Author

Topic: Driver lockup (Read 1430 times)

full member
Activity: 210
Merit: 100
January 10, 2012, 12:35:43 PM
#10
No problem.
I wish I'd gotten a bitcoin every time I made an ass of myself Smiley
newbie
Activity: 15
Merit: 0
January 09, 2012, 07:03:28 PM
#9
Okay, this is VERY embarrassing!!  Undecided

Seems I've fixed the "problem" now

I didn't realise when you install win7 on a desktop it's default setting is to hibernate after 30 min!
As I didn't have a screen or anything else connected to the rig and the rig stopped sending in shares and I couldn't connect via VNC I wrongfully thought that the computer had locked up.

It's been running now for 40 minutes with default clock so now to the overclockin'

Thanks again for all the help! I've never felt so stupid! Cheesy
full member
Activity: 210
Merit: 100
January 09, 2012, 06:04:34 PM
#8
Still crashes with just one GFX directly on the mobo :'-(

Ok, let's do something about that, shall we?
Do you mean that you tested the mobo four times, each time with a different GPU?

Do you think you could download a ready-to-use mining Linux distro, such as LinuxCoin(1) and try mining using that? LinuxCoin can be installed on any sufficiently big flash drive and won't modify your main OS.
The distro comes pre-equipped with amd drivers, sdk, and a miners.
Test your stability using that.

If the rig crashes again, we'll take a look at CPU and memory stability.


Links:
(1)  http://www.linuxcoin.co.uk/
full member
Activity: 140
Merit: 100
January 09, 2012, 02:52:48 PM
#7
Not sure what happened, but I have had a few abnormal crashes after 12.1, even after using drivers weep etc. strange.


It seems to be pretty stable now though, I left the miner off for a couple days.
newbie
Activity: 15
Merit: 0
January 08, 2012, 07:23:52 PM
#6
Sorry for double post but this is an update Cheesy

The rig seems to run stable with the one GFX card.

So I'll be adding more cards now but because the last PCI port on the card is dead I'll only be able to run 3 cards without extenders.
As I see it then I have to use 2 extenders if I want 4 cards on the board because of the layout of the ports.

I'll try with 3 cards directly on the mobo now.

Be back with an update in a minute!

-- Update --

The 3 cards are running now and I remember now why i bought the extenders in the first place. Much more quiet when the cards are not directly alongside each other Cheesy

I hope that I'll be able to use the 2 extenders so I'm able to use all 4 cards on there ..

-- Update2 --

Seems my PCI was not dead after all! Cheesy
All 4 cards are running directly from the mobo at this point with a little overclocking.
I'll test this while I sleep and then I hope I'll be able to get just 2 cards on extenders tomorrow!

-- Update3 --

Still crashes with just one GFX directly on the mobo :'-(
newbie
Activity: 15
Merit: 0
January 08, 2012, 06:08:56 PM
#5
Thanks for the great advice!

I've just run SIW on the rig with the miners working and the only difference I see is a drop in the 12v rail from 11,97 to 11,88.. So I guess the problem isn't that they are not getting the power they need?

Although SIW didn't report any problem i guess I should take the miner apart and put one GFX directly onto the board.

If I can't use the extenders then I have another problem! I'm pretty sure the PCI port forthest away fra the CPU is broken because I couldn't register any GFX on that port (I switched to the last x16 port with the extender but I'm not able to without the extenders)

Hope you stick around for some more advice because I really appreciate it :-)

-- Update --

The rig is now running with 1 GFX directly on the MB and I'll see if it crashes in the next 30 min or so.
full member
Activity: 210
Merit: 100
January 08, 2012, 05:45:03 PM
#4
What you can do quite easily, is run a program which will read the temperatures and voltages, for example SIW (System Information for Windows)(1)
That'll give you some idea as to what's going on under the hood.
The M1000, being a group-regulated design, wasn't the most stable of power supplies. I don't say it's bad, just imperfect.

The +12 volt rails should never drop under 11.4 or go over 12.6 volts.
5 volt rail should always be between 4.75 and 5.25 volts.
3.3 volt rail should be between 3.13 and 3.46 volts.
Power noise and ripple can't be measured without a oscilloscope.

If you see nothing out of spec, please try temporarily removing all the extenders and run a GPU plugged directly into the mobo.
If the rig hangs again, change that card to another.
Report whether or not you can get this thing stable.
Should the rig stabilize, add as many cards as you can without using the extenders. If everything still works fine, re-apply the extenders.

It's a lot of work but it's bloody damned hard to troubleshoot a (suspected) hardware issue remotely.
Best of luck!


Links:
(1) http://www.gtopala.com/siw-download.html    Choose the freeware version, the link is at the bottom of the page. No installation required.
newbie
Activity: 15
Merit: 0
January 08, 2012, 05:21:39 PM
#3
I have tried running only 1 card at a time but I'm still not seeing the stable rig I once had. (Had it running for 3 months without any hickups)
I forgot to include the PSU in the description. It's a Cooler Master M1000 (1000w)

I have also add'ed PCI extenders this time to get better airflow and lower temps but maybe they are the root to my problem?
I know it's possible to run a stable rig on windows so that is what I'm going for.
I don't have enough Linux knowledge unfortunately.

I'm sitting here thinking about removing the extenders to see if that would fix the stability problem but I'm so proud about my rig-enclosure Cheesy

Any other thoughts?
full member
Activity: 210
Merit: 100
January 08, 2012, 05:10:59 PM
#2
You should have thrown Windows out, not the plugs.

Drivers don't "just crash" out of the blue with such amazing regularity.
Let's face the facts, at least one of your cards isn't stable for whatever reason.
What PSU is running that rig of yours? Accurate model and wattage would be most welcome.

Try temporarily removing one card and see whether the problem persists.
newbie
Activity: 15
Merit: 0
January 08, 2012, 03:50:48 PM
#1
Hi there forum!

I've been setting up an old miningrig lately and since I threw out my dummyplugs I had to update the drivers. (The rig ran like a charm before)
I installed 12.1 and it BSOD'ed so I went ahead and installed the thing from scratch.
12.1 had horrible speed and crashed after not very long (no overclock)
I formated again and installed 11.9 because I have that running fine on my main rig. That restored the speed but it still crashed after 30-40 minutes.
After that I tried going oldschool with 11.6 and SDK 2.1 but then AOCLBF couldn't see all the GFX cards and I was lost as to what I should do about that.
I then installed 11.8 with SDK 2.5 and that is where I'm at now. It still crashes after about 30-40 minutes and is overall just running crappy.
I didn't know what to do so I ran win update just to try something but it didn't solve anything.

Windows 7 Ultimate
1gig RAM
Sempron 140
MSI 890FXA-GD70
4* Asus 5850 DirectCU

Phoenix 1.72 with AOCLBF 1.81 (phatk2, no overclocking, tried different AGG and worksizes but it hasn't made a difference.)
ATI Catalyst 11.8 (SDK 2.5)

I think the GFX driver crashes right before the rig locks up. I can't VNC to it and the mouse is frossen. (I have to restart)

Has anyone experienced this or does anybody know how to fix it? I don't know if I've missed something?

Thank in advance :-)
Jump to: