Is my PSU shagged?

I’ve been running a stability test on my graphics card after doing a mild overclock and have reverted back to normal settings. Still seeing the issue though, basically during a stability test with ungine the screen goes blank, then the picture comes back and the min frame rate is registered at 7 fps roughly, instead of 21.

At the time of the event, temperatures have been between 70 and 75 degrees on an AMD R9 280X GPU. The 12V voltage stays normal but the VDDC voltage drops from 1.148 to 0.945, whilst the VDDC current drops from 110A to 7-8A – consequently the GPU clock is throttled down to around 300mhz and the memory clock is throttled down to 160mhz.

Dodgy psu or undervolted?

How many watts is your psu?

110A? That sounds massive!
Have you got that right? That’s an Arc welder.

Corsair CS750M – 750 watts. Samuri, those are the numbers that GPU-Z is kicking out in my excel logs.

75C? That’s quite hot. Thermal limiting?

Gets too hot so turns down what it’s rendering, which results in less power.

When was the last time you attacked it with a vacuum cleaner?

Does it need a bigger heatsink?

Do you have any freezer spray?

I find over volts/clock the safety settings come on after a reset or the card fries.

Could be the power supply being dodgy although I’ve seen one puff out smoke and die.

Hopefully someone here will have better advice.

Thermal throttling, if the PSU was overtaxed it would cut out and the PC would just cut out and reboot would be my first choice.

However input your spec here and see what you need >>

http://www.extreme.outervision.com/psucalculatorlite.jsp

and be honest on the capacitor ageing drop down near the bottom

Throttling the PSU?

75 degrees is about 3 degrees hotter than it gets when stock. Reference cards with standard coolers hit 79, this one came with a massive heatsink preinstalled.

It’s not dusty, however the PSU fan fan is now making a ratling noise.

I’ve a R9 280X, what GPU and Memory speeds are you running it at. I’ll take a look at mine tomorrow and see what temps vs clock speeds its at.

Asus R9 280X with their bigger cooler. 1070mhz stock, ran it at 1130mhz before I had an issue and if I stress the card at stock for a long time I get it as well. No overvolting.

Been monitoring the system on idle and their appears to be some strange voltage drop offs as well.

Mines 1100 standard with 1150 boost, I’ll check the voltages and temps and let you know.

When I first installed it I was getting the random reboots and whilst my PSU max wattage was well above what was required for the card, I changed it out and no more reboots. All I could think of was that it was over 2 years old and had lots some of its max grunt.

One thing are you setting the power slider in overdrive to 20% when you do the overlock?

It sounds like thermal throttling but R9 cards run hot and a 280x shouldn’t throttle till it hits 95c. 750 watts should easily be more than enough to run a single 280x regardless of the other components and a Corsair PSU’s are high end (but could be on it’s way out I guess) so shouldn’t cause problems. So I guess I’m stumped unless the card manufacturer has changed the thermal limit? 110A can’t be correct though, sure it’s not watts?

Edit:Do you have another spare power supply to test with?

110 amps is the vddc current. That is the current supplied to the GPU itself. And it is being supplied at 1.148 volts, so the reported power consumption of the GPU core will be P = VI, or 126 watts. Not unreasonable for that GPU.

PSU should be fine for virtually any single card system, although you don’t specify what the rest of the components the system is running.

It sounds like the card is throttling to self protect although it’s unlikely to be the core temperature that is the problem. Could be the vrms or something else.

Check the windows event logs for graphics related errors. Try a different version driver.

If you think the problem is heat related then take the side off your case and blow air in with a large desk fan. If the time before throttling increases, then you have a heat related issue. If not I’d guess overcurrent protection but may very well be wrong.

Pretty sure a thread like this came up on the anandtech forum a while back. I browsed it but don’t have the card so didn’t really pay great attention.

Just saw that the 4amps being supplied to the my cards is just the VRMs so maybe 110amps isn’t impossible

PSU should be fine for virtually any single card system, although you don’t specify what the rest of the components the system is running.

I7 4670 running at 4.2ghz, stable under prime95 tests. 8gb of DDR3 running at 2000mhz, stable under memtest. One 500gb hardisk, 3 120mm fans, 1 140mm fan and 1 cd drive. Adds up to about 500 watts under load.

Cheers guys, running tests at stock again and at overclocked. I had data from my overclock that shows the first event when clocked at 1170mhz happening at 78 degrees celcius, some people seem to think that the cards throttle at 80c whilst others think 90c. So the scientist in me is going to try and repeat the issue and see if it’s related to heat, if it isn’t then the first thing I’m going to do is warranty the psu. I’ve been getting some strange GPU voltage drops to 0 when looking at the overclocking software from Asus when idle. However GPU-z doesn’t show them.

i7-4770K stock speed with 16GB of DDR3 @1866 1*1Tb HD 1*256GB SSD 1*BluRay drive 4*120mm case fans
R9-280X at 1100 GPU with 1150 boost with the memory at 6.4Ghz and a 760W PSU.

The GPU is at 28-29c Idle, running a 3D Mark session max fan was at 2777 with a max temp of 70c drawing a max of 12.9A on one PCI-E rail and 11.3A on the other which works out if I add in the other motherboard 12PCI-E connector at 1.8A at a draw of 312W

It’s the memory overclock that did it, after I rebooted it hasn’t happened at stock clocks after 1 hour of testing. Will try a 4 hour test, see if that does it and if not then I’ll call it a day.

Basically, when you start overclocking memory, even if the temperatures and voltages are fine, errors can be introduced by the memory. Average frame rates actually take a hit when the memory clock is taken beyond a certain point and instability kicks in.

Thanks a bunch of the help chaps.

What memory speed were you trying and whats your stock?