Random rebooting

[SmoulderingCircuits]

So i figured it'd be wise if i tested my second 5770 which arrived yesterday before i actually crossfire it incase is faulty. Now, it works fine on the desktop, fine watching youtube, but then when i played crysis demo rebooted soon as i put up to high settings (bear in mind my other one can handle maxed out). When it booted back up i checked temps and they were 50*c or below which is fine for this card at stock speeds (this is why i test before overclocking), so i thought maybe it's just a one off thing and fired up battleforge. No funny noises or nothing, fine, then died after five minutes again. This time it had shut down and wasn't booting, so i checked everything was connected and sure enough it was. The LEDs on the mobo were still lit so was definately recieving power. The psu fan carried on going to cool itself off afterwards so the fan hasn't conked out or anything like that. Still not rebooting. So then it occured to me, what if i flick the psu off, pull out the lead, then put back in and flick on again? Bingo, booted again and i have just ran cinebench open GL and cpu render tests without any abnormalities. So i am now wondering, should i attempt running games again, or is it likely to kill something? Remember all temps and voltages are fine. Case is tidy and hardly any dust.

 

[bigdog]

i would say check your driver on both cards if thay have the latest drivers installed let me know how it goes

 

[SmoulderingCircuits]

i would say check your driver on both cards if thay have the latest drivers installed let me know how it goes

I'm only running the new one atm, then i know it's definately the card and not crossfire causing any issues. Has run battleforge okay but my first 5770 never went much past 55*c and i am getting 60 - 65*c on the same settings with this new one. But seems to be going okay so far however i highly doubt i'll be able to overclock this one

If it can run crysis demo for an hour or two then i will deem it good to use

 

[SmoulderingCircuits]

Has ran battleforge for half an hour with fair amount of load, not exceeded 63*c, now on to crysis

 

[SmoulderingCircuits]

Crysis was stable the whole way through so now have both of them in the rig will test out later

 

[SmoulderingCircuits]

Ran AvP in crossfire and shut down again after about 15 minutes, psu dust filter is clear as a whistle and nothing felt hot whatsoever after opening case up

 

[zak4994]

So the newer, second card is essentially causing a problem?

I wouldn't risk it. Random shut downs and reboots are never good for the hard drive cos of the constant data access.

 

[SmoulderingCircuits]

So the newer, second card is essentially causing a problem?

I wouldn't risk it. Random shut downs and reboots are never good for the hard drive cos of the constant data access.

But strangely it seems to run for longer and longer periods before crashing the more i play the game, usually 5 mins, crash, 15 mins, crash, 45 mins, crash, two hours, fine. I don't think it's the second card though. As every other game has ran fine since crossfiring. If i keep switching out of games via the windows key never crashes. If i run for a while and don't exit the game then crashes after some time. But nothing gets hot, i don't understand it. I'll give it a week to see if sorts itself out.

Thinking of replacing the hard drive as is bit slow anyway and is the only part remaining from a previous rig.

 

[SmoulderingCircuits]

I turned off AA as it seemed to cause microstuttering, and cards have gone from 67*c top/ 51*c bottom to about 51*c each. Hoping that will solve the issue as I don't actually know what the thershold of a 5770 is but IF it is heat causing random shut downs then it would seem 70*c is the breaking point, however that doesn't seem logical considering the top card is hitting 67 in a room at 18*c!

But i'm becoming more concerned of the hard drive as seems quite apparent in lagging for five seconds (audio, frozen screen) then moving normal frames a second then repeats about four times whenever i reach a checkpoint in AvP.

This same thing happened when i tryed running Tiberium Wars maxed out with 8xAA on a Geforce 210

except the psu never needed plugging back in, only wait for the card to cool down

 

[zak4994]

What kind of hard drive is it and how old.

Might be able to give you a rough estimate on whether it's time to for the hard drive to die.

 

[SmoulderingCircuits]

It's the only remaining part from my old rig my bro built me about three or four years ago, back when i knew nothing of computers lol; Samsung Sata II 7200rpm 450gb and is either 16/ 32mb cache. I hear Samsung isn't that good for reliability and as it's approaching four years use anyway it doesn't take a genius to suggest it's five feet under.

Planning on ordering a pair of 640gb Sata III Caviar Blacks with 64mb cache each to stripe in Raid 0, then fresh install of windows. Hopefuly can copy my games across without having to spend a week installing them again

 

[SmoulderingCircuits]

Right I am now pretty sure the threshold of a 5770 is 70*c as when at 68*c if I go back into the game then save it shuts down (so more than likely hit 70). Now it isn't any specific one of the two cards that gets hot, they seem to switch between worker and helper roles every 15 mins or so. I have tryed running AvP on minimum settings bar 1600x900 resoloution and hasn't made the slightest scrap of difference to temps.

So I'm now met by a catch 22:

If I increase the fan speeds from the highest they are reaching now [50%] to 100% it may solve the heat issue but will also void warrenty, and should it fail to solve the heat issue i'm now stuck with two cards that only stay within thermal bounds in single mode.

But I don't like the idea of going back to slide show framerates with only one card so I'm thinking maybe unlock both cards, bump fans up to 100% (sound proof headphones + part deaf anyway) and lower cores to 750Mhz from 850Mhz and then if that solves the heat issues, slowly raise the bar back to stock core clock until gets hot again, then down a notch and leave it there stable.

What do you think?

 

[zak4994]

Might be your only choices.

This seems to be a very unique problem.

Sorry mate.

 

[SmoulderingCircuits]

Yeah it since crashed today straight after the loading screen in battleforge. I think it's two combined problems of a hard drive on it's way out, hence the lagg after checkpoints and crash at loading screens, and also the gpus do cut off at 70*c but my bro should be able to take a look tommorow (senior engineer for his company).

Thank you very much Xeno for your replies

 

[SmoulderingCircuits]

Oh and one last thing: if I do end up unlocking the cards, seen as I will have voided them anyway, do you think I would see much of a temp drop if I replace the stock thermal compound with some Arctic Silver 5?

Would that mean faffing about with those pressure pad thingamajigs in graphics cards?

 

[SmoulderingCircuits]

Has exceeded 71*c and stayed working so isn't overheating, however this was with AvP on minimum - though makes no difference to temps whatever the settings

 

[Rastalovich]

Have you got enough pcie<->molex power adapters to do some power testing ?

Even the beefiest of psus can have weak lines.

 

[SmoulderingCircuits]

Have you got enough pcie<->molex power adapters to do some power testing ?

Even the beefiest of psus can have weak lines.

There was one with each card and i just ran AvP on min for two hours straight with no probs whatsoever. I then go to start another session on min and dies within five seconds!

I think you're right, maybe it is the psu, coz temps aren't affecting anything, and while the hard drive is slow has happened at points where no loading/ saving is taking place. Psu has 2x6pin and 2x8pin. Has four individual +12v rails @18A.

But the only thing that doesn't correlate with the psu being the cause is I experienced this problem with the second card when using the exact same port and cable I fed the first card with which never had this problem, which leads back down the road of maybe it is the card

Man what a headache!

 

[SmoulderingCircuits]

I'm gonna use the 8pin slots instead of the 6pins, see if that provides enough juice should power be the issue

 

[SmoulderingCircuits]

Tryed it, didn't work, only stable for a couple hours at a time on minimum. I disabled crossfire in CCC and now runs AvP maxed with no instabilities at all. Frames aren't drasticly reduced either. Must be the card then.