) it refuses to go above 65c which is hard to stay away from with a 2900.
I already checked that and posted above but in case you missed it in my jumble of a post here it is again! :)

rthdriblcorejj6.jpg


All I can say is check it's not going above 75c.
It is going above 75°C ?
 
I have tried running the system at stock but the issue remains
run your kit all at stock and see if you still have the problem, work from there
:rolleyes: ;) :D

While people bang on about Orthos
My reference to Orthos was included to verify that my problem with the graphic hang is not down to some basic other problem like a general unstable system. My system is rock solid using a X1950XT and I've never had a *Hang* before however since popping in the HD2900XT the situatioj has changed.
 
But I'm running a massively overvolted E4300, loads of fans, 2 HDDs, 2 PCI cards with 2900XT @ 830/900 on a Tagan 530W and everything is perfectly stable?
 
Last edited:
Hmmmm, I'm not really sure what's going on then. Is there another machine you could try it in? If not, crank the fan speed upto 100% and see if you crash still. That's the only way to see if it's a heat problem.
 
What is the most likely cause for 3DMark hanging? The only things of that I can try now are:

  • Reseat the card
  • Fresh Windows Install
  • Try a New Meaty PSU?

Not sure what the exact cause was but after I finally did a fresh installation of WinXP its all sorted and working well!

Could have been a driver problem but I would have thought using driver-cleaner in safe-mode would have cleaned away any driver gunk but no?

The previous WinXP installation wasn't exactly old and crusty but I did install of load of random stuff recently like DivX codecs and the like?

Thanks for the help anyway, now I feel a bit silly for going and ordering a 1Kw PSU! :o :D
 
Can you not send the PSU back under the distance selling act?. You'd have to have ordered online and it must be in 7 days of receipt.

Also, nice one on getting the card up and running ;). Are you having fun with your new card?.
 
im on a...... hyper :eek: 580w here :) and all is good with my 2900xt, if you are botherd about temps etc (i think everything looks ok personally) then download ATI Tool, as there is fan controlls in there, i have my fan 100% silent idle. its idleing at 51c right now with a slower fan than stock, and the load fan speed is just standard (tiny bit audiable but not loud)
 
There was a suggestion about heat on here, and turning up the fan to 100% to check to see if it's heat related.

I also assumed once when my x1950xt at stock speeds, was totally unstable and would crash the system randomly and have artifacts in game. When I turned up the fan to 100% it cured the problem, but i wasn't happy as so darn noisy.

What it turned out to be was the bios didn't like the standard 100mhz on the pci-e bus. I also had to reset the bios as when I changed the graphics cards over (100mhz standard) it would corrupt my drives (Asrock 4coredual) and only a bios reset would solve the issue (then make all changes as necessary).

If heat doesn't seem to be the issue (i.e. it's within nominal range), then I suggest the problem is elsewhere and don't assume it's heat.

Matthew
 
There was a suggestion about heat on here, and turning up the fan to 100% to check to see if it's heat related.

I also assumed once when my x1950xt at stock speeds, was totally unstable and would crash the system randomly and have artifacts in game. When I turned up the fan to 100% it cured the problem, but i wasn't happy as so darn noisy.

What it turned out to be was the bios didn't like the standard 100mhz on the pci-e bus. I also had to reset the bios as when I changed the graphics cards over (100mhz standard) it would corrupt my drives (Asrock 4coredual) and only a bios reset would solve the issue (then make all changes as necessary).

If heat doesn't seem to be the issue (i.e. it's within nominal range), then I suggest the problem is elsewhere and don't assume it's heat.

Matthew

Not sure what the exact cause was but after I finally did a fresh installation of WinXP its all sorted and working well!

:p

Did you re-apply the thermal paste? You won't believe the kind of crap thermal material they put on these high end gpus.
 
Back
Top Bottom