Is GPU heat the cause of my crashing problems

[kinkladze10]

I'm Running a:

PK5
4 GIG Ram
8800 GTX
Quad Core 2.5 clocked to 3.0
26" Monitor
Windows XP SP3

I'm basically getting thrown out of Games, e.g. WoW with the same error "The memory could not be read"

All my drivers are upto date the pc is as good as gold when not running any games but I'm getting these crashes to the point where it is unplayable almost every 10-15 mins or so.

I've tried various display drivers and get the same results, I've done a virus/spyware search etc but I'm still having the problems.

I can only think it's either a ram issue or it's the heat.

A while back my GPU was running at 80-85 under full load (idle at 60-65) this was on a 22" monitor and I was getting regular crashes so I got a massive heatsink put on it with 2 fans from Vadim, since I've had it back I upgraded to a 26" monitor and I'm idling now at 50-57 and under load I'm at 70-75. It appears to crash whenever I hit the 75 temp mark or go around there, not straight away but that's when it tends to happen and the crash is totally random.

The improved cooling I've got has reduced the temps somewhat but it would appear not enough, I've taken the side of my case off and am considering just getting a desk fan pointed at the GPU until I can sort this but this isn't the answer long term.

I've done a few searches on this issue and there doesn't really seem to be a conclusive fix, I've even tried to set the priority to high in task manager and give it a single core affinity but no change.

I'm using Nvidia's Monitor View to record the temps and I've got all my fans set to 100% for max cooling but I really don't know what to do.

Other posts say 70-75 under load is way too hot but I have no idea how to get it down, the problem isn't helped that room has another pc in it and as I work from home it will have a laptop on as well sometimes so it's very warm in the room and not helped by the summer heat.

I was considering looking at getting one of those aircon unit but as it's a small room at best and to have a big hulking thing like that and then the big exhaust tube sticking outside the door is far from ideal, and again I don't really want to get this if it's not really a heat issue plus the wife would go ape.

I'm guessing I should run a memtest overnight and see if I get any errors to rule out the RAM? or would it be just easier to just take out one of the Ram sticks each time and try it till it crashes, then replace that stick and try another?

Any help/idea's would me most helpful on this.

*edit I can post my crash log and a Hijack this log if it helps?

Thanks

[Jay]

If you are running a 32 bit OS then you are not getting the most from your 4 gig. Your error could be to do with your overclock. Is your FSB and memory 1:1? What voltage and timings have you used?

[kalniel]

I would start by dropping to 2gb and going back to stock speeds. If you still have a problem then post back.

[kinkladze10]

Well I didn't set up on the timing and oc speeds, Vadim did.

With their reputation I do trust them to have done it correctly however that's not to say it could be causing complications down the line.

I'll be honest I have no idea about overclocking I wouldn't even know how to do it or undo it. I have no idea either about FSB or voltage/timings.

I'm only getting 3.25 GIG mem in XP, so should I take 1 stick of Ram out at time and see if I have a bad stick or is that a bad idea?

[kalniel]

Ah sorry, I didn't realise the whole thing was a Vadim computer - best thing would be go directly to them for support in that case, so that you can maintain warranty etc.

[kinkladze10]

Well that's the thing, it's been back to them, supposedly they put more cooling on it but since I've had it back the problem has not improved and this after a fresh install of windows !!

[kinkladze10]

Update, ran memtest, got lots of errors, not a good thing.

So at least I know the cause of the problem is the Ram, only thing is I'm not fully sure what it's telling me and how I correct it.

Is it due to just bad ram or are the timings wrong etc.

Here's some screenshots of the test, does it make sense to anyone?



And

[Pob255]

Well as you got a pre overclocked system from Vadim take it back and complain.
Ether the memory is bad or it's been clocked past it's ability to cope, the reason Why you get problems on games is that they are memory intensive (especially MMOs)
You could probably improve it your self by changeing the setting in the BIOS however you stated you don't know how to and more importantly you paid for an overclocked system so it's on Vadim to fix it.

[shaithis]

Just set the system clocks back to default and test again......if it passes, then you know its the overclock.

If it still fails, start removing 1 stick of RAM at a time and then re-run the tests.

Troubleshooting is a time-intensive process, there are normally no short-cuts and it normally involves discounting 1 thing at a time as the culprit.

[kinkladze10]

Well Vadim have just gone bust as of today so there goes all my comeback, not sure if I should laugh or cry !!!

[Pob255]

Ouch

well looks like your going to have to bite the bullet then and start digging around in the BIOS.

My 1st advice is start to read up on BIOS's it's often not as complex as it looks at first and most of the options can be left on there default settings.
Also have a look at the C2D overclocking thread the same stuff basically applies to the quad core as well.
http://forums.hexus.net/hexus-hardwa...beginners.html

[kinkladze10]

Cheers m8, I basically need a way to reset it to it default setting then memtest to see if the RAM is bad at stock speeds or if its just an overclock issue.

Last night I ran memtest and got RAM errors after 10 mins, I've taken 1 stick out so far ran for 20 mins and no error.

Dunno if I've just got lucky or not?

One thing I did notice was the RAM was redhot when I touched it, wondering if it is overheating that's causing the RAM to fail perhaps?

[kinkladze10]

Well major disaster, came to the pc this morning and it won't post.

Totally dead, have checked all connections tried reseting the CMOS but I have no idea what has happened.

Im totally screwed as Vadim went bust yesterday and I have a £2000 rig here which is now useless.

No idea what to do or who to turn to, I don't have the knowledge myself to rebuild it either.

I'm guessing I need to go to the yellow pages and reach for a pc expert ??!!

[kalniel]

If you need it quickly then yup. Otherwise we can take you through the various stages to test etc. but obivously it'll be slower. Also Vadim may still have something in place to deal with existing warranty/service requests.

Describe exactly what happens when you push the power button - any beeps? Any fans whirring at all?

[kinkladze10]

Power on the pc, fans go round lights come on inside the case etc.

But that's it no beeps no nothing.

No picture on screen either.

Have checked all connections have re-seated GPU and reset CMOS jumper and Battery too

No difference, it just won't post at all.

I have emailed Vadim but Im really not confident.

[chqshaitan]

That does not sound good, for it not even to provide a post bleep As that typically means something is loose onboard or has failed.

Have you tried removing all the ram and seeing if it bleeps when you try to turn it back on?