Sign in with
Sign up | Sign in
Your question

HD3870x2 + Asus P5K + Unstable -Help Please

Last response: in Graphics & Displays
February 25, 2009 3:47:16 AM


Recently bought 2 HD3870X2s and having major problems

I am running mainly Vista but also Xp and Windows 7 (32 & 64) in 2 gb with a 1000 Watt Jeantech Absolute PSU

Regardless of OS, my systm will periodically shutdown (graphics go, radeon fan speeds up slightly and system requires reboot)

Event view shows "unexpected shutdown" but nothing helpful to diagnose

I had thought that PSU (previous 570 watt) may have been failing at peak load so replaced with the 1000 watt -sadly this has not helped.

I would be grateful for any pointers to the problem - especially good suggstions for diagnostic tools / techniques

I am not using dual Cross fire and getting the problems with both cards independently - one card is Powercolor PCS and the other is Sapphire - i am getting to the stage where the only really common component is the motherboard as I have even swapped the 2 gb ram

Getting desperate guys as the 2 HD3870X2s plus the PSU are pretty serious investment (for me anyway :D  )


February 25, 2009 4:22:18 AM

This problem seem to be usual with people using 2 dual GPU's. May be voltage distribution process not working well with the PSU. Have you tried working with a single HD3870X2?
February 25, 2009 4:35:24 AM

Have you tried increasing the fan temps via rivatuner or CCC?

Also check your temps on those cards, and maybe run memtest for a few hours to test your system memory.
Related resources
February 25, 2009 12:09:56 PM

Thanks Guys

Just using one 3870X2 at a time but getting the problems on each

Have installed RivaTuner and also GPU-Z which is outputing to a log file.

Interestingly, there are some logged VCCD current spikes but I haven't had chance to compare with Vista Event logs to see if the resets are related.

I did notice that the GPU-Z readings for each core are radically different (temp and VCCD) - is this ok or should they be sharing the load equally with roughyl equivalent numbers ? My current thinking is that the core driving the monitor is the one doing the work and so higher numbers but this logic could be wrong
February 28, 2009 1:43:35 AM

In my system, the GPU labled "Core0" in RivaTuner is always the hottest one, and the core that gets all the load from watching movies or videos, etc. I have both monitors (1920x1200 and 1280x1024) running off the 3870x2.

Before I replaced the fans on the 3870x2 (ZEROTherm) I noticed that the system would lock if the GPU temp hit 92 celcius or so. That would happen with the fans fixed @40% by RivaTuner as the Catalyst drivers wouldn't spin up fast enough to make it more than 10 min or so during heavy gaming.

I don't get above 70 now on any core and I'm running the 3870x2 @855/955, and the 3870 @855/1170.

I get lots of spikes from the VCCD, and temps get pretty hot on them as well but it doesn't seem to hurt my system.

Have you tried another (spare, non 3870x2) in the system yet?
February 28, 2009 1:44:41 AM

By the way, when your system shuts down what are you doing? Games, movies, idleing at desktop...???
February 28, 2009 12:40:02 PM

Hey Firebird - I am really grateful for the good advice and time you are taking over my problems.

I appear to be making some headway as I have pulled a spare hard drive off my shelf and have carefully built up a new Vista 32 installation - this has now been running for about 18 hours without any spontaneous resets.

Previously the resets have happened both at idle, during gamplay, running ATITool tests etc. The only way I could get stability was to pull thew 3870X2 and revert to old cards (NVIDIA 8600 or ATI X1650).

I am also running an extemely lean setup - all of my non boot drives are disconnected to narrow down the active components

GPU-Z is reporting core temps of around 65 degrees on core 0 and 52 degrees on core 1 with fairly low fan revs. Rivatuner seems to be managing fan speeds effectively.

I have not yet tried to stress the 3870 with any intensive games but this is my next step (or maybe using a decent benchmark progamme if i can find one).

I am really a novice when it comes to fancy graphics cards so I really appreciate your guidance

March 1, 2009 11:34:47 PM

It's funny you mentioned the other hard drive. When I had issues with lock ups and crashes, I was also getting these wierd warnings related to the hard disk. I really thought my drives were dying, but after I got re-installed Vista for the 3rd time I decided not to use Everest for benchmarks and monitoring, and I've not seen the errors again.

BTW, what vid drivers are you using?
March 1, 2009 11:37:50 PM

Make sure you take the step-by-step approach with this still. If you're satisfied the system is stable, add components, software, and make any changes 1 at-a-time.

Also, make sure you stress test everything before adding anything.

It will take more time, but than you'll be more apt to find the problem.
March 2, 2009 1:11:56 AM

Hi Firebird

I have reverted to my old drive and guess what - the crashes have started again so back to the clean build for more evaluation.

What would you suggest for stress testing ?


March 2, 2009 1:12:08 AM

Hi Firebird

I have reverted to my old drive and guess what - the crashes have started again so back to the clean build for more evaluation.

What would you suggest for stress testing ?


March 2, 2009 11:30:07 PM

Prime95 and OCCT will stess your system nicely, as will 3dmark06 and 3dmark Vantage.

Your issue is a tough one though. What happened between the time you said the new HDD seemed ok and now? Vista updates? An O.S. setting maybe?

By the way, you might as well list out your system with hardware and software versions. This is getting to that point....