Diamond 5970 Locking up computer

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Hey everybody,

Just build a new rig and everything seemed to be working fine. I decided to do some stress testing to see if things were up to par.

My Diamond 5970 is having some issues locking up. I found a post specific to Diamond cards but am awaiting being activated as a member on that forum so i'll go ahead and say something here. I'm gettin worried I got a defective card!

1. I cannot run Crysis for more than 1 minute before the screen freezes and completely locks up my computer requires a hard reboot. The audio continues to run however when the computer freezes. Ctrl-alt-delete, nothing works I have to reset on the computer.

2. I tried some other games, CS:S worked fine for the 15 minutes I played it, and Oblivion seems to work fine with some graphics mods, only played for a few minutes though.

3. I was trying to get a little more power out of the card for Oblivion, so I read up on overclocking. I changed the GPU core voltage to 1.1625v from the stock of 1.05v or so. Did the first Furmark with this voltage, got same results and stock.

4. This is where it worries me, I cannot run Furmark in 640x480 for more than 50 seconds before the whole computer locks up and I have to do a hard reboot. Both times I did Furmark the GPU was running at stock speeds, got up to 64C, about 55 seconds in and the whole computer locks up!

I spent $670.00 on this card, please help me out! I don't know what to do!

Specs:
i7-920
Asus p6t Deluxe v2
Diamond 5970 2gb
6gb OCZ Ram
Windows 7 64bit

Thanks

-J
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Hey,

I must have mistyped, never did any actual overclocking per say. I was researching it, because I wanted some more oomph in Oblivion. This is running pretty much every graphics mod there is, plus some, to answer your question Kithzaru. Without mods I get about 150+ fps.

That aside, I jumped up the voltage to 1.625 from the stock of 1.100v of the core GPU clock. Nothing else. Ran Furmark and got lock up after a little under a minute, temps were at 64C at the time of lock up. It's not a temperature issue, unless the readings are somehow wrong, but I know the Antec 1200 has adequate cooling.

Had to hit the restart button on my case after that, whole computer locked up. I made sure everything was running at stock speeds, stock voltage after restart (never used Overdrive, just MSI Afterburner to raise voltage a bit.) And ran Furmark again, computer lock up at 55 seconds GPU temps at 64C. Had to do a hard reboot.

I do not lean towards ATI or Nvidia, I know the temperatures are fine, and I am not overclocking at the moment. I just want to figure out what is up with my card, cause I shouldn't be getting lock-ups like this with a brand new 5970.

-J
 

daedalus685

Distinguished
Nov 11, 2008
1,558
1
19,810
Assuming that you missed the extra 1 in the 1.165... why are you upping the voltage at all?

You are more than likely toasting your vregs.. the cooling in the card is not sufficient at stock to cool them during something like furmark if the voltage is increasing. However they should simply throttle the card and not crash, but one never knows.

The reason these cards run into OC limits is the Vreg cooling, not the GPU temps.
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Yes, I meant 1.1625v, but, your missing the point. I am NOT overclocking. I never did anything other than experiment upping the voltage to that number for a minute or two. I was using this guide here - http://www.guru3d.com/article/overclock-radeon-5970-guide/4 .

I am no longer overclocking. I will be tonight, testing the CPU/RAM with the Blend test in Prime95 to see if it is a cpu/ram issue which I don't think it is just to double check.

My brand new 5970 should not be locking up my computer running Furmark 640x480 after 55 seconds when the GPU is only at 64C.
 

Kithzaru

Distinguished
Apr 23, 2009
383
0
18,790
Unless upping the voltage hurt the card.

Still though, normally what I would do is double check drivers and such first. Or, maybe in your case if you have another ATi card you could shuv in there temporarily and see if it locks up?
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Don't have any other cards. I do have a slight update though.

In the ATI Overdrive menu, I checked manual fan control and slid the slider up to 100%. Meter on the right showed the fan go up to 11% for a split second then back down to 0%.

Also, the card is apparently currently underclocked to 157mhz gpu/ 300 mhz memory clock says the Overdrive menu. Overdrive itself is not activated however. The voltage was changed for a split second before, and nothing has changed since then, the small amount of upped voltage did nothing that wasn't already there.

Fans not working? (I can see the fan moving in my case)

Is there a way to disable the underclocking?
 

daedalus685

Distinguished
Nov 11, 2008
1,558
1
19,810


Upping the voltage still ups the temperature on the vregs, which is what fails when furmark runs. I did not say you were overclocking, calm down son. The GPU temperature is important, surely, but during furmark you have to watch the Vreg temperature, they should never get above 120, that is the throttle threshold.

You should not play with things like that if you do not understand them, you could have easily destroyed one of the vregs doing that and not even have a card that stalls on you.

Furmark crashes ATI cards, it always has. They supposedly corrected this by adding throttling in the 5000 series but it does not always kick in in time, the card crashes to save the voltage systems. Furmark is not an ideal test for GPU stability as it is FAR more intensive than anything you could dream of using. At stock it should not freeze though, as the cooling should be sufficient to run it, but with any overclocking (or even just a voltage bump with no overclock) it may not work and if the card is not throttling running it for any length of time with higher voltage coudl damage the card, since it is supposed to throttle this would be a flaw though.

Your card is probably faulty. If I were to guess I would say that the cooling is not contacting the voltage regulation properly, but it could simply be broken. You should RMA the card, but it would be best not to tell them that you changed the voltage as Diamond has terrible support as it is and doesn't need an extra reason to not replace it.
 

daedalus685

Distinguished
Nov 11, 2008
1,558
1
19,810


The underclocking is a power saving feature. You have to modify the cards bios to remove it.

By the by, ANY voltage modification can destroy electronics, whether for 10minutes or only 10 seconds. Too much is too much. Mind you it should be impossible to destroy a 5970 as there are hard caps in the voltage (unless something is funny with the diamond card) unless it is faulty to begin with. That being said, it still voids your warranty to do it even for a second.. which you should be careful of given the horror stories of diamond tech support.

Use GPUz to display the info as it will display current clock as well as set clocks, and should give a reading for the important thermocouples.
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Well that's good to know about the voltage. I figured it was safe as the guru3d guide said to do it and it was just a small adjustment. I keep reading more and more of people getting faulty 5970's, especially with so few of them being created, I guess they just rushed them out the door without proper testing.

May have to RMA, but with so few being made, when will I get my next one? A month from now?

Unfortunately I have no graphics card to replace it at the moment.

I'm going to try wiping the drivers w/ driversweeper and re-install the 9.12 without the hotfix.

Any other programs to use to check if my GPU is working properly, or not?

Thanks

-J
 
1. What is your PSU? Your PSU might not have enough juice to run it.

2. What are your temperatures?

3. The 5970 was specifically created and underclocked so it would draw no more than the 300watts of power provided by 2 six pinned PCIe connectors and the PCIe slot. Increase the voltage will increase the power draw to far over 300watts.
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
1. 850watt Corsair, supposed to be enough for two 5970's

2. Idle GPU temp 44C, when I get locked up in Furmark temps are 64-70C

3. Yeah, voltage is stock

4. Fan speed is stuck at 30% says GPU-Z. Regardless of what the fan speed is actually at, I CAN see it running, but cannot change it.

There's a large thread on Sapphire sending out upgraded BIOS to re-enable fan control for those that bought 5970's from Sapphire.

Diamond seems to fail at customer support around the board. I couldn't even register for an RMA, it kept saying I had not filled out all forms when I clearly had.

Does anyone know how long an RMA for a 5970 would take?!

I could imagine it taking months since there are so few, hell, you can't even find the damn thing on eBay anymore.

If you RMA to Diamond, do they send out the first cards they get to the people that RMA, or send them out to newegg or similar sites?

If I end up having to RMA it is really going to suck having just built this new computer. I won't have a graphics card for god knows how long.

Bah, very disappointing is a all. Had to vent a bit. I've read of many RMA'ing their 5970 and getting another faulty one, a few got two and even three.

I don't know.

-J
 

Kithzaru

Distinguished
Apr 23, 2009
383
0
18,790
Is it still possible to simply return it? I know it would be a downgrade, but with that many issues, maybe dropping to a 5870 would work out better for you.

Vent all you want, I don't think anyone here would blame you for being upset spending that kind of $$ and having this happen.

I'm not sure where you ordered your card from, but I think ZipZoomFly still has a few 5970's from XFX and HIS.
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
Strange as I ordered it from Ebay. Ended up getting a great price for it. It came in a newegg box pre-packaged un-opened. Not sure if it goes back to Newegg or back to Diamond. I THINK it only goes back to Diamond. I am going to try to get in contact w/ their customer support today.

I can't find a 5970 anywhere, except on Ebay.
 

Jamrock

Distinguished
Nov 16, 2009
233
0
18,690
So, I threw a 9500gt 1gb in and took the 5970 out. Installed the drivers and as we speak I am running the same Furmark test as I was with the 5970.

Same exact settings, been running it for 3 minutes now and the GPU temp is only at 55C and has been there for the past minute.

The 5970 would go from 41C-65C in just under a minute and lock up the computer. It got 200 (yeah, 200) more FPS while running Furmark, but nothing you can do about a faulty card.

I am going to do one more test, putting the 5970 in the 2nd PCIe slot as my 5970 was practically touching my auzentech 7.1 prelude before.

200 frames more than the 9500gt.... that's just wow. Here's to hoping Diamond responds to my e-mail in a timely fashion. Been reading stories of them taking a week or two to respond and giving an automated answer (i.e. did you check your cables etc.)

Anyway, thought an update would be nice. Since I got the card off Ebay I still don't know whether it should RMA to Newegg (it came packaged in a Newegg box, so I figure it was bought from there, the ebay seller has yet to answer my question on that) or to Diamond. I would bet that Newegg would be 10x faster, but I don't know because the 5970's are in such low stock what they do when they come in. Whether first priority goes to people who have RMA'd or to put up for sale.

I'd like to avoid Diamond at all costs if possible.