The Official Graphics Card and PC gaming Thread

Estuansis,

Your running the same Cpu clock, that im trying to stabilize. For the life of me, I cannot do it LOL! Other than your 206FSB, and 1.45Vcore, did you do anything else? NB, Ram settings, etc. Though, you do have a DFI with a different northbridge, and a different PSU...*grits his teeth*

 

OK, this is really starting to F me off. Yes, new drivers fixed the problem in fallout3. And crysis is very smooth. But every now and then, when playing crysis, power usage around 600w.... it just locks. I emailed evga tech support and they suggested running without the sli bridge connector - which I am currently doing, and it seems to have lasted a lot longer so far (hasn't crashed). But, with the bridge connector off, the power usage is about 100w less, so it doesn't bring me any closer to figuring out if its the bridge connector or the PSU to blame. I have a spare ttake 750w sitting here - is it possible/safe for me to short the black/green pin and run only the graphics cards from this PSU, or will the cards not know how much power they need seeing the PSU won't be plugged into a motherboard?

 

Where a GPU gets its power is of no consequence(provided it can put out). If you hotwire the OTHER PSU via THE green cable to any black cable(on the 20 - 24Pin), it will engage the secondary PSU. A PSU need not be connected to a MOBO to supply power

Here you go buddy
http://www.techwarelabs.com/guides/misc_mod/psumod/

I have performed this on numerous occasions, IT WORKS!!! Very well indeed.

 

Same result. Still get random locks. Although, instead of completely freezing the computer now, sometimes it just crashes to desktop. Anyone found a way to raise my NB voltage yet?

 

Is that a reliable result? Does it lock completely every time with one PSU, and only ever crash to desktop every time with the second? If it does, I still say try another PSU for your whole system.

 

Nope, it's random now. Sometimes it crashes to the desktop, sometimes it hard locks. I'm currently installing 32bit to see if that works. Evga seems to think its a PSU issue too, but they also said they'd replace my cards for me anyway, even if I don't think they are the issue. But I don't know. How would I be able to tell if the second PCIE slot is buggered? Or would the second card just not show in windows in the first place?

 

Graphics manufacturers ALWAYS say it's a PSU issue. ATI told me my 850W Zalman wasn't enough to run Quad CF when I had stability issues, and recommended I buy a 1KW PSU.
PCI express faults are hard to diagnose, both my Gigabyte boards refused to accept dual graphics without crashing, but there were no BSODS (well there were, but they were for a different reason), the bus just shut down which caused the cards to completely shut down too (Load power 700W, idle power 300W. Power usage when this happened - 220W).

 

Try turning on ACC in the BIOS. Phenom II chips are supposed to have it built in, but I found turning it on helped stabilize higher clocks. Also keep in mind that no chip is guaranteed to OC. I've seen E6600s that couldn't break 2.8GHz stable. But I got mine to 3.2GHz easily.

I forget what Southbridge does your board use? Mine is an SB750. There are a few 790 boards with the older SB700. 750 is supposed to OC better.

 

SB750. I wonder if yours being the 790FX and mine being the 790GX has anything to do with it.

I'll look for the ACC setting now

 

I would expect so. OBG has been known to cause overclocking problems in the past.

 

But...its disabled. How would that effect it?

 

It depends to what extent it gets disabled really.

 

Turning on acc, and an Auto setting...BSOD. And with that setting, I cannot see my individual core temperatures. Well...I gotta get busy with other things now. I'll look into it more later, thanks guys.

 

Well, thats sort of what's happening to me, but the card's arent't completely shutting down otherwise I'd be more in the range of 300w instead of 360w. Everything loads up to around 600w... play crysis for about... 30 seconds and BAM. Immediately back down to 360w and either CTD or complete lockup.

 

If it's the same figure as you get when just in windows, the cards are going idle (no input) not being shut down, but who knows, nvidia cards may not act the same way as ATI ones do.

 

OK, I'm currently running the CPU and 1st position graphics card off of my extra PSU. With only the motherboard, other card, and HDDs plugged in, crysis is drawing 400w at load (Sli enabled) Not a SINGLE freeze as yet. If you take 400w+200w for the other card, plus extra for the cpu thats a lot more than the 600w my computer usually crashes at. Think its the coolermaster not dishing out enough power for everything?

Edit: Plugged the CPU back into the coolermaster, after about 5 mins of play - at 440w - frozen. Surely 40w can't make that much of a difference- maybe I just didn't test it out long enough before.

Double edit: I've narrowed it down to either the PSU itself, or the set of cables I'm using. I'll borrow some cables out of my old psu and see how I fair.

 

Test more thoroughly with the dual PSUs, like for days.

 

Well, cross fingers, but looks like it might just be a faulty cable. Tested with my mates corsair cables, managed to play for a lot longer, and no crashes at all. Does it make sense that all these symptoms could point to a single faulty cable? Like... one cable would feed the card for a little while, then when it needs more power, the psu just conks out because it can't deliver all that through just one cable?
And, to further support my theory, Fallout3 now works. And I figured out why it would have been crashing at that same exact spot if the cable was faulty - because after the screen fades back from white, there is a 70w power spike - which I don't think the graphics card would have liked too much if one of it's cables wasn't working before

 

I suppose if a cable were faulty enough. It is possible. My fingers are crossed for you buddy As demanding as GPU's are, and how close most PSU's are to putting out the Bare requirement, if a cable were faulty, its pretty possible. I would not doubt that. Depends on the imperfection.

 

Hehe. Well, I've been playing fallout for a while without a hitch. And it used to fail every single time. In exactly the same spot. But it's doing fine! lol