5830 has become unstable

[Maniac479]

I've been happily mining away on my 5830 for over a month and then a few days ago it became unstable. After a couple or three hours the computer crashes and reboots. I haven't been around the computer when this happens, but when I get back the computer is no longer mining and is asking for the password to get back into my desktop. At first I thought it had something to do with the latest round of MS updates (Win7) so I did a rollback of my system, but that didn't help. I upgraded poclbm to the latest version and that didn't help. I use GUIminer.

I got a program called BlueScreeView to determine what was causing the crashes and it says it's the "ATI Radeon Kernel Mode Driver"

I'm running out of things to try, suggestions appreciated. 

[crazyates]

Mind posting your version of Windows/driver/SDK/miner?

It's also a significant possibility that you're card is on it's way out. Those 5xxx cards are a few  years old now, and bitcoin mining is hell for a card when run 24/7 for 18 months straight.

[bcpokey]

I've been happily mining away on my 5830 for over a month and then a few days ago it became unstable. After a couple or three hours the computer crashes and reboots. I haven't been around the computer when this happens, but when I get back the computer is no longer mining and is asking for the password to get back into my desktop. At first I thought it had something to do with the latest round of MS updates (Win7) so I did a rollback of my system, but that didn't help. I upgraded poclbm to the latest version and that didn't help. I use GUIminer.

I got a program called BlueScreeView to determine what was causing the crashes and it says it's the "ATI Radeon Kernel Mode Driver"

I'm running out of things to try, suggestions appreciated. 

Agreed with Ates, my suggestion would be to try turning down the clocks on your 5830, when mining see if that helps. Can also run GPU stress tests (like 3dmark) to see 1) if it crashes and/or 2) if you get video errors to help see if the card aging and abuse has finally gotten to it.

[superfastkyle]

5830's have a way of running at extremely high clocks 1000 plus for a few months and then you will have to lower it until they cease to function at all. You may have to lower your gpu clock to 950 or even lower to get it stable again. Its why I don't recommend 5830's for anyone, as all of mine have failed, most within a year. In fact its the only video card model I have had fail mining, thus far. If you bought it used chances are the previous owner already noticed it on its way out.

If I remember right 5830 is the only card that runs at a higher voltage (usually 1.163v) with a (correct me if I'm wrong) 3 phase vrm. Most have a 4 phase or run at lower volts (1.088, 1.05) Its probably best to run a 5830 undervolted for a longer life but it may be too late for that

[Maniac479]

Can someone explain clock speed? Is this the same as "core"? The slider on 50miner makes this easy to control- not the case on any other guiminer like GUIminer, so I'm switching from GUIminer to 50miner. I mine on 50btc anyway.

There are four sliders on 50miner:

Core ( 80-875 MHz)
Memory (150-1200 MHz)
VDDC (.950-1.163 V)
and Fan Speed (1-100%)

The only one I really understand is fan speed. I now think it's possible that the card was overheating- because without any tweaking (auto), the card used to run automatically at 70% fan speed. With 50miner set on auto fan speed the fan speed was down to 40%. Why would this be? I used to mess around with all these settings using MSI Afterburner or the overclocking controls that come with the driver, but the card used to run fine without any tweaking. Now it seems to want to run at a lower fan speed. So I have bumped the card up to 70% fan speed and lowered the other controls. And yes, I did get this card used on Ebay. I've been getting around 250 mega hash with it for a little over a month.

If this one fries or is fried, I will try another one... but I wouldn't want to spend more than $200 on one.

Also 50miner displays the temperature. I don't think GUIminer does. How do I know what temperature should worry me? It's running at 60C/140F

[Meatball]

You said you've only had the 5830 running for a few months.  Did you get it off of eBay or know if it was used/overclocked previously?

One possibility is the Radeon drivers.  The last few versions of the Catalyst drivers are a bit hinky with mining.  If possible, use the 12.8 catalyst drivers.

Also, the default clock speeds for the Radeon 5830 were 800 Mhz Core and 1000 Mhz Memory.  Try running at those speeds and if it still crashes, lower the core/memory in 25 Mhz intervals till you find a stable speed.

[bcpokey]

Can someone explain clock speed? Is this the same as "core"? The slider on 50miner makes this easy to control- not the case on any other guiminer like GUIminer, so I'm switching from GUIminer to 50miner. I mine on 50btc anyway.

There are four sliders on 50miner:

Core ( 80-875 MHz)
Memory (150-1200 MHz)
VDDC (.950-1.163 V)
and Fan Speed (1-100%)

The only one I really understand is fan speed. I now think it's possible that the card was overheating- because without any tweaking (auto), the card used to run automatically at 70% fan speed. With 50miner set on auto fan speed the fan speed was down to 40%. Why would this be? I used to mess around with all these settings using MSI Afterburner or the overclocking controls that come with the driver, but the card used to run fine without any tweaking. Now it seems to want to run at a lower fan speed. So I have bumped the card up to 70% fan speed and lowered the other controls. And yes, I did get this card used on Ebay. I've been getting around 250 mega hash with it for a little over a month.

If this one fries or is fried, I will try another one... but I wouldn't want to spend more than $200 on one.

Also 50miner displays the temperature. I don't think GUIminer does. How do I know what temperature should worry me? It's running at 60C/140F

Clock speed simply refers to the frequency at which computational device runs. In your case, my suggestion would be to reduce the clock speed of your Memory (Meatball's suggestion of doing this increments is good, but ideally you will want to be in the low triple digits, so 25Mhz at a time might be a bit extreme), and your Core. You can also play with the VDDC (lowering) as you lower. It will be a lot of little tweaking, to find a stable point, and you might not find one if the GPU is just dying.

60C is nothing to worry about. I believe most people suggest 85C as an upper limit, and I ran my cards over 90C for months (I regret this, but what's done is done).

I still also recommend a video stress test, as I suggested before. Visual feedback is very helpful, and easily understood (clipping/ errors / bad frames etc.)

[Meatball]

Yeah, I _used_ to run some 5830's and some were able to run at around 960 Core and 185 Memory and others were stable at 940/300.  That was a while back with Catalyst 11.7's, but I'd pull about 300-310 MH/s each with them.

[kwoody]

5830 has always been a strong card for mining, the 2 in my farm still function from 18 months of almost nonstop 960/300 clocks. for a long time it was the king of mhash/cost of GPU. $109 Sapphire models from newegg, for 300mh/s 2 years ago was a steal

[superfastkyle]

my $129 sapphire models from newegg (7/2011 I believe) were dead at the beginning of summer. But since I redid thermal paste to fix heat issues, no warranty.

5830 has always been a strong card for mining, the 2 in my farm still function from 18 months of almost nonstop 960/300 clocks. for a long time it was the king of mhash/cost of GPU. $109 Sapphire models from newegg, for 300mh/s 2 years ago was a steal

but back to the op try lower clocks if you still can't get a stable clock try 0.95v and 600 mhz I was able to run one of mine a few more months at those speeds

[stoppots]

Check your power settings on your computer. It sounds as if it may be going to the sleep setting and that can cause the screen asking for your password. When this occurs I believe the AMD Display Library or ADL is disabled which may cause your fans to no longer be controled properly.

This would then cause your display driver to crash along with possibly damaging your card although it should shut itself down at point prior to that occuring. I caught this same type of behaviour when using the switch user feature which also results in the password screen. Luckily I had my fans set high enough so that it was obvious that something had was not right when they reduced speeds immediately.