P55A-UD3R unstable under graphic load

Hi.

My PC built on the Gigabyte P55A-UD3R motherboard, and it has issues when playing games. Namely, PC is rebooting under (un)certain conditions. Generally, when it is loaded with some serious graphic.

Details. Some games, very demanding ones (Witcher 3), had no problems at all, or, if they did, then very rarely. Other not very demanding games (Skyrim, Mass Effect 3's Leviathan DLC) had very constant and repetitive issues. Skyrim was rebooting the PC when in in-game menu. ME3's Leviathan was rebooting at certain cutscenes, but not even loading CPU/GPU that much.

In medium cases it was also dependent on environment temperature. Cold = less problems, warm = more problems. Howewer, it was not helping even if I had downclocked GPU and limited its power consumption very hard.

PSU is 750W and fine, I did tested it with Furmark GPU+CPU mode, it holds. GPU was 6970 initially. Now I changed it with FURY X and it got much worse. With its advanced power management I believe I can limit consumption so that it will consume much less than 6970 did. But it doesn't helps at all (like, I can play Witcher 3 for a minute if FURY X is downclocked, or for a 15 seconds if it isn't). With 6970 it all was mostly playable, and with FURY X - mostly not.

But as it is essentially repeating on different GPUs I think GPUs are not the problem themselves. Again, it could work with 6970 at 100% constant load and whining fan, but reboots in another game when it is downclocked to 500MHz, -20% power and graphics settings so low that GPU was just at 40% activity and at 60 Celsius (which is low for a 6970).

So, what is left is motherboard. But how to diagnose it? All system temperatures that are monitored is fine. The chipset isn't even hot. Power circuits? Any other chips that can affect?

Any versions, suggestions, please?

P.S. It seems that I will have to upgrade MB/CPU, but as I already have 16GB DDR3 memory and i5-750 at 4.0GHz I feel, you know, greedy to change it with essentially the same stuff. So I prefer to find a solution rather than upgrade.

Can you post your full system specs like in my sig?
Please include PSU age also.

i5-750 [email protected] with Grand Kama Cross
P55A-UD3R rev. 1.0
Kingston Value RAM 16GB 1333@1200 (due to memory/bus clock multiplier)
Sapphire 6970/FURY X GPUs (not at the same time, of course)
Thermaltake Toughpower 750W W0116 (8-9 years old now)

1) I already did checked it with default clocks so probably overclocking is not the case.

2) PSU holds when GPU+CPU (6970 and i5@4,0, didn't tested FURY X still) 100% loaded with Furmark, so why it will fail with even moderate load in a game? And as FURY X has 275W TDP and 6790 has 250W TDP, when moving power limit slider to the minimum it actually makes FURY X less power hungry that 6970 at defaults, right?

3) PC case has pretty good air flow, a fan at front bottom sucks air in, PSU at rear top throws it out, and huge side fan adds some more air and mixes it all. No overheating at any sensor (that can be monitored with tools) was registered. No voltage deviations as well.

Replace the PSU, it's simply old and tired.

Sure.

I'm writing this post running FurMark test on FURY X. I set CPU burner to load cores 1-3 (just for additional consumption) and set GPU burner to 4th core so that it can deliver constant full load to GPU. GPU is set to +50% power limit, 1050MHz 99% of time. So in terms of power it is now working at the level it never actually achieves in real games.

And it doesn't reboot.

PSU really?

FurMark isn't a realistic load and is known to kill GPUs, quit using it.

Test with Heaven and Prime95 at the same time.

Sounds typical of PSU issues though, and your unit is quite old, especially for the quality of that unit.

Realistic or not, it is still Volts and Amperes, if speaking of tired PSU. And judging on temperatures, there was a lot of Volts and Amperes.

Speaking of various benchmarks. Okay, maybe I'll find one that reboots my PC too, just like in game mode (BTW Heaven didn't, maybe version was too old, or maybe too short run, but it is already longer than any game has run until reboot with FURY X). How is it supposed to help me? I already know there is some stability issues. I need to localize and possibly solve it. None of those benchmarks will tell "okay your PC is rebooting when PCI-E bus is overloaded or when rasterization load is high or whatever".

Take a read of this summary:
http://www.jonnyguru.com/modules.php?name=NDReviews&file=print&reid=7

Lists the exact problem you're seeing, high load at warm temps causes shut off.

Hm, possibly.

Maybe some of 4 rails is not so durable as others. And maybe in some scenarios there are surges of power on that exact rail, even if overall power draw is considerably low. Despite that in other high load scenarios it works well... I'll look into it, thank you.

Yeah, you should replace that power supply.

Most are meant to last 3-5 years, after that they're on borrowed time generally. Though there are some with 7-10 year warranties. Eight to nine years is a really long time for a power supply.

Even if the PSU ends up not being the issue, it needs to be replaced before it fails and it's most definitely on borrowed time already.

ATMINSIDE said:

Even if the PSU ends up not being the issue, it needs to be replaced before it fails and it's most definitely on borrowed time already.

Click to expand...

I believe the PSU just needs a recap at the worst.

RJARRRPCGP said:

I believe the PSU just needs a recap at the worst.

Click to expand...

Dangerous, and potentially life threatening, for anyone that doesn't know what they're doing.

Well, let's try compressed air first.

Okay, here is the conclusion. My idea about motherboard appeared to be at least reasonable, as it had catch some weird issues too.

First, I did cleaned PSU from dust, as it wasn't cleaned never before. I thought there must be a mice kingdom, at least, but dust wasn't too heavy, surprisingly. And also I cleaned the PC. Thought to see if PSU (if it is the problem) will possibly overheats less.

Looks like dust was the last thing that did hold it all together. When I turned the PC on it started only to shut down just as OS starts to load. I went to BIOS and reset all settings to fail-safe defaults. Something went wrong, PC suspended at reboot (black screen, no POST or anything). And after manual reboot it now shows a MESS screen instead of POST. However, I still could enter into BIOS settings, and it also displays the "loading operating system" message normally and then tries to actually load it, but again it just shuts down (no rebooting afterward).

At that point I psyched and went to store and bought the goddamn new PSU and the goddamn new motherboard and the goddamn new CPU.

Got rid of hardware - got rid of issues. Old ones are going to GULAG. That's how you solve stability issues, kids.

It's very possible that you hit it too hard with your compressed air and knocked something loose on the motherboard, at least it's working now though.

FYI, this is a family-friendly forum. Easy on the language.

ATMINSIDE said:

It's very possible that you hit it too hard with your compressed air and knocked something loose on the motherboard, at least it's working now though.

FYI, this is a family-friendly forum. Easy on the language.

Click to expand...

No, all the air went to PSU. But I had to, basically, disassemble the PC almost completely, because of very tight main power connector and lack of support under that part of MB. Maybe some manipulations damaged it, but I swear I was very gentle and polite.

But this old MB was some collection of bugs, as I see now that I have no those bugs with new one.
It recently started to load OS for very long time (more than a minute of black screen and still Windows logo). Gone when 6970 was replaced with Fury X.
But the I couldn't update GPU drivers because system was freezing in the moment the old driver unloads and the new one boots up. Had to load OS in safe mode to do the trick.

With the new MB, OS is loading instantly now without reinstall, AMD drivers updating well.

Guess it was the PSU after all eh?

I did not checked the old MB with the new PSU because of obvious failure of old MB.
And did not checked the old PSU with the new MB yet. Tired of jumping around the PC.

With the MB glitching so hard and finally failing, I bet the PSU could actually be OK.