After all my testing is it my PSU?

Plec · 10 Apr 2009 at 20:10

Spec:

XP Pro 32bit (i know i have 4gig but i'm awaiting a 64 bit OS)
Intel Pentium Dual-Core E5200 "LGA775 Core 2" 2.50GHz (800FSB)
Akasa AK-965 (CPU temp doesn't go above 30 degrees)
Gigabyte GA-EP43-S3L Intel P43
Corsair XMS2 4GB DDR2 PC2-6400C5 TwinX (2x2GB)
XFX ATI Radeon HD 4850 'XXX edition' 512MB GDDR3 (i haven't seen it go above 63 degrees)
PSU Corsair Memory 650W TX SERIES ATX (still under guarantee)
3 SATA HDDs
1 DVD-R

My computer crashes using Furmark and in Crysis. It crashes in Furmark after about 20 seconds and the temps are fine. Crysis crashes after about 10 minutes or so - but again the temperature of both Gfx card and cpu are fine - very cool in fact. Both types of crash are very similar in look and the fact that i have to do a hard reset. And other games so far seem to run fine - such as WIC and Bioshock.

However, i've checked the rails on my psu through Easytune and the BIOS itself and the values are as follows:

3.3v - 3.312v
12.0v - 11.795v

Obviously i'm not worried about the 3.3v rail but the 12v rail does all the grunt work and it seems quite low - is it within an acceptable tolerance - or is this the reason that my computer keeps crashing/freezing?

The computer is not clocked and i have run memtest and stressed the cpu. The above is the only problem i'm left with - apart from that the computer seems stable.

I have upped the cpu voltage and i have just upped the pci to 102 - no luck. I'm totally stumped - no idea what to rma even if it is a hardware problem.

Any advice grateful as this is proving to be the most frustrating build that i've ever done.

Plec

Sasahara · 10 Apr 2009 at 21:21

The voltages you've shown should no be causing the crashes.

Is the crash a complete lock up, or blue screen?

Plec · 10 Apr 2009 at 21:29

Thanks for the reply it's really appreciated. I'm fast losing my patience and sanity with this build. No sooner than i solve one problem then another props up - but this is the final provlem - and has probably been the only problem. I've already spent way too much of my holiday trying to trouble shooting this machine.

It's a complete lock up but not of the image it was last on. It's always a striped green screen on crysis and an orange striped screen on furmark but same stripe graphic - although furmark is sometimes just a grey screen. But no, no blue screen.

Any of the above of help?

Plec

Sasahara · 10 Apr 2009 at 21:37

Could be possibly be a RAM problem, system or Graphics card.

Easy to check the System ram, just use 1 stick and test.

If it's the card the only way to really check would be to put it into another system or try a different card.

Plec · 10 Apr 2009 at 21:47

Thaks again for the reply.

I've ran the latest memtest and had problems at first when run in dual channel - but after wiping the CMOS this was cured. I've also checked both stick independentlty since and they both check out fine. TBH i think the problem was brought about by the system freezing all the time due to freezing problem.

Hadn't thought of a gfx memory problem i was just focusing on heat/power etc. Is the problem i'm experiencing indicative of a Gfx memory problem? And if it is would it be easily simulated if i was to RMA the card?

Plec

Sasahara · 10 Apr 2009 at 21:56

I can't be 100% definite, but it is certainly one of the signs of a bad card.

Wait a bit for some other replays, someone else may have something to add, or backup what I've been saying.

Plec · 10 Apr 2009 at 22:04

With todays modern cards and the amount of memory they carry you would think that they would have their own utility for testing it's memory - like a gfx equivalent of memtest - it would take a lot of the trial and error out of testing new rigs...

Just about to try 9.4 cats - and then a complete re-install. After that it looks like i'll be on the long road of RMAing parts until i get the right one. When i get a build like this it almost puts me off building my own rigs and other ppls for good - just so time consuming and frustrating and a lot of other irregular verbs that can't be mentioned on this forum!

Thanks again.

Plec

RJC · 10 Apr 2009 at 22:08

Not sure if I can be of any help - Have you tested different drivers for the GPU or reinstalled current ones.

Like Sasahara mentioned can you test the card in another machine or try another card in your machine.

Was windows a fresh install.

Amleto · 10 Apr 2009 at 22:09

that 12v rail voltge is within the 5% atx specification. you've got a funny display on system freeze and haven't tried testing another gfx card? that would be the first thing I try tbh

If you cant borrow a mates, then the only thing you can do is eliminate other components which means you need spares of those - or just RMA the card and cross your fingers...

(as well as driver updates)

KingZeddy · 10 Apr 2009 at 22:17

have you tried monitoring using http://www.stvsoft.com/ sensorsview, and ATItools

Plec · 10 Apr 2009 at 22:18

Thanks for the replies.

Just tried the 9.4 cats and exactly the same problem.

RJC - trouble is my friends have either aging agp systems of laptops - i may try my loacal computer shop and see how much they charge for testing it. Windows was a fresh install but i'm about to do one last re-install before either taking to loacal shop or RMAing card.

Amleto - i was clutching at straws with that one but it's good to get confirmation. Again i know of no one with a pci-e gfx card so that test is a no go i'm afraid. So as above local shop or RMA.

KingZeddy, no i haven't - but tbh i've never heard of it but i'm willing to try anything, thanks. i've been keeping a close eye on my temps and voltages in easytune but obviously when it crashes i can only see what's in the BIOS after it starts to reboot.

At least the consensus of opinion appears to think it's the gfx card - so i'll start there - after my re-install.

Additional: one thing i will add is that i've had to up my vcore voltage to 1.17XX just so that it reads 1.15 in the bios hardware monitor and in easytune.

Thanks for your input, guys.

Plec

Plec · 10 Apr 2009 at 23:41

Well after yet another complete install - but this time with the CMOS cleared and the BIOS at standard settings - i have exactly the same problem.

I've only tested it with Furmark as the problem is a lot quicker to simulate than using Crysis but the freezing image is the same and also requires a hard reboot.

So, looks like i'll be RMAing this card back as it seems the most likely culprit - i just hope they can simulate the same problem so that i can resolve this problem without tearing other components out. I would be surprised though if it wasn't the gfx card.

Thanks for the help guys i'll let you know how it all pans out - i just hope OC's testing/turnaround is quick otherwise i'll have to beg, borrow or steal a laptop off somebody to get my work done.

Just thought, my brother-in-law has a relatively new system so there may be a chance I could test it in that system just to make certain that it is the card. If it does simulate the same problem then i may just buy a new card from OC's and send this one back for a refund as the turn around would be a lot quicker.

PLec

Plec · 11 Apr 2009 at 10:56

Ok, i have another rig i can try this gfx card in tomorrow. My question is; if i replicate the problem using Furmark in my brother-in-law's computer does that definitively prove that the card is at fault - even though i'm only testing one application?

Because, if it does definitively prove that it's at fault i may just buy a new card from OC's and return this one for a refund as the trun around will be a lot faster. (I've had the card for less than 7 days)

Plec

Quixote · 11 Apr 2009 at 11:08

Pretty much, yes

Plec · 11 Apr 2009 at 13:54

For pity sake! This build is going to trigger an aneurysm.

I had a theory that maybe, just maybe, i had a corrupt download of Furmark as when my system memory was at fault (sorted via a cleared CMOS) all my downloads were corrupted and wouldn't install. So, i d/led and re-installed Furmark and it runs it's tests fine - so far. I'm running crysis benchmarks and they work fine but i'm yet to test the game in play again as i just don't have the time to spend testing a game 20 minutes at a time to see if it crashes at the moment.

The one thing that is consistent is that after my system crashes a few times the dual-channel mode fails memtest - I can cure this by clearing the CMOS. So my question is, is it normal for a system that crashes a lot over a short period of time, for whatever reasons, to corrupt the dual-channel process and require the CMOS to be cleared ti fix it or could the problem be the motherboard slots? (I have ran the mobo with only 1 slot filled with memory.)

Basically, when a computer crashes a lot is it unusual to have to clear the CMOS to get the memory running dual-channel properly again? (It's my first dual channel board - and i've never had to clear the CMOS so often on other setups.)

I think i need to start a new thread listing everything i've done upto date with all the things i've ruled out and start again.

Thanks for taking the trouble to answer guys.

Plec

Plec · 11 Apr 2009 at 20:11

In hindsight i should have edited the above post with the below info - please read my last post and this post combined as they both have relevent information - apologies for adding further confusion to an already messy thread.

Well it just gets more confusing - I noticed the green speaker wire from my antec 5.1 speakers wasn't connected into the back of the pc after i last cleared the CMOS. So I connected it and the problem has re-appeared. (see above post)

So i thought, great, problem identified a faulty lead in socket for the green speaker lead on the motherboard, but no! I've unplugged it again but the problem is still there even after a reboot.

Was this just a coincidence? Was it the speaker connector port? Or is it just some of the problems in my above post?

Every time i seem to eliminate one problem another takes it place - but i'm pretty sure it's the same problem it's just displaying itself in a different way as i resolve different issues that it's causing.

I'm going to try clearing the CMOS again and re-installing Furmark and see if it works again. If it does what does that actually tell me? Does it start pointing to system memory incompatabilty?

Plec

Bob Wya Jnr · 11 Apr 2009 at 20:59

Plec

You think you've got problems... Try building a full watercooled build (ram, chipset, CPU and dual GPU) - then complain!! :eek:

You haven't isolated the problem down enough yet. I would try running some system stability tests that don't stress the GPU. (You know CPUburn, Prime95, etc.)

Unfortunately you can't beat being able to subsitute components. For example I have spare el-cheapo PCIe GPUs lying around to test in my rigs (bought off Ebay).

You got the latest BIOS for the MB I presume??

Have you tried the GPU in another rig yet??

Good luck!! Just be patient and methodical - you can't go wrong!! If all else fails just starting RMA'ing stuff!! :cool:

Bob

Plec · 11 Apr 2009 at 23:27

Bob Wya Jnr said:
You think you've got problems... Try building a full watercooled build (ram, chipset, CPU and dual GPU) - then complain!!

Jesus, if i had water pumping to each component and these problems i would have programmed my router to 'beachy head' 3 days ago.

Bob Wya Jnr said:
You haven't isolated the problem down enough yet. I would try running some system stability tests that don't stress the GPU. (You know CPUburn, Prime95, etc.)

Yeah, you maybe right i'll perform some longer stability tests tomorrow but i'm fast coming to the conclusion that i need to try some different memory.

Bob Wya Jnr said:
Unfortunately you can't beat being able to subsitute components. For example I have spare el-cheapo PCIe GPUs lying around to test in my rigs (bought off Ebay).

Yep, things would be a lot easier if this was the case but unfortunately i've upgraded from an ageing athlon slot 754 machine with an AGP slot - so apart from the PSU nothing else was interchangeable. Bummer...

Bob Wya Jnr said:
You got the latest BIOS for the MB I presume??

Yes, but i'm considering the beta version - but this is desperation on my part.

Bob Wya Jnr said:
Have you tried the GPU in another rig yet??

No, but i have the oppurtunity tomorrow or monday - but after watching it process furmark continuously for 20 minutes i'm beginning to suspect the system memory or the mb.

Bob Wya Jnr said:
Good luck!! Just be patient and methodical - you can't go wrong!! If all else fails just starting RMA'ing stuff!!

Bob

Thanks Bob, sound advice when you're feeling like throwoing £300 pounds of upgrade up the wall. I think i'll RMA the memory under the distance selling act and try some better quality stuff as this would be the cheapest first step.

Thanks for the reply Bob and i hope you're not speaking from actual experience with unresolved problems with a water cooled rig. I really couldn't imagine a more demorilising situation for a new build to go t*ts up.

Plec