This helps some what with cuda/mb hanging.
for windows XP pro these are the steps
First.. if you are not using a password with your Administrator logon, use this to add a name to the Administrator group. Using XP Pro, You will need a passworded Admin name for the Task Scheduler.

1> click start
2> right click my computer
3> click Manage
4> click on the + next to 'Local Users and Groups.
5> click the Users folder
6> right click in the window area and pick 'New User'
7> type in a name and password. (REMEMBER THESE!)
8> un check 'User must change password at next logon
then click create.
9> click the Groups folder to the right.
10> dbl click Administrators. A window will open.
11> click the 'Add' button'. A window will open.
12> Click Advanced. A window will open
13> click Find Now Button.
14> At the bottom of the window, look for your new created name from step 7 above. Double clikc that name.
15> click the 'Ok' button.
16> your new name is now part of the administrator group.

Next..
Create a scheduled task...

1> goto control panel..
2> dbl click Scheduled Tasks.
3> Click the Add Scheduled Task ICON. A window will open
4> Click next
5> click browse. Look for a program in the BOINC folder called 'boinccmd.exe' and dbl click it once its found.
6> in the next window. Check Daily then the 'Next' button
7> click the 'Next button again.
8> type in the new Admin name you created above or use 'the default if it IS passworded.
9> enter the password and confirm
10> click next.
11. Check the Open Adavnced properties for this task.... and then click the 'Finish' button
12. you sould see something like this...
"C:\Program Files\BOINC\boinccmd.exe"
add this to the end after the quote. --run_benchmarks
13> click the Schedual tap at the top.
14> click the 'Advanced' button
15> check the Repeat task box and set the every box to 2 and change the minutes to hours.
16> in the duration box, change that to 3.
17> click ok
18> click ok again.

Your task should be scheduled to run every 2 hours.
You can test it by right clicking on the task and picking 'run'
If you get a message to the right of the task saying 'could not start' or something close, double check the name and password you used to set it up.
Just double click the task to edit it.

I hope this helps..
until the bugs are worked out in CUDA and BOINC sharing properly.

---

Not Ready Reading BRAIN. Abort/Retry/Fail?

This was Borgholio's Idea. Credit where credit is do.
Thanks for the input in the other thread Borgholio.
So far this is working good.
Can't wait till the bugs are fixed with cuda and boinc.

---

Not Ready Reading BRAIN. Abort/Retry/Fail?

This was Borgholio's Idea. Credit where credit is do.
Thanks for the input in the other thread Borgholio.
So far this is working good.
Can't wait till the bugs are fixed with cuda and boinc.

Glad it's working for ya. I've set a batch file to run benchmarks on all my CUDA systems, so it's pretty smooth sailing. I have found however that there are occasionally some hanging workunits that fail to start even with benchmarks running. For those, you'll need to completely restart BOINC.

---

You will be assimilated...bunghole!

Instead of the "--run_benchmarks" command, could a person use the "--quit" command? I notice when I do this I disconnect from localhost but then like clockwork it seems 1 minute later the machine reconnects to localhost and restarts its tasks.

I tried this when 1 of my AP WUs got stuck and it seemed to restart it without any issues.


Is this a bad practice or is this ok?

Thanks

---

Boinc Wiki




"Great spirits have always encountered violent opposition from mediocre minds." -Albert Einstein

Instead of the "--run_benchmarks" command, could a person use the "--quit" command? I notice when I do this I disconnect from localhost but then like clockwork it seems 1 minute later the machine reconnects to localhost and restarts its tasks.

I tried this when 1 of my AP WUs got stuck and it seemed to restart it without any issues.


Is this a bad practice or is this ok?

Thanks

It does seem that way to me as well. I suspect that when the BOINC mananger is open and BOINC is not running as a service, it will auto-restart the BOINC client after one minute.

---

You will be assimilated...bunghole!

Instead of the "--run_benchmarks" command, could a person use the "--quit" command? I notice when I do this I disconnect from localhost but then like clockwork it seems 1 minute later the machine reconnects to localhost and restarts its tasks.

I tried this when 1 of my AP WUs got stuck and it seemed to restart it without any issues.


Is this a bad practice or is this ok?

Thanks

It does seem that way to me as well. I suspect that when the BOINC mananger is open and BOINC is not running as a service, it will auto-restart the BOINC client after one minute.

I noticed today that 1 or 2 of my rigs stalled after updating to 6.6.20. They stall after I run the --quit command. They simply don't reconnect to the localhost. I have changed the command line back to --run_benchmarks and they don't stall anymore. I also set my machines to restart every 6 hours just incase WUs are locked up and the --run_benchmarks command doesn't fix them.

---

Boinc Wiki




"Great spirits have always encountered violent opposition from mediocre minds." -Albert Einstein

After CUDA was announced last December I purchased 4 9800 GTX and GTX+ video cards and put them to work for Seti. One of the four GPU's started freezing up while crunching. Sometimes once per day sometimes many times per day. I dealt with it as many have dealt with the problem for a long time. Finally I decided to swap the card into one of my other computers and see if the freezing moved to the new computer or stayed on the old one. Immediately I had freezing problems in the computer that had never had a problem. To me this indicates a hardware problem, not a software problem.

These GPU's are marketed with CUDA capability being a large part of the marketing strategy. I requested and was granted an RMA from the manufacturer stating that the GPU consistantly froze while performing CUDA tasks as the reason. I have since received another identical GPU and it is crunching fine.

I would hope that those of you who are still in the warranty period would investigate the possibility of getting an RMA and exchanging the GPU. Don't just work around the problem with hacks and batch files unless you have no other choice.

---

Boinc....Boinc....Boinc....Boinc....

After CUDA was announced last December I purchased 4 9800 GTX and GTX+ video cards and put them to work for Seti. One of the four GPU's started freezing up while crunching. Sometimes once per day sometimes many times per day. I dealt with it as many have dealt with the problem for a long time. Finally I decided to swap the card into one of my other computers and see if the freezing moved to the new computer or stayed on the old one. Immediately I had freezing problems in the computer that had never had a problem. To me this indicates a hardware problem, not a software problem.

These GPU's are marketed with CUDA capability being a large part of the marketing strategy. I requested and was granted an RMA from the manufacturer stating that the GPU consistantly froze while performing CUDA tasks as the reason. I have since received another identical GPU and it is crunching fine.

I would hope that those of you who are still in the warranty period would investigate the possibility of getting an RMA and exchanging the GPU. Don't just work around the problem with hacks and batch files unless you have no other choice.

Actually, As a computer retailer and network consultant I have to say that creating an RMA is always my absolute last resort. There is nothing worse than neadlessly sending in perfectly good equipment before all possible testing is done. It might be easy for some people to simply walk into Future Shop or Best Buy and return a card with no further thought but as the owner, when I have to RMA anything I (my business) has to pay to ship the card back to the supplier.

If I shipped everything back that was thought to have issues, I would have gone broke a long time ago. Simple testing and/or experimenting with troublesome equipment has saved me thousands in shipping costs. Also, fixing problems for customers as they watch inspires confidence and encourages repeat sales.

I suspect that a large part of my particular problem is a direct result of overclocking. Although I consider the equipment I use to be top notch, I am sure that the farther I push it increases the chances of it having minor hickups. Since I can play games like Farcry 2, Crysis, Fear 2 and Conan without any hickups at all, I suspect that these hickups are likely to do with the CPUs and this 9800GTX+ DK being wound up to 100% usage almost 100% of the time.

Anyways, Thanks for the thought Geek@Play. It is easy for me to forget about hardware when I am geeking out with Seti. I like pushing my rig as fast as it can go. Being reminded that software isn't always the issue might help to save a lot of people a lot of time. At minimum, if they get replacement equipment and the problems persist, it might clue them in to look at other possible problems.

---

Boinc Wiki




"Great spirits have always encountered violent opposition from mediocre minds." -Albert Einstein