Ok, from reading other posts I have figured out that when I log into my PC using Remote Desktop, BOINC does not find the CUDA (because Windows uses different drivers on RDP), but what I have worked out that if I log into the real PC, let BOINC start, and then RDP to it, it seems to work fine, and I have been using it like this for a few months now (PC is downstairs, and I often RDP from laptop upstairs to it). However, I upgraded to BOINC 6.6.20 a few days ago, and today when I logged into the PC via RDP (not by going to the PC and logging in), when BOINC started, it didnt find the CUDA (as expected), but it then deleted all the downloaded, processed and part processed CUDA work units! There were about 40 of them, and now they are all gone.

This is what got dumped to the log file:

19-Apr-2009 08:33:53 [---] Starting BOINC client version 6.6.20 for windows_intelx86
19-Apr-2009 08:33:53 [---] log flags: task, file_xfer, sched_ops
19-Apr-2009 08:33:53 [---] Libraries: libcurl/7.19.4 OpenSSL/0.9.8j zlib/1.2.3
19-Apr-2009 08:33:54 [---] Data directory: C:\ProgramData\BOINC
19-Apr-2009 08:33:54 [---] Running under account *****
19-Apr-2009 08:33:54 [---] Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [x86 Family 6 Model 15 Stepping 11]
19-Apr-2009 08:33:54 [---] Processor features: fpu tsc pae nx sse sse2 pni mmx
19-Apr-2009 08:33:54 [---] OS: Microsoft Windows Vista: Ultimate x86 Editon, Service Pack 1, (06.00.6001.00)
19-Apr-2009 08:33:54 [---] Memory: 3.25 GB physical, 6.68 GB virtual
19-Apr-2009 08:33:54 [---] Disk: 74.53 GB total, 20.77 GB free
19-Apr-2009 08:33:55 [---] Local time is UTC +12 hours
19-Apr-2009 08:33:58 [---] No CUDA devices found
19-Apr-2009 08:33:58 [---] No coprocessors
19-Apr-2009 08:33:59 [---] [error] App version uses nonexistent coprocessor CUDA
19-Apr-2009 08:33:59 [SETI@home] [error] No app version for result: windows_intelx86 608 cuda
19-Apr-2009 08:33:59 [SETI@home] [error] No app version for result: windows_intelx86 608 cuda
.... -- This messages repeated 108 times total
19-Apr-2009 08:34:15 [SETI@home] [error] No app version for result: windows_intelx86 608 cuda
19-Apr-2009 08:34:15 [SETI@home] [error] State file error: result 20fe09af.32073.6736.14.8.27_0 not found

Very annoying, now I'll have to download more units on my painfully slow dialup internet connection (I cant get high speed in my area for a few more months). I am postive this didnt happen in previous versions, is this a new 'feature'?

---

I can confim that this occured on my desktops earlier today too. I think its when I remote from my non cuda laptop to my cuda PC then starting bionc on the cuda pc via remote desktop connection. Bionc seems to think the PC doesnt have cuda device and deleted the wu's, could it be looking at my non cuda laptop I wonder? The processor is registered correctly though.

Luckily I had just backed up my Bionc directory so it was an easy fix for me.

...could it be looking at my non cuda laptop I wonder?...

No it's the remote desktop connection, which starts other drivers.

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

SETI@home classic workunits 3,758
SETI@home classic CPU time 66,520 hours

To maybe make it a bit clearer, remote desktop uses it's own drivers to run the graphics. They are not CUDA capable. When Boinc starts it checks the graphics drivers, sees the Remote desktop drivers and chokes.

---

PROUD MEMBER OF Team Starfire World BOINC

I've seen the same thing. At first I had my machine set to keep about a days worth of work downloaded, but when I RD into the machine everything disappeared (But still showed up on my account, about 30 items). It makes sense that BOINC can't access the GPU, but why does it have to delete the GPU WU's? I assume this is more of an issue with BOINC than SETI@home. Isn't there some way to let it keep the WU's, some configuration setting, so they can be continued when the GPU is available again?

If it were so easy.

What happens is boinc sees no CUDA device and deletes the work units.
That is needed for when someone removes a GPU.

One possible work around is an CUDA app from Raistmer. I dont think It will delete them if its in an app info file.

~Bob

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

One possible work around is an CUDA app from Raistmer. I dont think It will delete them if its in an app info file.

By using an app_info.xml you will tell BOINC to use the applications pointed out in that app_info.xml file, but that won't help if the coprocessor isn't being detected. Even then the work appointed to the coprocessor (in all cases app version 6.08) will be deleted as there's no 'hardware' that can do it.

It's probably better to tell Microsoft to upgrade their RDP driver to include CUDA capability.

---

Thanks for the feedback. Its interesting though that when I log on physically to the PC, I can then RDP to it fine again and BOINC continues to process on the CUDA. From reading the setup options, if I set up BOINC to run as a service, it wont use/find the CUDA either?

---

If it were so easy.

What happens is boinc sees no CUDA device and deletes the work units.
That is needed for when someone removes a GPU.

One possible work around is an CUDA app from Raistmer. I dont think It will delete them if its in an app info file.

~Bob

Tried Installing the Raistmer CUDA app and it deleted all 60+ Work units, and didn't report them as error or anything.. wondering if that is why the server won't give me any more work units, gotta wait til mid May for those other ones to timeout on the server maybe?

---

www.GreenClimate.net

What reason does it give for not getting more work?

I guess its time to search out an alternative to remote desktop... I did see one posted around here somewhere...

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

What reason does it give for not getting more work?

I have about 18 hours of work left for astropulse_v5 5.03 but it seems to be picking up when they hit about 80% they are finishing in about 20h vs the 200h it estimates.

4/23/2009 1:44:19 PM Starting BOINC client version 6.6.20 for windows_x86_64
4/23/2009 1:44:19 PM log flags: task, file_xfer, sched_ops
4/23/2009 1:44:19 PM Libraries: libcurl/7.19.4 OpenSSL/0.9.8j zlib/1.2.3
4/23/2009 1:44:19 PM Data directory: C:\ProgramData\BOINC
4/23/2009 1:44:19 PM Running under account Rob
4/23/2009 1:44:19 PM SETI@home Found app_info.xml; using anonymous platform
4/23/2009 1:44:19 PM Processor: 8 GenuineIntel Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz [Intel64 Family 6 Model 26 Stepping 4]
4/23/2009 1:44:19 PM Processor features: fpu tsc pae nx sse sse2 pni
4/23/2009 1:44:19 PM OS: Microsoft Windows Vista: Ultimate x64 Editon, Service Pack 1, (06.00.6001.00)
4/23/2009 1:44:19 PM Memory: 5.99 GB physical, 12.15 GB virtual
4/23/2009 1:44:19 PM Disk: 931.51 GB total, 606.61 GB free
4/23/2009 1:44:19 PM Local time is UTC -4 hours
4/23/2009 1:44:19 PM CUDA devices: GeForce GTX 260 (driver version 18250, CUDA version 1.3, 896MB, est. 85GFLOPS), GeForce GTX 260 (driver version 18250, CUDA version 1.3, 896MB, est. 86GFLOPS)
4/23/2009 1:44:19 PM Not using a proxy
4/23/2009 1:44:19 PM SETI@home URL: http://setiathome.berkeley.edu/; Computer ID: 4875506; location: home; project prefs: default
4/23/2009 1:44:19 PM No general preferences found - using BOINC defaults
4/23/2009 1:44:19 PM Reading preferences override file
4/23/2009 1:44:19 PM Preferences limit memory usage when active to 3067.09MB
4/23/2009 1:44:19 PM Preferences limit memory usage when idle to 5520.76MB
4/23/2009 1:44:19 PM Preferences limit disk usage to 10.00GB


<skip a bunch, and force a update>

4/23/2009 10:39:35 PM SETI@home Resuming task ap_01mr09ab_B3_P0_00348_20090422_18395.wu_1 using astropulse_v5 version 503
4/23/2009 10:39:40 PM SETI@home Sending scheduler request: To fetch work.
4/23/2009 10:39:40 PM SETI@home Reporting 1 completed tasks, requesting new tasks
4/23/2009 10:39:45 PM SETI@home Scheduler request completed: got 0 new tasks
4/23/2009 10:39:45 PM SETI@home Message from server: No work sent
4/23/2009 10:54:54 PM SETI@home Computation for task ap_04fe09ac_B4_P1_00187_20090422_03708.wu_0 finished
4/23/2009 10:54:56 PM SETI@home Started upload of ap_04fe09ac_B4_P1_00187_20090422_03708.wu_0_0
4/23/2009 10:54:59 PM SETI@home Finished upload of ap_04fe09ac_B4_P1_00187_20090422_03708.wu_0_0
4/24/2009 1:21:09 AM SETI@home Sending scheduler request: Requested by user.
4/24/2009 1:21:09 AM SETI@home Reporting 1 completed tasks, requesting new tasks
4/24/2009 1:21:14 AM SETI@home Scheduler request completed: got 0 new tasks
4/24/2009 1:21:14 AM SETI@home Message from server: No work sent

I have about 18 hours of work left for astropulse_v5 5.03 but it seems to be picking up when they hit about 80% they are finishing in about 20h vs the 200h it estimates.

That could be the reason you aren't getting new work. It thinks you have way too much already. Let it complete a couple of APs so that it will see the right time and it should start asking for more. It takes awhile for the DCF to settle.

---

PROUD MEMBER OF Team Starfire World BOINC

I have about 18 hours of work left for astropulse_v5 5.03 but it seems to be picking up when they hit about 80% they are finishing in about 20h vs the 200h it estimates.

That could be the reason you aren't getting new work. It thinks you have way too much already. Let it complete a couple of APs so that it will see the right time and it should start asking for more. It takes awhile for the DCF to settle.

Yea, but I have 0 work for CUDA, and I have a total of 8 AP jobs, this machine has 8 cores, 6 are running AP, 2 are running Climate, it picks up the AP's when it is about 80% done with a active AP.

You seem to have a lot of MB work you picked up on the 20th/21st. Are these showing as 6.08 or 6.03? Are they showing up at all in your Boinc Manager?

---

PROUD MEMBER OF Team Starfire World BOINC

You seem to have a lot of MB work you picked up on the 20th/21st. Are these showing as 6.08 or 6.03? Are they showing up at all in your Boinc Manager?

OK, back to the start when I entered this thread. Installed Raistmer CUDA Processor on the 21st... BOINC Deleted all the MB Work, SETI Server thinks I have 60+ MB jobs, I have ZERO Jobs on the actual Computer. Not getting any more MB jobs after fixing Raistmer application.

Suspect that either it is still installed wrong even though it shows 2 CUDA devices there, or SETI Server isn't giving me anymore work til this other 60 MB jobs expire sometime mid May.

Which answers the original reply, If you Install Raistmer Application, BOINC still has the potential of deleting all your MB Jobs.

All right, I'm up with you now. Didn't understand what you were looking at.

I probably would set no new tasks, run down the APs that are left, remove your app_info file and the other files you will need and do a detach/reattach to the project. Also set no network activity. Check your app_info for mistakes, put everything back in,set your cache for no more than two days,make sure Boinc Manager is running correctly, resume network activity and allow new tasks. This will get you started all over again and will clear all those WUs seti thinks you still have.

Make sure you are putting the app_info and everything in the right folder. It goes in the program data/ boinc/projects/setiathome folder.

---

PROUD MEMBER OF Team Starfire World BOINC

The app info is in the right spot.. boinc has detected it.
I'd say that you just had some bad timing getting tasks as the feeder ran dry for MB just as you requested.
It wants work it just isn't getting any.

---

Do you Good Search for Seti@Home? http://www.goodsearch.com/?charityid=888957
Or Good Shop? http://www.goodshop.com/?charityid=888957

Dunno.. has been since the 21st of April since I got a MB Job. It looks for work every hour for 4 days now.. I should be done with these jobs in 16 hours, my newest job is elapsed 03:21:22 at 14.701%

Amazing how the Raistmer Code took my AP work from 100 hours to 20 hours, and they are verifying good.. 1260 x 8 = 10,080 credits every 20 hours

After about 50 restores.. I think I finally got it. Took stopping network communications aborting about 40 Work Units.. and the most important was the app_info from another thread http://setiathome.berkeley.edu/forum_thread.php?id=52589&nowrap=true#888092

the stock APP_INFO does not work from Raistmer it is missing important information letting the seti server know you can process type 603 and 608 work units.

other thing that is nice to do at least, is to abort and upload all your work units before trying this. so you don't have ghost work units.