• 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.

GTX570 issue or PSU issue? or Other?

iB CoMpL3XiTy

iB CoMpL3XiTy

iB CoMpL3XiTy

Associate
Joined
10 Sep 2010
Posts
1,533
Location
Cornwall, England
Right Let me start by listing my PC spec's as I think my issue may be PSU related.

Core i7 920 @ 4.2Ghz 1.360V
POV GTX570 Beast (Factory overclocked)
POV GTX570 Beast (Factory overclocked)
3 x 1GB OCZ 1333Mhz 8-8-8-24 @ 1200Mhz 6-7-6-18 1.68V
Gigabyte GA-EX58-UD5
Samsung Spinpoint F3 500GB
Lamptron FC6 Fan controller
10 X 12V Fans
18W Laing DDC Pump
10W Laing DDC Pump
Corsair Obsidian 800D
Corsair HX850
Custom Water For CPU and first GTX570.

Ok So I recently decided to buy a second GTX570 for SLI, When I first booted into Windows with both cards installed I noticed that I didn't have any option to enable SLI (Although the SLI bridge was installed), I didn't think much of it and decided to uninstall the nVidia driver, reboot and install the latest driver, but I still didn't have any SLI option.

Now because the first card is watercooled, its not easy to swap the 2 cards around to check if the new card was working properly (would involve draining the loop, uninstalling the card, installing new one, leak testing for 12 hours...)

So I went into the BIOS and to my surprise there's an option in there called "init Display first" which allows you to tell the BIOS which PCI-E Lane to boot from, So I chose PCI-E x 16 (2) which allowed me to boot from the second card, PC booted fine and when I got into windows I was able to enable SLI, However I noticed in GPU-Z that the BIOS for the new card was "unknown" and in Device manager there was a warning sign next to the new GPU.

I decided to do some testing, I ran Unigene Heaven V2.1, about 15-20 seconds into the run my PC crashed, although I could still hear Unigene's music... I later decided to run MSI Kombuster which also failed after about 4 minutes.

The next day I checked GPU-Z and to my surprise it was reporting the BIOS correctly, and Device manager no longer had a warning sign next to it.
I then set the fan speed of the new card to 100% to elimate any possible heat related issues and ran 3DMark11, 3DMark Vantage, Metro 2033 Benchmark (3 runs), and Unigene Heaven 2.1 (3 runs) without issue, So at this point I thought that I had solved the issue by setting the fan speed to 100%

Later that day I decided to play some metro 2033, after 10 minutes my PC shut down....

I then decided to disable SLI and just test the card on its own, So far I've played about an hour and a half of metro 2033 without issue...

Sooo.... thoughts and opinions please!

Thank you for reading my ridiculously long post.
 
Are u using the sli bridge that came with the board

Are u using latest motherboard bois

Use msi afterburner to check temps and voltages to get a clearer picture of what's happening when under load
 
Yes I'm using the SLI bridge that came with my motherboard

Yes, I'm using the latest motherboard BIOS

VID for both cards is 1.050V, haven't changed this at all, Temps for watercooled card are 26C idle and 35C under load, and for the Air cooled card idle temp is 46C and 58C load (100% Fan speed for load temp)
 
My Guess, PSU not up to the load.

1 card will run fine, and as the 2nd is not under load you will not experience an issue, but almost definately something in your psu has said enough.

I had a similar issue with 2 4870's in xfire with an overclocked 965 BE @ 4.0ghz on a 1000w psu, just started shuting down or not detecting 1 card. So my guess is a rail went down.

Very similar symtoms you describe.

Also try each card individually in the primary GFX slot, if they both play nice it's defo new PSU time.
 
hmmm ok thanks for the reply

So do you think the PSU was never up to the job, or has become faulty?

I'll do some more testing later, maybe reset my CPU overclock to stock so that it uses less power, then test SLI again....
 
Cool Breeze said:
My Guess, PSU not up to the load.

1 card will run fine, and as the 2nd is not under load you will not experience an issue, but almost definately something in your psu has said enough.

I had a similar issue with 2 4870's in xfire with an overclocked 965 BE @ 4.0ghz on a 1000w psu, just started shuting down or not detecting 1 card. So my guess is a rail went down.

Very similar symtoms you describe.

Also try each card individually in the primary GFX slot, if they both play nice it's defo new PSU time.
Click to expand...

I disagree... an 850W PSU should be fine for 2 GTX570's which are fairly low power.

How do you have the power connected, are they all PCi-e leads from the PSU or are you using Molex to Pci-e

Edit, mind u - the 2 pumps and 12 Fans will be adding to your load and the CPU overclocked to 4.2 will be drawing a lot of power.
but if you are taking power to the GPU from a molex and that lead is driving a pump it maybe maxing out that circuit.
 
Last edited:
SimonMaltby said:
I disagree... an 850W PSU should be fine for 2 GTX570's which are fairly low power.

How do you have the power connected, are they all PCi-e leads from the PSU or are you using Molex to Pci-e
Click to expand...

read the post man, I illude to the fact that it has failed, not that 850W is not man enough.

Jeez:rolleyes:
 
The CPU, motherboard and the 4 PCI-E 6 pins from the PSU have sleeved extension cables, however I don't think this would cause a problem, Not using any molex to 6 pin adapters.
I've done some rough calculations:
Core i7 920@ 4,2Ghz 1.360V = 250W
GTX 570 Beast Ed = 230W (Slightly more than ref design because of overclock and high VID) x 2
Mobo + RAM = 100W ?
pumps = 28W total
Fans = no idea, maybe something like 8W each? = 80W
HDD = 20W?
Total = 938W

I'll reset my CPU back to stock which would put it at roughly 130W, and a total of 818W and then test again in SLI
 
Ok so I reset my CPU back to stock and then played about 40 minutes of metro 2033 before getting bored, I then decided to let Unigene heaven run on absolute max settings, it lasted almost 2 runs before the monitor lost signal (could still hear unigene running)

Sooo... still at a loss here chaps.
 
Have you tried using the Antec Power Supply Calculator?

http://www.antec.outervision.com/index.jsp

I did a rough estimate on your rig details given to us and came out with approx 820W :eek:

Sadly, I could not see on the PSU calculator a choice to show overclocked speeds for your GPU's (as your 570's are OC'd?) which will surely increase the wattage further :eek:

You have a go and see what wattage you come out with ;)
 
Last edited:
Just did that PSU calc on my system, and it came out as 940w, when Im running furmark and Prime on the other 3 cores the most ive ever seen my system draw from the wall is 936w, so take efficiency into account I can only draw about 748w max.

So that calc is a little off, but obviously it needs to allow a bit of headroom etc.
 
iB CoMpL3XiTy said:
Would a faulty PSU cause random crashes whilst under load? Not really sure how to test further, any advice or opinions are welcome

Please help!
Click to expand...

When you really push a PSU hard the heat can effect the efficiency badly and then everything goes to hell.

It could be that the supply is struggling and your getting big fluctuations. I would pull one card and and stress test.
 
Ok thanks for your help guys, MUCH appreciated

I filled in the Antec PSU Calc and for 10% cap aging (lowest amount selectable) it said 896W recommended minimum, and for 20% cap aging it said 973W minimum recommended...

With no CPU overclock and only 10 cap aging (best case senerio) it said 831W minimum recommended.

I don't know how fast the caps will degrade, I've had my PSU for about 1.5 Years now and my PC is on for about 4-5 Hours a day...

Each card works fine on their own, and I don't get any crashes when I'm playing a game that has a FPS Cap (like NFSHP & CoDBO MP) and so limits the GPU usage (GPU'z at 52% load each in NFSHP and about 30% each in CODBO MP). I take it that less GPU usage = less power usage?
 
Do you possibly have a second PSU you could run one of the cards off? You could use that to determine if it's your current PSU or not, sounds like it is since they both work by themselves.
 
Unfortunately I don't have another PSU : /

You guys have any recommendations for a PSU, preferably 1200W+ ? I guess I'll be selling my current HX850 on the bay.
 
All of the 1200w on OCUK are good, so just pick one you like the price/look of, read a review to check it does everything you want etc and all cables are long enough and whatnot, then buy one.
 
You must log in or register to reply here.
Back
Top Bottom