Bitcoin Forum

So, I'm having issues with my rig.  2 5970's and a 5830 via extension. They are unstable even when running stock everything (clock, memory, voltage) but with applying flags -v -w128. I don't believe it is a hardware issue.

Here are the details.

windows 7 64 bit
catalyst 11.9
guiminer opencl


what happens is the cards will mine happily (at stock) for awhile but the screen would go black, and recover and the windows message (ati driver had some issue but recovered) and sometimes the entire system would just freeze.

When I first put the cards in, I anticipated relative ease as i've had with my other rigs so started them OC'd to 850 for each 5970 core and 900 for the 5830 core. (5970 stock core is 725 i think).

I've tried reinstalling catalyst a few times, and that doesnt work.

Any ideas? Whover is able to solve this for me i'd be happy to donate!!  :D

also: MSI 870-G45 AM3 AMD 770 ATX AMD Motherboard

Cheap PSU?

PSU?

Heat?

PSU is 1000w rosewill. Used a killawatt and stock values (no flags, no overclocking, no underclocking the memz) while mining it uses at most 750w.

It crashed again during this case as well, meaning that the power draw is not the cause of the crashing. ??? ??? ???

Hmm.. what temps are your GPU cores hitting?  What about VRM temps?

no gpu temp is able to cross the 80c barrier before crashing.

how do I check vrm temps?

my only thought is maybe the motherboard I'm using can't handle 5gpus? I use the same motherboard for other rigs with no problem (3x6950, 3x5830), but that's only 3gpus.

motherboard: MSI 870-G45 AM3 AMD 770 ATX AMD
http://www.newegg.com/Product/Product.aspx?Item=N82E16813130290

if that is the problem I will likely start sobbing, as I'll have to ditch either a 5830 or (gasp) a 5970, plus I'll have extra juice from my power supply going to waste. (I know it's not actually 'wasted', but i would want to utilize everything i have at my disposal.)

thanks for any advice

I have the same mobo, but only 4 GPUs (one 5970 and two 5870s).  You can check the VRM temps with CPU-Z (gonna have to Google that one).  Normally you'd want to see those at 100C or under.  Have you tried taking the 5830 out of the equation and just running on the 5970s?

I'll try that once I get back from work. I can tolerate losing the 5830.


have you ever heard of a cpu you not being powerful enough? I haven't, but thought of trying a f# flag like f11 or higher. I'm glad to hear that mobo can handle at least 4 gpu's though.

I'll report back later with either drunken stories of victory (champagne) or defeat (whiskey)

Set the fans to fixed speed of 70% on each card and see what happens after like 6 hours.  The way your driver is crashing is the way my 5830's crash when they are too hot or the clock speeds are too high.

I hate to say it, but it just might be the PSU...Rosewill does have some winners, but every once in a while you get one that the 12v rail is extremely noisy. Try a different PSU

I've run 3*5970 on this exact board, so > 4 GPUs does work.
Yep, PSU is a possiblility, there's 2 current Rosewill 1kW units:
RBR-1000M = Andyson OEM. Had half a dozen 1kW PSUs based on the same Andyson platform crap out after ~3 months while only running dual 5970s.
LIGHTNING-1000 = SuperFlower OEM. Seems pretty good, so far had zero of those die, and thats with 2*5970+2*5770 or 3*5970 on em.
Anything older... not sure, probably not too good.

I have gotten them stable, overclocked the 5970's to 800 (mem to 300) and the 5830 to 925 (mem 300) with -v and -w128. The solution was reverting back to catalyst 11.8.

Artforz: they are the RBR-1000M, so that is disappointing. Hopefully they die within the 30 day refund policy from newegg  :D