Help me troubleshoot my 5970.

[Ashkael]

Hello all,

I recently bought an MSI R5970 but I'm having some issues with my card. Before I continue, let me please state that RMI is not an option for me, as I do not live in the US and my card had to take 2 week trip through the mailing system to get to me.

I'm running the card in Windows 7 Ultimate 64-bits with a HX 1000W Corsair PSU on an Asus P6T Deluxe V2 mobo, with an Intel Core i7 920 processor OC'd to 3.8Ghz.

My problem: the card, freshly installed and with all settings at stock, causes an ATI driver crash 1-3 minutes into any game I've tried to play. As far as benchmarks go, the drivers are crashing on 3D Mark Vantage at the very beginning of the second GPU test. Prior to the driver crash, the games work perfectly fine with sky-high FPS. Drivers also crash shortly into a Furmark dual GPU benchmark or stability test. Note that the Furmark driver crashes come accompanied with a bit of artifacting. So far I've only managed to complete Furmark's 1-minute benchmark test once.

Coupled to this is the fan control on CCC. If set to manual, no matter where I move the slider it results in the fan going to 100% speed. If set back to auto, the fan eventually speeds down. MSI afterburner is not letting me change fan settings.

MSI Online Live update keeps crashing on IE (I believe it doesn't work on 64 bit systems), and MSI LiveUpdate 4 which came with my 5970 is not finding any BIOS updates for my card.

I tried uninstalling all drivers, then running Driver Sweeper, then installing Catalyst 10.4. No dice.

Anyone else have any ideas on what could be happening? Or do I just got me a $700 door stopper?.

EDIT: And I just noticed, the VDDCI Phase #1 Temperature in GPU-Z is showing zero.

 

[thilanliyan]

What are the temperatures like before crashing?

 

[happy medium]

Are you using the 10.4 drivers ? Try the 10.3
Do you have direct x 9 installed .? You should.
Does it only crash in 3d? (Gaming or benching)

Temps in GPU-z?

Is your cpu overclock stable? Did you try it with no cpu overclock?

 

[Ashkael]

Highest temperature in GPU-Z right before my system froze during a Furmark Stability Test was 81C for the VDDC Phase #2 Temperature. The GPUs don't go over 75C, and the other VDDCs stay at around 75.

Yes, I'm using Catalyst 10.4. Was there some bug introduced with 10.4?

One thing I should mention, after uninstalling my drivers and reinstalling everything again, now I just crash. There's no driver recovery like before. Hard system lockup forcing me to reboot.

Given the card was putting out some amazing FPS but crashing, I decided to underclock it a bit to test. Furmark crashed again, but the really GREAT thing is after returning the clocks to stock values, the card is now acting as if it was STILL underclocked! Moreover, trying to run 3D Mark Vantage after this now causes some WEIRD bugs with the water in the first GPU test: it's all distorted and looks more like stretched jello than water. D:

As far as my CPU overclock, yes it is stable (passed OCCT multiple times and endured hours of prime95 with no issues, peaking at 69C for Core 1). Haven't tried with an underclocked CPU.

There's also the issue of fan control. MSI afterburner lets me move the fan on my 5870 freely, but not on my new 5970. Trying to change it through CCC results in 100% fan speed no matter where the slider is.

 

[Madcatatlas]

put in a spare card, do a complete drivesweep and all software sweep. Then install the card using as few software addings as you can. this means no CCC , just put the card in and let win7 take care of the rest.

Try a few games like that. depending on your results, update the drivers to the 10.2 then 10.3 and 10.4, this is to see if any of the drivers come in conflict with your hardware setup.


This is just how i would do it, nothing i have done.

 

[Kenmitch]

Maybe try the show hidden devices thing and clean out the system

Click Start, point to All Programs, point to Accessories, and then click Command Prompt.
At a command prompt, type the following command , and then press ENTER:
set devmgr_show_nonpresent_devices=1
Type the following command a command prompt, and then press ENTER:
start devmgmt.msc

 

[Ashkael]

put in a spare card, do a complete drivesweep and all software sweep. Then install the card using as few software addings as you can. this means no CCC , just put the card in and let win7 take care of the rest.

Try a few games like that. depending on your results, update the drivers to the 10.2 then 10.3 and 10.4, this is to see if any of the drivers come in conflict with your hardware setup.


This is just how i would do it, nothing i have done.

Just tried this.

1) I put in my 5870.
2) Ran the ATI Install tool and did an express removal of "ALL ATI" software.
3) Ran Driver Sweeper and cleaned all ATI remnants.
4) Turned off my PC and switched to the 5970.
5) Turned on my PC and let Win7 do its thing.

Windows 7 identified my new device as some sort of "ATI HD audio" device or something. I couldn't change resolutions other than the default one nor run any games nor benchmarking utilities, so I went ahead and installed MSI's driver from the DVD that came with my card. Rebooted, and back to square one: the card still crashes, and oddly enough it's STILL acting as if it was underclocked!

 

[Ashkael]

Maybe try the show hidden devices thing and clean out the system

Click Start, point to All Programs, point to Accessories, and then click Command Prompt.
At a command prompt, type the following command , and then press ENTER:
set devmgr_show_nonpresent_devices=1
Type the following command a command prompt, and then press ENTER:
start devmgmt.msc

Tried this. My device manager is showing TWO HD5900 series adapters every time I put in the R5970. Maybe one for each GPU? I honestly don't know. Trying to remove/disable either of them immediately resets my resolution to default and shuts down my access to any 3D applications.

 

[Kenmitch]

Maybe a fresh install on a spare drive is the next logical move.

Or you could revert to generic vdeo driver then remove all ati software. Run driver sweeper then do the show hidden devices and remove all ati in device manager. Both video and audio devices. Reboot and try to install drvers again.

 

[Puffnstuff]

What ps is the op using?

 

[Ashkael]

What ps is the op using?

Same as yours sir, a Corsair HX 1000W. It should be enough to run a single HD5970 just fine. I think. :\

 

[Ashkael]

Or you could revert to generic vdeo driver then remove all ati software. Run driver sweeper then do the show hidden devices and remove all ati in device manager. Both video and audio devices. Reboot and try to install drvers again.

I already did that sir. It didn't work.

What's more baffling is that the card is STILL acting as if it was underclocked from when I did so earlier today to test its stability. How the hell can the video card still be running at underclocked values despite having removed it from its socket, having used a different card to wipe all existing ATI drivers and software, and then done a clean reinstall of everything?

 

[SolMiester]

I already did that sir. It didn't work.

What's more baffling is that the card is STILL acting as if it was underclocked from when I did so earlier today to test its stability. How the hell can the video card still be running at underclocked values despite having removed it from its socket, having used a different card to wipe all existing ATI drivers and software, and then done a clean reinstall of everything?

Seems the only thing left to do before you re-consider your RMA stance is to test the card in another PC!

 

[Kenmitch]

I already did that sir. It didn't work.

What's more baffling is that the card is STILL acting as if it was underclocked from when I did so earlier today to test its stability. How the hell can the video card still be running at underclocked values despite having removed it from its socket, having used a different card to wipe all existing ATI drivers and software, and then done a clean reinstall of everything?

I'd suggest fresh install on a spare drive. Easiest way to see if it's the card or not I'd think.

I went thru a similar situation with my 5850 caused by a series of bios flashes and CCC crashes due to overclocking. Tried everything which didn't work in the end. Thought maybe killed my card but just did fresh install on spare drive to test it and it worked fine. Wound up just doing a fresh install on my SSD in the end.

Have you tried a registry cleaner program?

 

[Ashkael]

I'd suggest fresh install on a spare drive. Easiest way to see if it's the card or not I'd think.

I went thru a similar situation with my 5850 caused by a series of bios flashes and CCC crashes due to overclocking. Tried everything which didn't work in the end. Thought maybe killed my card but just did fresh install on spare drive to test it and it worked fine. Wound up just doing a fresh install on my SSD in the end.

Have you tried a registry cleaner program?

Well, I'm back to square one with the video card. I just did another fresh reinstall of everything, but this time I just installed the driver and no CCC or anything else for that matter. The card is back to working at its core clocks, but still crashes seconds into any game or benchmarking application.

I also went ahead and reset my BIOS to factory settings to get rid of my CPU overclock, but no dice.

I'm going to try running it with a spare drive I got around with another copy of Win7 64.

EDIT: Yes, I've tried running CCleaner.

 

[Qbah]

Is the PCI-e slot running at x16? Did you try using a different PCI-e slot perhaps? Check if the card's power cables are connected properly. Also, you might want to check how the voltage fluctuates when you load the card.

A fresh install and a recurring problem suggests a hardware issue here. It is possible the card arrived defective. Also check if the voltage changes much when you load the card (so a bad PSU). Was the CPU overclock fully stable? How long was the computer working before you got the card?

 

[Ashkael]

Is the PCI-e slot running at x16? Did you try using a different PCI-e slot perhaps? Check if the card's power cables are connected properly. Also, you might want to check how the voltage fluctuates when you load the card.

A fresh install and a recurring problem suggests a hardware issue here. It is possible the card arrived defective. Also check if the voltage changes much when you load the card (so a bad PSU). Was the CPU overclock fully stable? How long was the computer working before you got the card?

CPU-Z shows the PCIe link width at x16. I haven't tried using a different PCI-e slot, mostly because my motherboard recommends using PCI-e slot 1 for the main VGA. I've checked and double checked the cables; everything is connected properly.

The CPU overclock is stable. Like I said before, I've left Prime95 running for hours with no issues at all. The machine was working for 4 months prior to me acquiring the card; I put it together myself. I've been using a Diamond HD5870 all this time with no issues at all.

I'm gonna fire up a 3D Mark Vantage bench since my card crashes there once it hits the 2nd test. I'll log the results on a text file using GPU-Z and post them.

I just tried getting this card to work on a separate drive with a different installation of Win7 Ultimate 64 and fresh drivers with no luck. About the only things I haven't tested yet are 1) a different OS, 2) a BIOS flash (maybe that would solve fan control? MSI's Live Update tool insists I got the latest BIOS though), and 3) plugging the card in a different system (pretty much impossible, as my computer is guaranteed to be the only motherboard with a PCI-e slot where I live; hooray for 3rd World!).

 

[Qbah]

CPU-Z shows the PCIe link width at x16. I haven't tried using a different PCI-e slot, mostly because my motherboard recommends using PCI-e slot 1 for the main VGA. I've checked and double checked the cables; everything is connected properly.

The CPU overclock is stable. Like I said before, I've left Prime95 running for hours with no issues at all. The machine was working for 4 months prior to me acquiring the card; I put it together myself. I've been using a Diamond HD5870 all this time with no issues at all.

I'm gonna fire up a 3D Mark Vantage bench since my card crashes there once it hits the 2nd test. I'll log the results on a text file using GPU-Z and post them.

I just tried getting this card to work on a separate drive with a different installation of Win7 Ultimate 64 and fresh drivers with no luck. About the only things I haven't tested yet are 1) a different OS, 2) a BIOS flash (maybe that would solve fan control? MSI's Live Update tool insists I got the latest BIOS though), and 3) plugging the card in a different system (pretty much impossible, as my computer is guaranteed to be the only motherboard with a PCI-e slot where I live; hooray for 3rd World!).

I'd say try the slot change anyway. You can then have a closer look at the card. Perhaps some of the signal tracks are cut? Something's not making contact with the PCI-e slot? Pull the card out and inspect it. Also, check the voltage drop - though since the OC was stable, I'd say the PSU is fine - worth a check though! You should have some monitoring software together with your motherboard. Just run something in 3D windowed so you can see current readings.

 

[Ashkael]

Here is the GPU-Z log while running 3d Mark Vantage. This time, I actually managed to get through about 30% of GPU Test #2 before I crashed.

 

[Kenmitch]

I guess if the fresh install on another drive didn't fix it then at least you know it's hardware related.

Things you could try possibly:

MB bios is it current version....May or may not help tho
Video card bios reflash.... Possibly corrupted
Try different power connectors from your power supply....Possibly a bad connection

You can find 5970 bios here http://www.techpowerup.com/vgabios/...ufacturer=&model=HD+5970&interface=&memSize=0

Of course you'd have to do some research on how to flash if you have never done it before. Best would be a bootable USB stick.

 

[Ashkael]

I guess if the fresh install on another drive didn't fix it then at least you know it's hardware related.

Things you could try possibly:

MB bios is it current version....May or may not help tho
Video card bios reflash.... Possibly corrupted
Try different power connectors from your power supply....Possibly a bad connection

You can find 5970 bios here http://www.techpowerup.com/vgabios/...ufacturer=&model=HD+5970&interface=&memSize=0

Of course you'd have to do some research on how to flash if you have never done it before. Best would be a bootable USB stick.

My mobo's BIOS is updated to the latest version.

As far as reflashing my video card's BIOS, well, yeah. My card's BIOS is currently the same one as the MSI and ATI ones listed on that link. ATIWinflash is not letting me flash because it insists I got the same version, so I guess the only way would be to flash it the real way using a bootable USB drive. I don't think I want to go there just for a chance it might work.

I was doing some digging around and found talk about people having the same problem I have with the fan control with Sapphire's 5970. They solved it by flashing the BIOS. Hell, even Sapphire themselves recommended doing so. Sapphire's version of the BIOS is more recent though.

I think I'm just gonna give up and just try to work a RMA with the person that helped me buy the card.

 

[Kenmitch]

As far as reflashing my video card's BIOS, well, yeah. My card's BIOS is currently the same one as the MSI and ATI ones listed on that link. ATIWinflash is not letting me flash because it insists I got the same version, so I guess the only way would be to flash it the real way using a bootable USB drive. I don't think I want to go there just for a chance it might work.

Bootable USB device is the safest way to flash the bios.

I was looking for the post but I thought I read somewhere here that another person had similar issues with this 5970. Turned out that disabling the onboard sound in bios on his MB cleared them up. Worth a try at least.

 

[MarcVenice]

In all honesty, with all the artifacting going on even when it runs semi-stable, i'd say it has a bad memory-chip. RMA is probably your only option. It sucks, but it's better then throwing it in the trash ...

 

[Piotrsama]

I'd suggest fresh install on a spare drive. Easiest way to see if it's the card or not I'd think.

This.
Or try it on another PC.

 

[Ashkael]

This.
Or try it on another PC.

I tried on a spare drive. I'm currently unable to try on a different PC because of where I live (small town in a 3rd World Country - good luck finding somebody ELSE with a PCI-express motherboard )

Anyhow, after more testing I think I've narrowed the issue to the card itself not being able to handle the temperatures it's supposed to be handling. I've been logging GPU-Z reports and the card hard locks my system every single time the GPU temperatures get around 72-75.

For a final test, I opened the case and positioned it very close to my AC unit set at the lowest possible temperature. I managed to get the idle temperatures all the way down to 40C. I then bumped the fan to 100% and loaded 3D Mark Vantage. This time I managed to complete GPU Test 1 and most of GPU Test 2 before the inevitable hard lock, which once again happened when the GPUs hit around 74. This time it took a bit longer for the GPUs to rise due to the open case + AC + 100% fan combo.

So yes, I'm chalking this one up to the card itself being defective. It's clearly capable of producing the FPS it's supposed to be producing, but it just can't handle the temperatures it's supposed to be running at.

Anyway, I've contacted my seller and the card is taking a long trip to back to the US. It will take a bit, but I'm going to get it changed for a Sapphire 5970 this time. Hopefully I get a working unit this time.

Thank you very much to all the peeps that provided input and info.