Author

Topic: Connection times out on ONE GPU (Read 1366 times)

member
Activity: 77
Merit: 10
June 16, 2011, 09:55:56 AM
#9
I am getting this with stock clocks as well :/

I am currently working on a batch files, that restarts GUIMiner every 10 seconds
sr. member
Activity: 378
Merit: 250
June 15, 2011, 08:39:43 PM
#8
Overclocking a card too long or too much heat could be having a detrimental effect on it. 
member
Activity: 77
Merit: 10
June 15, 2011, 08:16:28 PM
#7
I just don't understand why I am getting this all of a sudden, when I have mined before without issues!
sr. member
Activity: 378
Merit: 250
June 15, 2011, 07:45:42 PM
#6
Yeah, that verification failed is almost always a card overheating/oc'd too high.  Especially when you have multiple cards in a machine those cards in the middle get their airflow choked off and the OC that works fine for the cards on the outside is just too high for that middle card.

You could set the OC individually for each card and lower the speeds on just the card that is having problems.  Other option is to get more airflow through the cards.  Get better fans or even pull the cards out of the case with some cable extenders and give them some room between each other to breath.
member
Activity: 98
Merit: 10
June 15, 2011, 12:18:09 PM
#5
I had the same problem, 1 GPU kept disconnecting after 20-30 mins.  At first I thought it was the pool, so I switched pools.  Same problem.  Lowered the overclock and problem went away.
sr. member
Activity: 378
Merit: 250
June 14, 2011, 10:21:56 PM
#4
That error is almost always a hardware issue.  Either it's overclocked too high or not getting enough power.  I'd drop the OC down 10 MHz and let it run for a while and if it still crashes, try another 10.  Also, with mining you don't need memory clock speed.  Underclock that.  I have my 5850's running at 860 GPU Clock and 300 Memory clock and they've been rock solid, getting me almost 350 MH/s.  (If you're using poclbm, use -v -w 256 and -f 1 for the flags).  Of course, depending on the card, your mileage may vary, but definitely try to get the memory clock lower.  It'll run much cooler.

Make sure you're using MSI afterburner for overclocking as well.
member
Activity: 77
Merit: 10
June 14, 2011, 05:38:17 PM
#3
Bump
member
Activity: 77
Merit: 10
June 13, 2011, 01:24:36 PM
#2
This is the error i get before it shuts down:
2011-06-13 20:22:36: Listener for "Mine Prim": Verification failed, check hardware!

Has it been overclocked too much?
member
Activity: 77
Merit: 10
June 13, 2011, 04:19:16 AM
#1
Hello!

I am currently mining with my two HD5850 @ BTCMine, and ONE of my GPU's keeps timing out!
It's overclocked to it's absolute MAX, 870 Core Clock and 910 Memory Clock, pushes out 340Mhash/s, has it been overclocked too much? Is this what's causing it to crash or something?

Why does my GPU time out, when my other one (not overclocked), works properly?
Jump to: