Sign in with
Sign up | Sign in
Your question
Solved

EVGA GTX660 SC constant GPU failure, App & Sys crash

Last response: in Graphics & Displays
Share
January 11, 2014 11:42:52 PM

Calling all graphic & display experts, specifically on the 'voltage' and 'clocking' side of things. I appreciate all the help and edumacation I can get from everyone.

System specs:
GA-P55USB3 rev.1 (which i plan to upgrade to a PCIe 3.0 mobo asap)
i3 LGA1156 @ 3.0 GHz
OCZ Gold series RAM overclocked @ 1600MHz 1.65v
Corsair TX850W PSU (less than 1yr. old)
VGA listed in thread topic @ factory settings (less than 1yr. old)
OCZ Vertex3 SSD w/ fresh Win7 x64 install

Alright, so this card seems to be defective in the sense that it's failing whenever it's put under stress, and sometimes not at all. The display will die, flicker, come back to life -or- the app will crash, the system will lock up entirely displaying only a 'colored' screen of whatever color it chooses.

These are some of the apps running in full screen 1080rez. that are seemingly causing my GPU to fail:
Metro 2033 (1st title)
CoD MW2
Media Player Classic

I started monitoring the GPU while playing CoD to see what was up. This is what I got after 30 mins:
GPU Temp: 45C min - 67C max
Fan speed: 37%/1290RPM
True Current Clock:
-Core: 1188MHz (base clock: 1045MHz / boost clock: 1110MHz)
-Mem: 3004MHz
-VDDC: 1.175V
GPU Power: Avg. 30%-70% TDP
GPU Load: Avg. 50%-99%
Limiting Policies (NVIDIA): GPU voltage limit reached (occasionally, then failure)
**that last one sounds problematic to me**

I've already gone through the whole driver song 'n dance, and it's clearly not a software issue. I've also tried disabling that 'Powermizer' jazz and that did jack-squat.
I'll mention that I'm a certified PC tech and that my system was running like a dream until I put this *NEW* video card in, especially since the fresh OS on a new SSD.

I guess what I'm looking for, is a bit of advice regarding the clock/voltage settings and if I should tweak them to potentially fix my problem. OR I just get on the horn with EVGA and attempt a warranty return/refund while trying not to throw my phone and computer out the 3rd floor window.

Best solution

January 11, 2014 11:58:17 PM

you should check your power cables to make sure they are secure and reseat the gpu
could fix the problem
also you can download evga precision and tyr overvolting http://www.evga.com/precision/
then test the system again
if not then call evga
also it would be better if you can run metro 2033 and post a gpuz screenshot if it doesnt freeze on you
Share
January 12, 2014 12:30:29 AM

Thanks, I'm checking that stuff and will try overvolting, but I would want to avoid frying the card, voiding the warranty and pissing $220 out the window. Will also grab a screenshot of Metro 2033 specs while under load if possible.
By "reseat the GPU" are you talking about physically removing the processor and reseating it like you would a CPU?? ..or reseating the card in the PCIe slot?
m
0
l
Related resources
Can't find your answer ? Ask !
January 12, 2014 9:05:31 AM

not the processor just take the `gpu out of the pcie slot
then just put it back
you have evga if you ask them they encourage overclocking
i actually have 2 gtx 760s
i overclock the crap out of mine and they still rma'ed it
m
0
l
January 18, 2014 1:23:25 AM

Yeah, so I tried all that jazz and in the end, I've noticed that changing the voltage settings using PrecisionX voltage tuner tool makes a considerable difference in "not crashing" my shit.
What I still don't fully understand (and maybe signifies something wrong with the card) is that the card @ stock settings wants to spike the voltage to 1.150v and above, as high as 1.175v. This is where the instability happens and it's not temp related.
When I use the voltage tuner, I found that anything @ 1.10v and above is trouble, so I ran it on lower voltages. I was using FurMark benchmark tool to test diff levels and that's as high as I could get it without crashing.

Does this seem normal to you for a 660 class? I guess I'm at a better position to call EVGA now and see if those values add up to spec.
m
0
l
January 18, 2014 1:32:07 AM

can i know of you rm'ed it or just overvolted to keep the card i would personally rma the card
m
0
l
January 18, 2014 2:30:49 AM

Yeah, I haven't called them but it's going back. I can't even load a game in MW2 anymore without it crashing and locking up my system. Pretty f'ing annoyed.

Thanks for all the advice. They had better send me a proper working card or no more evga for this guy.
m
0
l
January 18, 2014 11:03:20 AM

ok cool good to know it was just a bad card and not something else with the computer
m
0
l
January 19, 2014 1:58:37 AM

Yeah, it's definitely just the faulty card. I put my original Radeon HD5770 back in and no failures. Best way to isolate the problem. I'll be upgrading the mobo + CPU soon, but they run 100% stable since the original build.

This is kinda off topic.. but what would you recommend for custom GPU cooling? During testing, my card would reach 80C+ and I want to avoid that completely. It puts too much wear on the electrical components over a short period of time.
m
0
l
January 30, 2014 3:07:47 AM

Yeah, would be for the same card once they return a "working" one back.. I can't believe EVGA would let a high performance GPU reach temps of over 80C (under standard, non-oc use) without adequate cooling. I mean.. wtf. Maybe that's not as big a deal as I think it is...

Thanks for all your help/advice.
And to anyone experiencing the same problem, just RMA the damn thing. EVGA doesn't actually make a big deal about it. If you can simply swap your video card with another working one and your problem disappears.. blam. Problem isolated. RMA the defective junk. Amen.
m
0
l
January 30, 2014 5:39:44 PM

i cant asure you that you would receive the same model maybe a better one
but you should wait before ordering the cooler
m
0
l
February 2, 2014 7:38:44 PM

Yeah, I'm gonna wait for sure. Now that I know the vga will be replaced, I'm gonna upgrade the mobo+cpu. The cooler will be after. I don't intend on burning the piss out of it, playing games for 8+hrs straight. I just want it to run as cool as possible.
Any recommendations for a standard Z77 mobo? (with i5 cpu most likely)
m
0
l
February 2, 2014 7:52:08 PM

i would recommend the i5 3570k best bang for the buck you could even get the none overclock able version which would be cheaper but the overclock able one would benefit in the future
m
0
l
February 3, 2014 11:57:58 PM

Cool, I would consider buying that one if I jump up from an i3.
But what about the mobo? Stick with Gigabyte or maybe go with an ASUS, etc?
m
0
l
February 4, 2014 1:59:28 PM

asus have very good mobos as do gigabyte that is all on perference and what you want on the mobo
m
0
l
February 10, 2014 11:55:08 PM

Noted. I'll chose one based on the components and BIOS management.

Get this: EVGA is sending me a 660Ti replacement instead of the original 660SC. They don't even have the 660Ti listed on their website anymore? I'm not really sure how to feel about this...
m
0
l
February 11, 2014 1:38:28 PM

its fine the gtx 660ti is probably refurbished
which means it will be in good working order when you get it
if not called them back up and tell them that you want expedited shipping if its bad
m
0
l
February 14, 2014 2:04:43 AM

Yeah they sent me a used 660Ti. I can tell because it's all scuffed up and the fan makes a humming noise. Are they f-ing playing me?
My card was only 9 months old, in pristine condition. It was defective from the manufacturer, not from overuse & abuse... and they try to replace it with a used, refurbished product that has apparently been removed from their product line. This is how evga handles warranty coverage?
Oh yeah, and my display is flickering to black and on again 3+ times.

They're gonna get my foot up their ass, plain n' simple.
m
0
l
February 14, 2014 6:59:56 PM

Just call them and tell them this time it will better
hopefully they give you a gtx 670 now
m
0
l
!