Sign in with
Sign up | Sign in
Your question

EVGA GTX660 SC - Voltage problems?

Last response: in Graphics & Displays
Share
December 9, 2012 1:05:40 AM

First of all, I'm not a great fan of asking for help, but I've been trawling the net for about 2.5 hours for an alternative answer to "buy a new power supply".

I recently bought an EVGA GTX660 SC, I had a Radeon HD4850 before this and an X1950XTX before that.

When I bought my 1950, I had to upgrade my PSU but couldn't as I was using a micro atx case. So I opted to use a 5.25" Supplemental PSU (http://www.fspgroup.co.uk/u.k/02_newproducts/BoosterX3....)

It's done me well ever since, now on to the problem at hand, the first thing I noticed when I got the GTX660 was the single PCI-E connector, but I figured my GFX PSU would have enough juice. I believe I'm wrong.

At first I was just getting random driver crashes, then artifacts on screen along with sometimes full system crashes when playing World of Warcraft, I changed all sorts of settings but anytime it done anything remotely graphics intensive it'd crash the driver, screen would go black and 4 - 5 seconds later it'd be back up again. To replicate the issue, I've taken to running FurMark.

Run a stress test for a little while and sure enough artifacts come on screen, lock up, screen black, something something kernel driver has crashed. I've tried underclocking it back to stock 980MHz, I tried giving it as much voltage as I could, eventually with the help of nvidiaInspector's graphs I noticed as soon as FurMark was stressing the card, the clockrate would dip and so would the voltage. Making me think it's power related.

Long winded as hell, but really my only question is... Would it be possible to take the second PCI-E 6-Pin connector from my supplemental PSU and use some sort of "joiner?" to make dual connectors a single connector to run the card?

Original question about a cable to combine 2 PCI-E 6 pin to 1 PCI-E 6 pin still stands, but I just done more underclocking, the card is still crashing even with the lowest possible settings, -549 MHz on the Base Clock, -2194MHz on the Memory, 110% Power Target and -112.5mV on the voltage. The TDP was only around 65% when it crashed too... I'm starting to wonder if I have a bad card. :( 
December 9, 2012 11:09:19 AM

Edit 2: Decided to say bugger it, who needs money right? Just got a new 450W XFX PSU, 34A on 12V, tried the burn in again, crashed after about 10 seconds, no drop in voltage, no drop in speed. Artifacts only in FurMark's screen then black and crash.
- Could it just be I have a screwed card? Because I'm really getting pissed off with this nVidia crap.
December 9, 2012 11:23:46 AM

Its not the psu. I have that same FSP gpu psu. Sounds like a bad card to me. RMA.
Related resources
December 9, 2012 11:28:02 AM

Excuse me, but i am honestly extremely unclear of your system power supplies. Could you reply with a COMPLETE system spec? Specify your current power supply u have in your system.

That graphic card needs a 500watt power supply to run. Yes you could use that power booster thing you have, but i recommend you using a single power supply.

I have a suggestion for you. How about you use your xfx 450 watt power supply to power the entire system EXCEPT FOR THE GRAPHIC CARD. Then use that add on power supply to power only your graphic card. FSP is a good brand and that add on psu will be enough to drive a single gtx 660
December 9, 2012 11:51:59 AM

Motherboard: Asus M4A78LT-M
Processor: AMD Athlon II X2 250 3.0GHz
RAM: 8GB Kingston HyperX DDR3-1600
HDD: Sandisk 128GB Pulse SSD & 2x1TB SATA-II Storage Drives
GFX: EVGA nVidia GTX660 SuperClocked Edition (2GB GDDR5)
PSU: XFX PRO450W Core Edition /w potential 300W Supplemental PSU (FSP BoosterX 3)
OS: Windows 7 - 64bit Enterprise Edition

- I've just uninstalled all drivers, used driver sweeper and a registry cleaner in safemode. Gonna reinstall 306.97 drivers from nVidia's site and hope to hell it'll work. But I'm doubtful.

I'm just a bit frazzled that the first time I've had an nVidia card in about 10 years has decided to give me this much hassle.
December 9, 2012 12:10:31 PM

Decided to install 305.27 because that's what came with the card on the driver CD, still same problem, luminous green artifacts over the whole screen now and black screen crash back to desktop with a driver recovered error.

TDP was at 101.9% when it went, but this doesn't seem to matter, as I've seen it under 100% when it happens, temperature isn't an issue, it's crashing at around 45 degrees.

Just consistant high load on the GPU that's causing these crashes, there's a few other things I'll try, I've got a spare computer I'll try and see if I can get it to run in that, but it is a lot older and weaker, so I'm not even sure it'll work. Other than that, I could try and disconnect all other devices apart from my SSD and graphics card.
December 9, 2012 1:45:12 PM

Found out what the problem was, I'll be returning the card to ebuyer or getting it RMA'd.

Running it in a completely different computer system didn't solve the problem, what did was after it had crashed, I didn't boot up the overclocking software again and just left it at default 324MHz memory clock, ran FurMark again and also EVGA OC Scanner, no crashes now. Looks like when I underclocked the memory before I only brought it to around 800MHz. So it seems like a hardware fault with the memory on the card.
!