• 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.

gtx 670 help

Associate
Joined
8 Jun 2009
Posts
515
Location
London
Hello guys,
I really need some assistance because I am not sure what to do now.

My system is:
  • i7 3770K @ 4.6 GHz @ 1.275v - stable 12hrs prime95
  • samsung green ram @ 1800 mhz
  • asus p8z77v pro motherboard
  • corsair atx 850w psu
  • crucial m4 ssd
  • 1920x1080 res

I had two 5870s before in crossfire, which worked perfectly fine, but I decided to upgrade to a single card, and bought Gigabyte gtx 670 windforce.

Received a card, even made a fresh windows install, latest drivers (301.42) and libraries - but the card performed extremely slow. In heaven, when I launch the benchmarking it starts with about 40 FPS, and gradually drops as it goes through the scenes, so by the end of benchmark it is crawling, freezing and stutterring at like 10 fps. Moreover, if I try to launch the benchmark again right after, it will start at those 10 fps and crash shortly after. The same thing happens in 3dmark and several others. Same happens in game - it starts diablo with 150 fps, 3 min later its at 40 fps, 10 min later its unplayable at 8-15 fps.

First I thought maybe beta drivers or something, so sweeped and installed older ones from the disk supplied - same thing. Then I thought maybe windows messed up, and performed another clean install of a different windows 7 distributive - still same thing. Performs a lot worse than even a single 5870.

So then I thought its RMA time - I could just DSR it, but I wanted the card so I asked for a replacement. But here is what I got:

The product GX-095-GI on RMA220449 has been returned to you, this will be sent on a new order with order number 4031533.

Here is the technicians Test Report:

"Graphics card passes all Tests with no issues, Tests Run: Crysis WarHead, 3D Mark 11, Furmark, 3Dmark Vantage, AVP, Lost Planet 2, Stone Benchmark MSI Kombuster, Resident Evil Benchmark, ATI Mecha, A"

(Graphics card passes all Tests with no issues, Tests Run: Crysis WarHead, 3D Mark 11, Furmark, 3Dmark Vantage, AVP, Lost Planet 2, Stone Benchmark MSI Kombuster, Resident Evil Benchmark, ATI Mecha, ATI Ladybug, Streetfighter Benchmark, Stone Giant and heaven benchmark tool, these were all tested at 1920 x 1200 resolution running maximum settings, tested on DX9, DX10 and DX11 where applicable, there was no issues apparent.)


OCUK shipped me the same card back. So apparently the card is fine and fault is on my end. I just received a card, sweeped off ATI drivers with driver sweeper (was running old 5870 while RMA), restarted, installed new nvidia drivers, restarted - same thing...

Here is a heaven screenshot:
n0Asu.png
This is about 4 times worse than everyone else with 670 is getting. Funny enough, the card is not even overclocked, if I try to boost the core by even small 10 mhz - it crashes the bench instantly.

So yeah guys, I don't know what to do :( Any suggestions what should I try?
Sorry for such long wall of text, and thanks in advance.
 
Last edited:
Associate
Joined
14 Oct 2004
Posts
979
Have you got any monitoring software installed? Your FPS would be about right for the lowest power saving mode. There have been reports of cards being stuck at this setting.
 
Associate
OP
Joined
8 Jun 2009
Posts
515
Location
London
Yeah, I've tried all available pci-e slots, but no difference.
Games perform the same as benches, starts diablo with 150 fps, 3 min later its at 40 fps, 10 min later its unplayable at 8-15 fps.

I have afterburner yeah, should I do another test and post a log here?
 
Soldato
Joined
17 Jul 2005
Posts
9,683
Using Afterburner can you see what temperatures the card is running at.

Normally a drop in performance such as you are seeing is indicative of thermal throttling (card clocking down due to temperature).
 
Last edited:
Associate
OP
Joined
8 Jun 2009
Posts
515
Location
London
Well, it's corsair 850w, and it was good for 2x5870 with a 280 for physx for about 2 years, so I would assume it should be fine for 670.

I've just played Diablo until it crashed (about 8 min) - uploaded a hardware monitoring log from evga precision - http://www.sendspace.com/file/gn5mro

It seems like it randomly goes to power saving mode and drops voltages and clocks
Temps are at low 50s
 
Man of Honour
Joined
8 Nov 2007
Posts
16,034
Location
Outer Space
But why is it dropping, if OCUK didn't have any problems with it? Can you try different power leads to the card, either different PCI-E cables from the PSU or use the molex to PCI-Es that normally come in the GPU box, this would use different rails on the PSU?

Also, was your RAM overclocked when you had the 5870s? Maybe run them at stock and the CPU for now, just trying to eliminate things 1 by 1? :)
 
Soldato
Joined
17 Jul 2005
Posts
9,683
Trying to narrow down the issue here:

1) Test with CPU at Stock.
2) Test with Memory at Stock (along with CPU at stock).
3) Test single and dual HD5870's to check their behaviour when gaming.
4) Try using different PCI-e cable combinations to rule out a damaged or faulty cable.

The following are only possible if you have access to them:

5) Test with different PSU - the fact Overclockers sent it back as fully functional points to an issue with your system rather than the card itself (although I wouldn't rule out a card issue at this point).

6) Test the card in a different PC / friends PC / Friendly Overclockers PC.

The symptoms you are experiencing suggest the PSU may be at fault (as the card passed testing at Overclockers). You have confirmed it isn't a temperature issue - although could you just confirm that this is the temperature noted during testing and not just when running at lower clocks (it could be spiking and the dropping as the card clocks down - I can't view your file here at work).
 
Soldato
Joined
18 Oct 2002
Posts
7,614
Location
Sutton Coldfield, Birmingham
Tried different drivers?

If it works on OcUK's system then it points to an issue with your setup. Worth doing a fresh install? This may seem a bit drastic but I tend to do a fresh install every now and again anyway as with SSDs the time it takes to install everything is hardly anything.
 
Associate
OP
Joined
8 Jun 2009
Posts
515
Location
London
Thanks for suggestions guys, gonna those that I can.
Here is an afterburner monitoring:

701b19f1d80bf8a1a23b70071a8ddf22.png


I've done 2 fresh installs already James, and different drivers as well.

1) Put memory and cpu at stock, heaven crashed at scene 14/26:
f61d490a0675b94da42df8fdfc598f11.png


gonna try recable with molex now
 
Last edited:
Associate
OP
Joined
8 Jun 2009
Posts
515
Location
London
Trying to narrow down the issue here:

1) Test with CPU at Stock.
2) Test with Memory at Stock (along with CPU at stock).
3) Test single and dual HD5870's to check their behaviour when gaming.
4) Try using different PCI-e cable combinations to rule out a damaged or faulty cable.

The following are only possible if you have access to them:

5) Test with different PSU - the fact Overclockers sent it back as fully functional points to an issue with your system rather than the card itself (although I wouldn't rule out a card issue at this point).

6) Test the card in a different PC / friends PC / Friendly Overclockers PC.

The symptoms you are experiencing suggest the PSU may be at fault (as the card passed testing at Overclockers). You have confirmed it isn't a temperature issue - although could you just confirm that this is the temperature noted during testing and not just when running at lower clocks (it could be spiking and the dropping as the card clocks down - I can't view your file here at work).

1) no change
2) no change
3) work perfectly in the same system (they draw a lot more power than single 670, and i've used them with i7 920 @ 1.4 volts for nearly 3 years, so it seems like psu should be sufficient?)
4) tried different combinations, tried supplied molex adapters as well, tried different pci-e slots on the motherboard - no change


I paid for the card on 20th of May, received on 22nd, shipped back on 24th, and just received it back on 31st. Do you think I can still DSR it? Kinda sucks though, since I've already paid £13 shipping back to OCUK, then they charged me £10 inspection + shipping back to me.. and now if I DSR it, I'm gonna have to pay another shipping to them lol

Could there be some kinda z77 bios problem? Maybe ivybridge gpu conflict or something?
 
Man of Honour
Joined
8 Nov 2007
Posts
16,034
Location
Outer Space
If you received it on the 22nd then yes you can still DSR it, but as you say you'll have to suck it up on the postal charges, though if the fault lies with your PC then another card could be exactly the same.

Can you try the card in a mates or family members PC?
 
Soldato
Joined
25 Jun 2011
Posts
16,799
Location
Aberdeen
You could try force the p2 state with nvinspector and apply clocks?

Sounds like a power saving feature is screwing up the boost states on the card.
 
Back
Top Bottom