I've been having this problem for some time now and it's caused me to stop running Seti here on my work computers :(

I have two systems that are on 24/7. One I built just for Seti, my other is my workstation. Specs are outlined below for reference.

Both systems have fresh installations of Windows 7 Ultimate 64bit SP1 with all the updates and Microsoft Security Essentials antivirus. Everything is up to date and nothing is running in the background. Zero chance of infection.

Both systems have driver version 14.1 (CCC ver 14.6) which is the newest from ATI/AMD's site. I did also try the newest beta driver, 14.2 (CCC ver 14.7), with no change.

The systems boot up normally and the client loads in the system tray with a red mark to show that it's still loading. Soon as the pause mark goes away the system starts lagging badly. Very soon after, the screen goes black and you're left with a frozen cursor. The CPU and GPU fans still ramp up and down and there seems to be hard disk activity every now and then, but as far as I can tell the system is hung. Nothing changes if you wait even for days.

One time I did see a popup from either Windows or AMD that said the display driver had stopped responding and was recovered.

I'd chalk this up to some incompatibility with the system or its drivers and the Boinc software but I have TWO that do the same thing!

Neither system is overheating either!

Are there logs I can check to possibly find the solution?

System specs:

System 1:
Asus P5Q-E
Intel Core2Quad Q9400 (stock clock, Zalman cooler, new paste)
8gb DDR2
Sandisk Ultra 120gb SSD
2x AMD Radeon HD 5770 1gb in Crossfire

System 2:
Foxconn A74ML-K
AMD Athlon 64 X2 5000+ (stock clock, stock cooler, new paste)
4gb DDR2
WD Velociraptor 80gb
AMD Radeon HD R5 230 2gb

Such technical issue is best discussed in Number Crunching

Post there with link to this thread (to not repost the same info)

You didn't say what brand/model are the PSUs

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

Do know that the default BOINC preferences set up is that the GPU is only used when the system has been idle for N minutes. If you changed that to "Use GPU always" you will experience the lag at Windows start-up. Perhaps for the course of testing time to set it back to "Use when idle"?

Edit: Or if you must want to start BOINC at Windows login, have it start with a delay. You can do this by using Windows Task Scheduler, create a new task, have it start BOINC Manager (boincmgr.exe), Trigger: At log on, Delay task for xx minutes (I'd say the default 15 minutes to allow for all background programs to fully load).

Black screens sound like technical (hardware) problems, or that the power supply unit isn't strong enough to cater for the GPU(s). Checking your system's tasks, you only have one error thus far in this result, which has likely more to do with the values for Spike count: 2, Autocorr count: 6, Pulse count: 8, Triplet count: 5, Gaussian count: 0 than anything else. So on the whole, the systems do seem to crunch OK, it's just that you don't see anything, which leads back to a technical (hardware) problem.

I have a HD6850 here which will cause the whole desktop to go black unreadable, when I uninstall its drivers prior to installing new drivers. The desktop is still there, I can move the mouse over it, I can start programs, I just can't see them as everything is pitch black. That HD6850 has always done that, in 2 different systems and it'll probably go do that in the TV server, if I ever get enough time to take that down for a day and swap cards over. (Problem there is that the LED-TV that's on it is non-standard, so that really takes a long time to reinstall everything and get it working, hence why I am waiting.)

Anyway, sounds familiar?
My solution at the time was to buy a HD7870. :)

---

I have tried setting the program to both "Run always" or "Run when idle" but that made no difference.

I know what lag you speak of when referring to bootup and program loading, but this isn't that. By lagging I mean the system is barely usable before it hangs fully.

Power supply on System 1 is a 700w PCP&C unit; more than sufficient.

Power supply on System 2 is a basic 450w; also sufficient.

I should also state that NO other 3D programs or games cause EITHER system to crash or behave abnormally! Furmark & Crysis... work fine.

I dont know what all options are in the motherboard bios, but my son's PC had been crashing in games and trying to crunch Seti on the GPU, doing just what yours did, I was going threw the bios on his and in the PWM settings, it is a DFi 790FX-M2R, there was a voltage for PCIe, a little searching I found it was way to low, cranked it up a tad and now runs flawless.

Like I said I dont know your bios, but maybe have a look around in the power settings, there could be something there....

---

I have tried setting the program to both "Run always" or "Run when idle" but that made no difference.

If it still happens at Windows login, with BOINC only being able to use the GPU(s) when the system is idle, then you'll have to look elsewhere for causes.

What does Windows Event Log say for errors on System and Applications?

---

No no, I meant the problem will happen regardless of when I start Boinc. If I disable the two entries for it in MSCONFIG the PC boots and behaves normally.

I will check event viewer later at work today and let you know, but I do know it's the video driver that's crashing as I did see that error once before the screen went black.

That's why I tried a fresh install of everything to see if it was corruption; but it wasn't.

Wasn't sure if there were known issues with the newest ATI drivers or anything like that.

What happens if you only allow BOINC to start CPU apps?
(i.e. 'Suspend GPU' from Activity menu)

Your 'Core2 Quad' system have Last contact 12 Jul 2014
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7254457

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

CPU only works fine. I'm 100% sure it's a driver issue with the ATI software as that's the only common denominator between both systems having issues!

Like I said in the initial post, both systems have fresh updated installations of Windows, drivers, and the Boinc software.

OK, then it may be possible to reduce the load/demand on the GPU/driver


1) From ReadMe_MultiBeam_OpenCL_ATI.txt
"
-period_iterations_num N: Splits single PulseFind kernel call to N calls for longest PulseFind calls. Can be used to reduce GUI lags or
to prevent driver restarts. Can affect performance. Experimentation required. Default value for v6/v7 task is N=20. N should be positive integer.

-cpu_lock :Results in CPUs number limitation for particular app instance. Also attempt to bind different instances to different CPU cores will be made.
Can be used to increase performance under some specific conditions. Can decrease performance in other cases though. Experimentation required.
"
(-cpu_lock also reduces lag)


To apply those Settings - go to SETI@home directory (<BOINC_Data>\projects\setiathome.berkeley.edu\)
and find a file with similar name: mb_cmdline_win_x86_SSE_OpenCL_ATi_HD5.txt (it will be empty, have zero length)
and put the following line in it (higher N = less load) (to make the app see the new cmdline - pause/resume the task):
-cpu_lock -period_iterations_num 80


2) It seems you don't accept AstroPulse now
ATI AMD GPUs are most effective and have much less lag for AstroPulse (compared to SETI@home v7 (MultiBeam))

From ReadMe_AstroPulse_OpenCL_ATI.txt
"
-unroll N :sets number of data chunks processed per kernel call in main application loop. N should be integer number, minimal possible value is 2.
"

If/after you start to 'use' AstroPulse - Find a file with similar name: ap_cmdline_win_x86_SSE2_OpenCL_ATI.txt
and put the following line in it (lower unroll = less load, I think -unroll 2 is the default now for AstroPulse v6 but the default will change for AstroPulse v7 (still not deployed here))
-cpu_lock -unroll 2

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)