I have to say 11.65v is ok and within spec and your PSU has a single +12v rail rated at 960W 80A so it really shouldn't see dips down to 11.13V. On the other hand I can't really say that it is or isn't a bad PSU anything in your system could be dragging down that voltage not just the GPU. One thing you can try is re-running all the cables and disconnecting anything un-necessary in your system like dvd drives and other pci/pcie/USB cards/devices "I know USB is not 12v but it's still a load on the PSU" and try testing again then add your other devices one at a time and testing every time. God that sounds tedious. Edit: Granted I'm still not sure it's a voltage problem, at least in my case. I haven't seen any really low voltage drops on my system and my black screen is not instant and is avoidable by running the fan curve starting at something like 40% and steeply going up to 100% rather early at something like 75-80C. Not those are really ruff numbers I haven't even attempted to fine tune them at all.
Yeah, same. Let's see if the future "fix" drivers change that, if they don't come out tomorrow, I am returning the card as I already talked to the store and I will exchange it for a GTX780 Ti.
There is no "in spec" until the droop has been measured while under load. At 11.6v idle it's unlikely droop will stay above 11.4v which is the lowest allowed.
Now you're just nitpicking. But yeah I assumed under load. Edit: Spec is normally. Nominal value: 12.00V Allowed voltage variation: +/-5%, 11.40V - 12.60V Preferred voltage variation: +/-3%, 11.64V - 12.36V Allowed ripple: <120mV Preferred ripple: <80mV And yes that's a shameless copy paste and god I need sleep. Edit9000ish: I forgot to add you can read the proper voltages with a multimeter and an unused Molex connecter. This will be far more accurate than software reading cheap & uncalibrated sensors off your MOBO and GPU
Yes software readings can't relied on for 100% accuracy, they only as good as sensors on the motherboard which could be little wonky DMM will give accurate results
While measured voltage may be off by few 100s mV, 500mV drop is not inaccurate reading. My 750W PSU has 200mV drop from 12.05V to 11.85V under maximum CPU+GPU burn test. 500mV is way too much. And in such case I would use voltage measurement points on MB which is safest way to measure +12V since you can short circuit molex if you are not careful.
I have to agree with Pill Monster, here. I mean no offense, whatsoever, but you're being really quite obtuse and stubborn. We still know jack-squat about the black screen issues. However, as pointed out by myself and others, the low 12v readings are suspect. You seem intent on ignoring this. You also have a flashy motherboard with 4 (4!?!?) Aux power connectors that you more or less refuse to utilize. Based on principle, I would never, in a million years, intentionally not provide power to those hookups, regardless of how "optional" you feel that they are. Furthermore, you claim that since you had success with a 7970 that your PSU must be in perfect working order... even when you admit to seeing readings far below the allowed variance in the 12v specification. This is a type of fallacy referred to as Post Hoc. For your immediate situation, you need to actually do the testing rather than just saying you don't want to, or shouldn't have to. It's almost as if ego is preventing you from following proper scientific method. Plug in your motherboard and remove the iron gate that's barring you from making progress with this endevour.
Ok, then I apologize. From what I had read, you mentioned that you could not hook up some of the connectors for various reasons. I'd like to see this resolved as much as anyone and I don't even have a card. :nerd:
Yeah, at first I thought they were MOLEX connectors, but I remembered that was on my old P67 motherboard, then realized they were standard connectors like the PCIE 6 pin, I tried it, but it didn't help at all. The reason I say I don't trust software voltage readings is because I had 2x 7970's, even 3x 580's before in this same PSU and it never failed. One PSU doesn't die from one day to another with a GPU change, I think if it was faulty it would have failed on my 2x 7970's as well... but it didnt, EVER. If I were the only person affected by these black screens I'd definitely try to blame it on the PSU, but with so many other users with the same issue, I can't. I already asked the store to allow me to return the card for a refund (So I can get a 780 Ti), I am holding one or two days until those promised AMD drivers release, and if they still don't fix the black screen issues, I'll try another GPU.
Can't disagree with that. What I'm considering, which might be pretty obvious to most, is that there is some variance in the cards themselves that causes this excessive 12v drain and it is that drain that leads to the black screens. Through manipulation and testing of the 12v (via mm readings, software readings, psu swapping) perhaps more could be learned. I've yet hear if anyone had a "normal" 12v prior to their 290x's and it's only now that they're seeing these relatively large droops.
If I had the money, I'd just buy a new PSU from Amazon and test it, and then return it, just to see if my PSU is also part of the culprit, but I can't do that...
May I ask you: "If software show you 13.5V which puts your entire system in danger of burning, would you take multimeter and measured real voltage?" And while 3x GTX580 which is only 3x 244W if not overclocked can draw at worst 732W + 250W remaining components is still below 1000W. It for sure was not easy life for your PSU. And as 780Ti draws about same as 290x you may be very close to end of life cycle of your PSU. And I personally would not risk entire PC by not confirming health status of PSU.
Do you have any of your old cards left over still? Maybe it'd be worth noting what the 12v readings are at idle/load with a previous generation. Maybe you'd still be sitting under 12v and thats perfectly normaly for your rig. Maybe the rail will be 12v+.
You have a good point, but I don't have a volt meter neither know anyone near me who has one... And unless I can PROVE it's the PSU without that tool, there's not much I can do. No, I sold the 7970's right away. I'll ask a friend see if he can let me borrow his 5850, but that card doesn't draw much power so not sure if it'll be helpful?
I will, if the store allows me. And IF then I still get failures with my computer, I'll RMA' the PSU, but right now I don't want to RMA it because if it's not faulty the store will charge me.
I am sure they will charge you more than cost of cheapo multimeter. And even those can measure in 20V measurement range with error about 0.01V. You do not need expensive stuff capable to measure transistors and capacitors.
You know I'm tempted to go back to NVIDIA as well I have had 5 ATI/AMD GPUs over the years and I have never liked their drivers. And yet I have had no major problems with NVIDIA's drivers. The early GTX480 drivers were a little dodgy but still overall it was ok. It also looks like they are making better use of the computing power available on their GPUs by slowly adding more features to even cards as old as my GTX 480. Well except ShadowPlay and G-Sync the GTX 480 isn't getting that but still. I expected some small problems with the early reference cards and drivers but not this. The only reason I went with the 290x over NVIDIA offerings was of all things the one thing I don't really expect to be adopted by many developers "TrueAudio" yeah I'm a sucker for sound and really a sucker for good positional audio. This is a problem most developers could have fixed if they just used something as simple as OpenAL. Granted I did have my first GTX 480 go bad after 3 weeks and had to RMA it. In the mean time I guess I can take my system to work and see if I can take my power supply out of the loop. I think an Agilent 6032A bench power supply will power the 290x and maybe I will poke around on the board and see if I can find the problem. But I dont think I can get to that today.
i have 290 and no black screens but i played around a bit in last days and if you with black s. issues could plz test something at your own risk. i played around with aux/vddci and found that even the smallest undervolt(ex:minus-6mv) would cause this exact black screen. so maybe you could monitor or test for yourself if up (or downclock) makes a difference for you. i would not take BF4 as test because i played it on nvidia and amd and had probs before patch(but 2 hours or more playable) and after patch it got worse for me while others have no prob from beginning.