Posts by Bernd Schnitker
log in
1) Message boards : News : New MAC OS X OpenCL Version (Message 14392)
Posted 2020 days ago by Bernd Schnitker
Sorry I would prefer to report better news, I appreciate the big work you do and I can imagine how disappointing it may be to get only complaining.
I still have problems with the app. I'm not sure if I could do anything.
two issues:

1.) If the app is paused, most time the result is not valid.
(see: [url]http://boinc.thesonntags.com/collatz/result.php?resultid=118760490
[/url] for an example)

2.) I get messages to update the driver:
Collatz Conjecture: Motice from server
ATI GPU: auf neuesten Treiber aktualisieren, um die GPU diese Computers nutzen zu k├Ânnen.


How can I upgrade the driver?, - AFAIK Mac drive is coming from OS
What can I do.

Because the Mac stalls if GPU is running everytime I changed the setting to pause gpu if computer is in use
- which results in more invalid results, even if calculating long lasting collatz (see 1)

Any hints what I can do or where I can find some instrcution, are welcome as I' have insufficient knowledge about BOINC and GPU.




I know that for NVIDIA GPUs you can get newer drivers on the dev website even for OSX. I have an older Mac mini (2010) and I can manually get newer drivers for my integrated 320M from the Nvidia website. I would look around the ATI/AMD website for a dev forum for Open CL.
2) Message boards : Number crunching : MAC OS X 32-bit CUDA App (Message 13868)
Posted 2092 days ago by Bernd Schnitker
I have started to see a bug I thought was fixed. I am having collatz GPU apps that die silently. I see the wall clock continueing to run but GPU has stopped doing any work. When the WU is suspended the following message is shown in BOINC.
Thu Apr 26 19:04:39 2012 | Collatz Conjecture | Task collatz_2375033642863672666472_824633720832_1: no shared memory segment
Thu Apr 26 19:04:39 2012 | Collatz Conjecture | Task collatz_2375033642863672666472_824633720832_1 exited with zero status but no 'finished' file
Thu Apr 26 19:04:39 2012 | Collatz Conjecture | If this happens repeatedly you may need to reset the project.

I'm not sure why this is happening. If I don't catch it the WU runs until the max elapsed time safety fires and kills the WU.
3) Message boards : News : Server Back Online (Message 13757)
Posted 2107 days ago by Bernd Schnitker
I posted to the BOINCStats and also the SETI.USA team's message board on Saturday to let people know it would take a couple days to get everything back up and running. I should have notified Bok so he could post something on the FreeDC stats site as well.

Maybe also look into posting on the BOINC project forums in the projects outage thread. http://boinc.berkeley.edu/dev/forum_forum.php?id=11 the thread is News on Project Outages.
4) Message boards : Number crunching : Mac OSX no longer allowed to run Collatz CUDA? (Message 13216)
Posted 2206 days ago by Bernd Schnitker
Working now. According to a PM slicker fixed it. Config error
5) Message boards : Number crunching : Mac OSX no longer allowed to run Collatz CUDA? (Message 13206)
Posted 2208 days ago by Bernd Schnitker
Boinc start up message
Mon Dec 26 15:23:32 2011 | | Starting BOINC client version 6.12.33 for x86_64-apple-darwin
Mon Dec 26 15:23:32 2011 | | log flags: file_xfer, sched_ops, task
Mon Dec 26 15:23:32 2011 | | Libraries: libcurl/7.19.7 OpenSSL/0.9.7l zlib/1.2.3 c-ares/1.6.0
Mon Dec 26 15:23:32 2011 | | Data directory: /Library/Application Support/BOINC Data
Mon Dec 26 15:23:32 2011 | | Processor: 2 GenuineIntel Intel(R) Core(TM)2 Duo CPU P8600 @ 2.40GHz [x86 Family 6 Model 23 Stepping 10]
Mon Dec 26 15:23:32 2011 | | Processor features: FPU VME DE PSE TSC MSR PAE MCE CX8 APIC SEP MTRR PGE MCA CMOV PAT PSE36 CLFSH DS ACPI MMX FXSR SSE SSE2 SS HTT TM PBE SSE3 DTES64 MON DSCPL VMX SMX EST TM2 SSSE3 CX16 TPR PDCM SSE4.1 XSAVE
Mon Dec 26 15:23:32 2011 | | OS: Mac OS X 10.6.8 (Darwin 10.8.0)
Mon Dec 26 15:23:32 2011 | | Memory: 8.00 GB physical, 231.85 GB virtual
Mon Dec 26 15:23:32 2011 | | Disk: 297.77 GB total, 231.61 GB free
Mon Dec 26 15:23:32 2011 | | Local time is UTC -5 hours
Mon Dec 26 15:23:32 2011 | | NVIDIA GPU 0: GeForce 320M (driver version unknown, CUDA version 3020, compute capability 1.2, 253MB, 91 GFLOPS peak)
Mon Dec 26 15:23:32 2011 | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 31149; resource share 100
Mon Dec 26 15:23:32 2011 | | General prefs: using separate prefs for home
Mon Dec 26 15:23:32 2011 | | Reading preferences override file
Mon Dec 26 15:23:32 2011 | | Preferences:
Mon Dec 26 15:23:32 2011 | | max memory usage when active: 5734.40MB
Mon Dec 26 15:23:32 2011 | | max memory usage when idle: 7864.32MB
Mon Dec 26 15:23:32 2011 | | max disk usage: 7.00GB
Mon Dec 26 15:23:32 2011 | | max CPUs used: 1
Mon Dec 26 15:23:32 2011 | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
Mon Dec 26 15:23:32 2011 | | Not using a proxy

Again hope this helps
6) Message boards : Number crunching : Mac OSX no longer allowed to run Collatz CUDA? (Message 13205)
Posted 2208 days ago by Bernd Schnitker
Just tried to do Mini-Collatz WUs on GPU and am also getting the same error message.
Fri Dec 30 21:59:20 2011 | Collatz Conjecture | Task collatz_2373217925579359889768_824633720832_0: no shared memory segment
Fri Dec 30 21:59:20 2011 | Collatz Conjecture | Task collatz_2373217925579359889768_824633720832_0 exited with zero status but no 'finished' file
Fri Dec 30 21:59:20 2011 | Collatz Conjecture | If this happens repeatedly you may need to reset the project.
Fri Dec 30 22:45:59 2011 | Collatz Conjecture | Restarting task collatz_2373217925579359889768_824633720832_0 using collatz version 202
Fri Dec 30 23:05:53 2011 | Collatz Conjecture | Computation for task collatz_2373217925579359889768_824633720832_0 finished
Fri Dec 30 23:05:56 2011 | Collatz Conjecture | Started upload of collatz_2373217925579359889768_824633720832_0_0
Fri Dec 30 23:05:57 2011 | Collatz Conjecture | Finished upload of collatz_2373217925579359889768_824633720832_0_0
Sat Dec 31 13:27:43 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 13:27:43 2011 | Collatz Conjecture | Reporting 1 completed tasks, requesting new tasks for NVIDIA GPU
Sat Dec 31 13:27:46 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 13:38:53 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 13:38:53 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 13:38:54 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 13:59:35 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 13:59:35 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 13:59:38 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 14:45:26 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 14:45:26 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 14:45:27 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 16:20:46 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 16:20:46 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 16:20:47 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 16:20:47 2011 | Collatz Conjecture | No work sent
Sat Dec 31 16:20:47 2011 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sat Dec 31 16:20:47 2011 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sat Dec 31 19:04:48 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 19:04:48 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 19:04:50 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 19:04:50 2011 | Collatz Conjecture | No work sent
Sat Dec 31 19:04:50 2011 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sat Dec 31 19:04:50 2011 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sat Dec 31 23:18:42 2011 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sat Dec 31 23:18:42 2011 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sat Dec 31 23:18:44 2011 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sat Dec 31 23:18:44 2011 | Collatz Conjecture | No work sent
Sat Dec 31 23:18:44 2011 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sat Dec 31 23:18:44 2011 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 03:14:38 2012 | | Project communication failed: attempting access to reference site
Sun Jan 1 03:14:39 2012 | | Internet access OK - project servers may be temporarily down.
Sun Jan 1 08:41:48 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 08:41:48 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 08:41:49 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 08:41:49 2012 | Collatz Conjecture | No work sent
Sun Jan 1 08:41:49 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 08:41:49 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 10:05:55 2012 | Collatz Conjecture | update requested by user
Sun Jan 1 10:05:59 2012 | Collatz Conjecture | Sending scheduler request: Requested by user.
Sun Jan 1 10:05:59 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 10:06:00 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 10:06:00 2012 | Collatz Conjecture | No work sent
Sun Jan 1 10:06:00 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 10:06:00 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 10:20:45 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 10:20:45 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 10:20:47 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 10:20:47 2012 | Collatz Conjecture | No work sent
Sun Jan 1 10:20:47 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 10:20:47 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 10:49:33 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 10:49:33 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 10:49:34 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 10:49:34 2012 | Collatz Conjecture | No work sent
Sun Jan 1 10:49:34 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 10:49:34 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 11:29:09 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 11:29:09 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 11:29:10 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 11:29:10 2012 | Collatz Conjecture | No work sent
Sun Jan 1 11:29:10 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 11:29:10 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 12:34:10 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 12:34:10 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 12:34:11 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 12:34:11 2012 | Collatz Conjecture | No work sent
Sun Jan 1 12:34:11 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 12:34:11 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 15:16:37 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 15:16:37 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 15:16:38 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 15:16:38 2012 | Collatz Conjecture | No work sent
Sun Jan 1 15:16:38 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 15:16:38 2012 | Collatz Conjecture | Message from server: No work available for the applications you have selected. Please check your project preferences on the web site.
Sun Jan 1 19:44:44 2012 | Collatz Conjecture | update requested by user
Sun Jan 1 19:44:49 2012 | Collatz Conjecture | Sending scheduler request: Requested by user.
Sun Jan 1 19:44:49 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 19:44:51 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 19:44:51 2012 | Collatz Conjecture | No work sent
Sun Jan 1 19:44:51 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.
Sun Jan 1 19:57:11 2012 | Collatz Conjecture | Sending scheduler request: To fetch work.
Sun Jan 1 19:57:11 2012 | Collatz Conjecture | Requesting new tasks for NVIDIA GPU
Sun Jan 1 19:57:12 2012 | Collatz Conjecture | Scheduler request completed: got 0 new tasks
Sun Jan 1 19:57:12 2012 | Collatz Conjecture | No work sent
Sun Jan 1 19:57:12 2012 | Collatz Conjecture | Message from server: collatz is not available for your type of computer.

This is all the Collatz messages since the last successful WU was done. Hope this helps
7) Message boards : Number crunching : Mac OSX no longer allowed to run Collatz CUDA? (Message 13203)
Posted 2208 days ago by Bernd Schnitker
I am now getting server errors saying I can no longer crunch Collatz longer work units. Yesterday their was a science app to allow it and today the app seems to be missing. Is this on purpose or just an accident? The project says I may crunch Mini Collatz on my GPU but not Collatz what is up with that since the app is the same but the only difference is in WU length?
8) Message boards : Number crunching : OpenCL Apps (Message 12330)
Posted 2441 days ago by Bernd Schnitker
I have a bit of curiosity with Open CL on OSX. When 10.6.x came out Open CL was supposed to be supported on Nvidia and ATI/AMD GPUs. Supposedly the same code could run on both has this come to pass or even come close? Does Lion( 10.7.x ) look better on this front? Have you looked at the Dev builds of Lion apple has released?

I have an Nvidia GPU and use it here with OSX. I have a slow memory leak on the GPU and need to reboot/ sleep wake it ever 2-3 days or it runs out of memory and trashes work every 30 seconds or so.
9) Message boards : Number crunching : MAC OS X 32-bit CUDA App (Message 11887)
Posted 2485 days ago by Bernd Schnitker
I have Collatz working on a 320M GPU with the 3.2.17 drivers. I just need to pay attention to WU after running some games and occasionally after running flash games/movies. Sometimes it looks like memory is not cleared after the program quits. Sleeping the system after the WU trashing or right after quitting the game before resuming GPU work seems to fix this. Collatz is hit more often with this bug than is PrimeGrid. I have even been able to run one of the PrimeGrid sub projects at the same time as playing Civ IV. Collatz gave an immediate out of memory error and trashes the WU.
10) Message boards : Number crunching : New workunit size/type for Collatz (Message 10982)
Posted 2561 days ago by Bernd Schnitker
My Mac Mini 320M went from 3h 20 min to around 13hours 30 min. I am noticing Video lag for the first time in a long time. Will wait and see if that keeps up.
11) Message boards : News : Planned Outage (Message 10821)
Posted 2567 days ago by Bernd Schnitker
All I am going to do is wish you good luck and may you remove no more skin from your knuckles than the IT gods demand. The case requires a blood sacrifice every time it is opened.
May it happen as planed.
12) Message boards : Macintosh : Collatz goes into an endless error loop after playing Civ IV (Message 10593)
Posted 2597 days ago by Bernd Schnitker
I have been noticing a that the Collatz science app for CUDA does not recover from being suspended to allow a Game of CIV of any type to be played. I noticed it erred if run at the same time as the game. So I started to suspend the app before starting a game. Now I quite the game and resume the Boinc Collatz science app 50% of the time it errors out right away. It then downloads a new WU an it errors out right away also. This can continue indefinitely as I found out when I went to bed right away after a game. Had 6 or so dead work units in the morning. The only cure I have found is to sleep the mac and then it seems to be fixed or restart the computer.

My question is their any other trick to fix this problem? Restarting is such a pain in the *SS.
Restarting is the only 100% sure fix sleeping does it most of the time.

link to a WU
http://boinc.thesonntags.com/collatz/result.php?resultid=67462791

Edit clarification and some spelling
more info Prime grid CUDA app does not have this problem 100% able to work. In fact often Collatz errors twice and then Prime grid downloads one and it runs clean. Then Collatz tries again and errors again and prime grid grabs one and it runs fine. This cycle is what happened when I had the overnight oopsie and the 6 dead WU.
My mac is a 2010 mac mini 8GB memory OSX 10.6.5 Nvidia 320M chipset/GPU 256 MB video ram.


The problem is that your GPU has not enough RAM. Just take a look at the error message:

cudaSafeCall() Runtime API error in file <collatz.cu>, line 494 : out of memory.


You need to uncheck "Leave application in memory while suspended" in your boinc managers settings.

I have that option already unchecked. I think it maybe the Civ Game leaving the GPU in an unusual state after use. I do see vRam being freed up after the game is shutdown or as it ends most of the time the Aspyr port of Civ IV crashes even though it has been out for years with many bug fixes. Aspyr is known to be crashy. I need to check the ram usage before and after more carefully. Free ram runs from 160MB fresh running CUDA to under 60MB after/during the Game. After the game Quit/ crashes for several minutes it sit at 60 free and goes back up after sleep wake or restart.

Edit more info vRam reading from the ATI NVIDIA Tool 0.93 for OSX
13) Message boards : Number crunching : MAC OS X 32-bit CUDA App (Message 10590)
Posted 2597 days ago by Bernd Schnitker
I have Collatz running with CUDA 3.2
Mac reports CUDA driver version 3.2.17
GPU Driver Version 1.6.24.17 (256.00.15f04)
says I am all up to date for Mac Mini 2010. OSX 10.6.5 8GB Memory GPU 320M 256 MB
Have problems resuming work after suspending to play games. ( Civ IV) Sleeping/restarting fixes the problem.
14) Message boards : Macintosh : Collatz goes into an endless error loop after playing Civ IV (Message 10589)
Posted 2597 days ago by Bernd Schnitker
I have been noticing a that the Collatz science app for CUDA does not recover from being suspended to allow a Game of CIV of any type to be played. I noticed it erred if run at the same time as the game. So I started to suspend the app before starting a game. Now I quite the game and resume the Boinc Collatz science app 50% of the time it errors out right away. It then downloads a new WU an it errors out right away also. This can continue indefinitely as I found out when I went to bed right away after a game. Had 6 or so dead work units in the morning. The only cure I have found is to sleep the mac and then it seems to be fixed or restart the computer.

My question is their any other trick to fix this problem? Restarting is such a pain in the *SS.
Restarting is the only 100% sure fix sleeping does it most of the time.

link to a WU
http://boinc.thesonntags.com/collatz/result.php?resultid=67462791

Edit clarification and some spelling
more info Prime grid CUDA app does not have this problem 100% able to work. In fact often Collatz errors twice and then Prime grid downloads one and it runs clean. Then Collatz tries again and errors again and prime grid grabs one and it runs fine. This cycle is what happened when I had the overnight oopsie and the 6 dead WU.
My mac is a 2010 mac mini 8GB memory OSX 10.6.5 Nvidia 320M chipset/GPU 256 MB video ram.
15) Message boards : Number crunching : MAC OS X 32-bit CUDA App (Message 10270)
Posted 2624 days ago by Bernd Schnitker
I updated OSX to 10.6.5 and my CUDA App is still working. I have the CUDA control panel complaining that the driver needs updating but no new driver is available. So I have not updated anything and it is still working.




Main page · Your account · Message boards


Copyright © 2018 Jon Sonntag; All rights reserved.