Computer randomly turning off after GTX 970 Installation

Hi guys,

I got an Asus GTX 970 today to replace my aging GTX 560 Ti. It's a powerful card that TBH is at odds with the rest of my system (which I'm replacing bit-by-bit). After I installed the card I found it worked great, running pretty demanding games such as Shadow of Mordor maxed out.

But then came the random turning-offs. It could happen whilst in BIOS or in Windows (even during Safe Mode sometimes). Typically what happens is the screen will freeze, and then a second or two later the entire system power is gone.

I'd be tempted to think this was a PSU issue if not for the fact that I was indeed playing games fine and not only that, but there seems to be absolutely no correlation between how demanding the process on screen (more demanding would seemingly = more power draw = greater chance of shutdown).

Having said that switching back to onboard graphics or the old GTX 560 Ti seems to fix things. I've done the whole rigamorole of undoing drivers, registry editing and Ccleaning and all that jazz to get a fresh install (even using Nvdia's fresh install option when I did install the newest drivers from their site). This initially fixed the issue... for a while. And then the issue returned.

My Specs are:

Case: Be Quiet! Silent 800

PSU: Be Quiet! CM BQT E9-CM-580W PSU Straight Power

Motherboard: Asrock kn10n78

Boot Drive: Samsung Evo 840 240GB

Extra Storage Drive: 5400rpm 1TB

Processor: AMD Phenom II X4 965

Cooler: Cooler Master Hyper 212 Evo

RAM: 8GB DDR2

My CPU temperature is sitting in the 40s, going up to the 60s during load, so I don't think that's the issue.

At the moment (owing to not having enough fan headers) one of the two front 140mm fans and the rear 120mm aren't running. Air flow isn't optimal so this could actually be the issue. If it is I'll find out tomorrow when I get some molex/3-pin fan connectors in, which should help airflow in the case in general. But I doubt this is related many shutdowns happen with the GPU as low as 31 degrees. And I don't think I've ever managed to push it past 60.

I posted this question on another forum and one user recommended turning up the voltage and underclocking. So, using Asus GPU Tweak, I turned the voltage up one increment from 1175 > 1185 and turned the two clocks down one increment each. Again, this was fine for a bit, even when playing demanding games. Works great I figured. So I put the panels back on the case, neaten up my desk and restart. Guess what? Problem comes back because the tweaks to voltage done by the Asus' software only kick in once the software loads when you enter Windows... which isn't much use if the system shuts down before you even get the chance to get into windows.

For whatever reason doing a whole driver scrub seems to get me stability, at least for a while, so I'm open to trying things. Any ideas?

max temp for the phenom 965 is 63c you don't want to see the 60's with that cpu. I know because it was my first, and I still own it.

could be that your computer is shutting down to prevent damage to the cpu?

edit: after reading more, I dunno. Never used Asus, maybe someone else could help with that.

Have you tried MSI Afterburner? since you complain about the Asus software.

Is there anyway you could try a XP/Vista/W7/W8/W8.1 bootdisk so you could rule out that something weird is going on with the OS.

You cant try a VM or anything, just so it boots from a USB or DVD.

You likely just need to raise your NB voltage (not CPU-NB) a bit so it can handle the upgrade...

@kitty: Yeah I'd imagine the CPU is fine because I ran it for years with the stock heatsink and fan *shudders* and it was 70... idling on BIOS. Funny seeing as according to Core Temp 70 is its TJMax (that's the level where it starts throttling it's own performance if I understand any of this stuff). Yet in all those years I've not had a single heat related shutdown before, at least not to my knowledge.

I've returned the card for now because I'm not happy with fiddling around with this stuff as well as the case I had with it. My old case doesn't have the space necessary to house the GTX 970 so it makes no sense to hold onto it. Figured I'd try upgrading again a few months down the line and do the motherboard, RAM and graphics card all at once.

Like he said anything over 60C on phenom II cpus is uncomfortably high since the limit is 62-63C so that might be a problem. Also you cant rule out Psu even if it shuts off randomly and in non intensive tasks ( heck with my psu before giving up i could play hours upon hours without a problem yet it would crash doing something simple like browsing or watching movies )

Its one of two things \

1.unstable cpu from high temps

2. or its the PSU unable to supply the power, GTX 560ti uses less power less a GTX 970 at peak loads.

Pretty sure GTX 560 TI has higher maxTDP than GTX 970

Nope, Peak power usage of normal GTX 560ti is 147w while ASUS GTX 970 uses around 170w. Now a GTX 560ti overclocked model or even the 448 core version can see 180w+ usage.

I can safely rule out the first one based on experience of running the CPU hotter for longer and having never had an issue.

I'm also sceptical because this issue has little to do with graphics card load. It can occur repeatedly on BIOS/before Windows loads. But conversely I can play something like SoM maxed out for over an hour and it will be fine (at least for the little while after a drivers scrub).

In any case I looked over the specs on Nvdia's page (granted these are for reference cards) and found that in terms of power draw the two cards are pretty much tied:

560ti | GTX 970

Max Temp: 99C | 98C

Max Graphics Card Power: 144W |145W

Minimum System Power Requirement: 500W | 500W

If anything I'd imagine the newer Maxwell card would be the more energy lenient of the two. Having said that Adamosmaki's comments might be onto something. I'll know when I get another card down the line and have the same issue. If I do I'll try jumping up to a 680W PSU.

Edit: Figured I might as well jump up to a new PSU now. Will return the old one as it's been less than a week and pay a bit more on top for a bit more.

^^ It's funny you should mention that. When I checked the Amazon page the other day they'd halted selling it because they'd gotten a complaint about it not being as described. I figured it was just one disgruntled customer who'd actually gotten the wrong item, but thinking about the wider context I'd imagine the returns have come in at a rate/number that alarmed Amazon.

I can say with near certainty that your SPP (NB) voltage just needed to be raised. Motherboard chipsets, especially nVidia, need more power to transport that much data. Older AMD and Intel motherboard northbridge chipsets need higher voltages than "auto" will provide in many instances as well, especially when combining powerful cards with high CPU overclocks...

@superclocked: Yeah, I saw your message a little too late as I'd already returned the case by then because of all the trouble it was giving me (cable management in name only). Without the case there was much point holding on to a graphics card I couldn't use. Hopefully by the time I get a card like that again I'll have a newer motherboard in any case. But I'll keep this page bookmarked for reference.