5870 going insane. And taking me with it!

Firenze · 6 Mar 2011 at 17:50

ok so both mates forgot to bring their towers for me to test the card in a different PC... Joy.

I'm very wary of playing with volts tbh, could do with a little advice?

Any other suggestions at this point?

Thanks

Firenze · 6 Mar 2011 at 21:23

I've been had some friends over for a bit of a celebration this weekend, so haven't done any thing since underclocking the CPU and GPU in turn...

Like I say I'm wary, mainly because I've no real experience in playing with the volts. What are the areas I should test? and by how much?

Currently I'm running a Memtest suggested by OCUK, on its 3rd pass atm, still no errors. Quite confident its not a memory issue myself...

Firenze · 6 Mar 2011 at 23:37

well we underclocked the CPU to like 1.2 GHz, and the problem went away, and the volts and temps were normal, then set it back to stock and underclocked the GPU by 250MHz and the results were the same. I set both back to stock, and the problem came back...

Doing a memtest as although I'm sure it'll be fine, I want to run 10 passes just to be 100%.

Firenze · 7 Mar 2011 at 01:13

ok Memtest is nearly done, what run do you want me to do on Prime95? (there's like 3 isn't there?) will make sure its priority 1 in the morning.

Firenze · 7 Mar 2011 at 09:50

I set the CPU to 3.6GHz, (400 from 333), and set it going, monitoring temps as it went, core 3 has just stopped, an error during 10k testing... I'm a little worried.

The temps got to at a maximum 64 on the first two cores and a maximum of 56 on the second two..

The other 3 cores are still testing.

Firenze · 7 Mar 2011 at 13:03

the VCore is at 1.408v but there's little jumps to 1.424v...

Firenze · 7 Mar 2011 at 14:35

JAKUS said:
What ?? why ? that is way high...I suggest You load bios setup defaults and do some googling on overclocking and stability testing for Your motherboard setup.

I said up the Front side bus 50mhz , I don't know what the Multiplier is on that cpu of the top of My head, 8.5-9.5 ? well the FSB x the multi gives the CPU frequency, problem is most motherboards automatically ramp the voltage when you increase speeds ( makes it hard to test stability if the goal posts move)

I'm assuming You are still at school ? do You have any smart mates to help you ?
It sounds like a simple problem but without the familiarity with high end PC's and their settings it's gonna be tough for You to nail down and Me to help.

OK, I appreciate the help you've provided, but the condescending tone is definitely not appreciated.

I've never played with the voltage specifically as I've pointed out before, I did up the FSB as you suggested, and provided the results. Reading on a review site (the first after googling) they put the volts up to 1.45v to maintain an OC past 3.55Ghz, so doesn't seem like its that mental. Also, why would it be the issue under load, we've already looked at the volts under the issue, and they were fine? I've never played with the volts at all, so why would it suddenly become the problem when I've replaced the graphics card? especially when its with the same card?

Firenze · 7 Mar 2011 at 14:39

Oh, and I'm a University graduate actually.

Firenze · 7 Mar 2011 at 15:13

Treznor said:
I've had the exact same thing with a XFX card that I recently had back after an RMA, it wont run at stock clocks, all my research has suggested that it needs more volts, prolly a bad batch of cards. I'm quite fed up of of the 5870 too, had problems with XFX, never going to buy that brand again

Currently the only way I can keep the GPU stable is to underclock to 825mhz on the core...I should just RMA it again, as it wont run at its default clock and advertised speed

Yeah I'd do that. Its my first XFX card, and the reason I got it was because of the premium warranty that OCUK offered on it. They've been a bit cheeky with it, but a small nudge in their direction has always set them straight.

I have to say I'm glad someone else feels that JAKUS was being a little bit out of line. I may have found the solution however, even if he was wrong at the end, we missed it half way through, when monitoring temps and volts. Mainly because I was unsure what figures I should have been looking at, but my CPU on stock clocks, was being run at 1.5v, instead of 1.3v.... I've set my mobo straight and it seems to have solved the issue, and brought my CPU down a few degrees! currently idling on 10C. not too shabby on air!!

So thank you JAKUS, however, in future a little bit more of a bedside manner and a little less hypocrisy would go a loooong way.

Firenze · 7 Mar 2011 at 16:14

OK, although the game no longer crashes at the map change, there is still a flash of artifacting. I'm worried about this, and wonder, is there any suggestion as to where the problem might lie?

Firenze · 7 Mar 2011 at 16:28

JAKUS said:
Your Pc fails with the new GFX card because all of a sudden it's fitment has increased Loads on the PSU , Loads on the Cpu and motherboard and thermal loads on everything.

More so than the exact same card that was fitted before hand?

JAKUS said:
Lot's of people fall into the same trap...It was fine before, card is "bricked" RMA
In Your case the chances of all 3 cards being "bricked" is next to impossible !

next to, but not impossible. Hence why I came to the forums looking for some suggestions as to how to single out a decent testing method that would actually pinpoint whether the GFX is definitely duff or not. Baring in mind that I bought the original card Xmas 2009, and that the replacement was issued in April, and thus there being a fair amount of time between cards. (The first being considerably different to the others)

JAKUS said:
What we know thus far is it's the motherboard/settings or PSU not coping with full load, we have to assume the GFX card is OK

Right, so why? I'm not trying to argue, its just that so far all you've done is said "Do this.... Do that...." and I've no real idea as to what for.

Firenze · 7 Mar 2011 at 17:44

JAKUS said:
I basically don't know weather to laugh or cry ...Why were You doing 10 passes of memtest, Why did Overclockers tell you to do so ? If all you had changed was to a replacement Identical card which was fine with the system.

Here's the rub, I'm now speculating that something changed in the PC "miraculously, or otherwise) that started the chain, and a good chance the original GFX card was fine.. but that's history.

Regarding the "do this, do that" I would not have approached the fault finding any differently with more information on the problems history and I'm sure I always tell people why i would try testing this or that....normally

When you have Your own business You kinda learn to read between the lines with customers problems

Anyway is it all ok now ?

Not really. Battlefield no longer crashes to desktop like it did, but there is still flashes of artifacting on the screen as it changes map. I'm worried that there is something still wrong somewhere in the system and that it is going to get worse, or even take other components with it if it does...

Any suggestions?

Firenze · 7 Mar 2011 at 20:01

cranked up the core and memory by 50/50 seemed alright, then bugged a little as usual when changing maps, then corrupted textures in the next map, and caused the computer to crash when I tried to exit the game..

Firenze · 7 Mar 2011 at 23:09

*shoots self*

Firenze · 7 Mar 2011 at 23:34

nothing, just frustrated.

Just literally tried a game of battlefield with some friends, all on stock settings etc, and the hung completely on the map change, no flicker of artifacting or anything, just screen froze, audio on a horrible noise and system became completely unresponsive.

So although I thought it was at least better after sorting the CPU volts a bit, it appears its not.

Firenze · 8 Mar 2011 at 00:23

I'll be honest, I've no idea how long the CPU has been running at 1.5v, I've never had any issues, and haven't played with it really, the CPU was a prize at i32, and I built the rest of the PC to hold it.

Got to visit Sheffield on Wednesday for a job interview, so I'll try and see a friend and try it in their PC and see if it works... If it does, its upgrade time, if not, its RMA time.

Firenze · 8 Mar 2011 at 17:07

I've run every benchmarking program on Guru3d today, and not one of them seems to have had a problem, loaded up battlefield, completed a round, got to the map change and it broke on me again. There's obviously something wrong somewhere, will have to try the card in another PC to make sure, but I am quite confident the problem is elsewhere..

Firenze · 8 Mar 2011 at 22:01

Firenze said:
, GPU fluctuation is a min of 0.95V to 1.16V,

and the clocks stay at stock, 850/1200

Like I say, I'll take it with me to sheff tomorrow and try in a friend's PC and see what happens..

Firenze · 9 Mar 2011 at 21:01

Just tried the card in my mate's pc, and had the same issue, definitely the card. Just have to get ocuk to check bf, cos the bench tools don't cause a problem...