Juha - that sounds like a plan.

It will be a bit before I can risk my main machine.

This is purely a 64-bit problem though. I peeled the extra RAM out of the Quad Core. Wiped it and installed 32-bit Mint17 KDE (main machine has 64-bit Mint17 KDE). Applied all updates. Installed the 331 driver. Rebooted. Installed BOINC. Added my projects. The NVIDIA Server info thing shows my CUDAs are now being used and the BOINC startup log shows the card being found.

Gandalf, if you can live with a pathetically small 4Gig of RAM, the "quick" or rather "safe" fix is to back down to a 32-bit version of Mint 17. The 64-bit stuff was obviously not tested, at least with respect to BOINC and CUDA.

---

Thanks Roland, but I was hoping to augment my CPU WUs with the GPU WUs.
By limiting my RAM, would that not affect the CPU WUs already running?

The short answer would be no.

There is some formula which you tweak with your settings about % of CPUs and % of memory and the checkbox for "leave WU in memory...", not to mention those checkboxes for "while computer is in use" and "use GPU while computer is in use", etc. That has more to do with your total units.

My Quad Core when reduced to 32-bit OS and only 4G of RAM limited to 90% of CPU time is still running only 4 WU. It was running the same under 64-bit with ^G of RAM.

The Quad Core, however, is only running BOINC to keep the parts alive. It ceased being my main machine a long time ago.

I do software development for clients and author books, both geek books: [url]http://theminimumyouneedtoknow.com/ [/url] and some general interest fiction: http://infiniteexposure.net/ http://johnsmith-book.com/

At any point in the day I could have dozens of document files open or a build running which is trying to chew through (many) hundreds of source files using a high -J option. If I dont want to watch everything come to a crawl while swap gets beaten into submission, I need to provide enough breathing room.

Now, if all you do is answer email and occasionally surf the Web, you should have few issues. If you stretch the definition of "occasionally surf the Web" to include "watch streaming high def video" then...

---

I don't believe I'm going to try the probe option on the Six Core. Before I started messing with this I only suffered the "NVIDIA Crash" about once per week.

For those who don't know the "NVIDIA Crash" is where the mouse will move around on the screen, but take no action. You can SSH into the box from another computer and do stuff as long as it doesn't need any part of the video card.

I have had that crash at least 4 times today. I have tried many of the drivers from the edgers PPA which claim to fix the problem. At this point I have rolled the machine back to the default neavu (sp?) driver and did a command line purge of all things NVIDIA after purging the edgers PPA from this system.

BOINC will have to live without CUDA on this machine because the 64-bit stuff simply hasn't been tested. The 32-bit stuff seems to work great, but I cannot roll back to a 32-bit OS on this machine. On the BOINC only machines, who cares, but on my primary machine it is simply not an option.

Perhaps, 4 years from now, when the next LTS comes out, BOINC and CUDA will work correctly in the 64-bit version????

---

Roland - what happens, if I replace the 64-bit LM with a 32-bit LM. Will the PC function the same with the CPU WUs plus allow the GPU WUs to work?

Thanks

Gandalf, if you are considering switching to 32-bit Mint you could just as well try the nvidia-modprobe. Even if it turns out it doesn't work and you need to reinstall you aren't losing anything.

Gandalf,

Just be certain to backup or copy off to another disk anything you do want to save, such as Documents, music, etc. _before_ you try the modprobe.

FYI, my Six Core is busily humming away with KDE BOINC 64, one text editor and this browser. I opened up the monitor because a few things seemed sluggish for a second. (Turns out BOINC was pulling down a pile of WU from Poem.) Only 1.8 Gig of my RAM was being used. I have BOINC on this machine configured with the default of 0 for % of processors but 80% for amount of CPU.

64-bit implementations of Ubuntu (and YAU distros) tend to be the red headed step child chained in the basement. The only real testing appears to happen on the 32-bit stuff.

---

Gandalf,

After looking at my schedule today I decided it was going to be a bust trying to get any real work done. I backed up my box while riding my exercise bike after installing the 331 drivers from the repo and rebooting. Manually added the repo in software sources.

sources->add repository then paste in the line which was so graciously provided earlier in this thread.

BOINC is finding the CUDA cores on the 16 core card I stuck in this box.

I'm going to try getting this card to work:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814121778
then I can get around to real work.


Note: The info for nvidia-modprobe will freak you at first. The current version in the backports is 340 something. It does appear to work with the 331 driver.


I will report back tomorrow if I suffer any NVIDIA Crashes. You may wish to wait until then, or, do a full image bare-metal-restore backup and try it yourself.

---

Roland - I'm not a Linux Guru like you guys. I currently have an i7 PC running LM - 64 bits. After all the currently downloaded WUs have reported, I will replace the LM 64-bit with the 32-bit LM.

While I have tried to follow your above information, most of it was greekish to me. :-) If I try to extract the proper changes from the above instructions, I could mess something up. (could?) More like will.

I'd appreciate it, if you could do a brain dead step by step guide, to implement the needed changes to take a new LM 32-bit install with the 32-bit Boinc program installed, to the desired configuration to get the CPU and the GPUs to run WUs. Yea, when it comes to Linux I'm really that dumb. I'll understand, if you don't have the time or if you can't get to it right now. I'll be ready, when/if you are.

Thanks again.
Gandalf

PS - even magic doesn't work with the 64-bit version.

I'm not Linux Guru. If this was OpenVMS I would gladly delve into the depths, but it is not.


Your 32-bit instructions:

After installing Mint 17 KDE:

Click on the little blue start button in the lower left corner.

In the little search bar at the top type "drive" and you should see an option for "Driver Manager" appear. Click it.

Enter your password when prompted.

Go get a cold can of Diet Mt. Dew. When you get back the window you really want to see might have appeared. There is no "busy" indicator or anything, it just smokes half a pack of cigs before it gets around to showing you anything.

Select nvidia-331 DO NOT SELECT the -updates version.

Click on apply changes.

Finish drinking your Mt. Dew while this works.

When it completes click the x to close the window then click "ok" on the main window.

Reboot

Open your BOINC manager.

Click Advanced->Event Log

Scroll to top of log (it always shows last entry when it opens) You should see something like:

Tue 10 Mar 2015 06:35:09 AM CDT | | CUDA: NVIDIA GPU 0: GeForce GT 630 (driver version unknown, CUDA version 6.0, compute capability 3.5, 2047MB, 1973MB available, 692 GFLOPS peak)

near the beginning.


++++
Want to know the only steps different to make your current 64-bit system work?

After rebooting so nvidia-331 takes full effect.

Click on the little blue start button in the lower left corner.

In the little search bar at the top type "sour" and you should see an option for "Software Sources" appear. Click it.

Click the "Additional repositories" button on left.

Click the "Add a new repository" button at bottom.

paste in the following:

deb http://archive.ubuntu.com/ubuntu trusty-backports main restricted universe multiverse

Click "Ok"

Click the "update" or "refresh" button on upper right side of screen. (This will be greyed out with "No action required" until you actually add something and I don't remember what the text chagnes to.)

Exit out of all windows.

Click on the little blue start button in the lower left corner.

Applications->System->Synaptic Package Manager

Enter password when prompted.

After the index finishes updating type "modpro" in the "Quick filter" field

Mark nvidia-modprobe for installation, then click "Apply changes"

Exit everything and reboot

Open your BOINC manager.

Click Advanced->Event Log

Scroll to top of log (it always shows last entry when it opens) You should see something like:

Tue 10 Mar 2015 06:35:09 AM CDT | | CUDA: NVIDIA GPU 0: GeForce GT 630 (driver version unknown, CUDA version 6.0, compute capability 3.5, 2047MB, 1973MB available, 692 GFLOPS peak)

near the beginning.

---

All was well today. No "Nvidia crash" to report. I did have to tweak my processor usage settings:

x While computer is in use
x Use GPU while computer is in use

Switch between applications every 240 minutes
on Multiprocessor systems, use at most 0% of the processors
Use at most 80% CPU time


If you put anything other than 0% for processors, say 70 or 80, blithely thinking, oh, that leaves N CPUs for "me", you get a system which is almost impossible to use. Typing in a text editor or word processor remains 1-3 characters behind AND NEVER CATCHES UP.

With these settings I barely know BOINC is running. Applications are a little slow to initially paint and start menu is a tiny bit sluggish, probably due to fact I'm running at 93% GPU utilization, but, I really put that card in here so Einstein, Cosmology, GPUGRID, and WCG would have some place to play.

If it wasn't for BOINC I wouldn't know anything about CUDA.

---

I've redone my number two Linux PC to LM 17.1 32-bits Cinnamon.
The two GPUs are GeForce GTX285

Updated the system.
With Synaptic Package Manager, installed...
Boinc
Boinc-app-milkway
Boinc-nvidia-cuda

Driver Manager says...
xserver-xorg-video-nouveau

GPUs are not seen by Projects.

Tried...

sudo service boinc-client restart

No joy.
Event Log says...

Thu 12 Mar 2015 05:28:28 AM PDT | | Starting BOINC client version 7.2.42 for i686-pc-linux-gnu
Thu 12 Mar 2015 05:28:28 AM PDT | | log flags: file_xfer, sched_ops, task, sched_op_debug
Thu 12 Mar 2015 05:28:28 AM PDT | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Thu 12 Mar 2015 05:28:28 AM PDT | | Data directory: /var/lib/boinc-client
Thu 12 Mar 2015 05:28:28 AM PDT | | No usable GPUs found

What is the next step, please.

Some times the package manager provides too much information.

I install only Boinc. It then installs boinc-manager, boinc-client, and libboinc7.

I add my projects from within boinc, I do not install anything else which has "boinc" in its name. I would humbly suggest you un-install those other things, especially that boinc-nvidia-cuda. I don't have it and I'm using cuda just fine.

You definitely skipped the instruction steps surrounding installation of nvida-331 driver. the nouveau driver does not provide any cuda support that I'm aware of.

---

Roland - Thanks for the guidance. This is why I am doing this step by step.
I have un-installed the boinc-nvidia-cuda and the boinc-app-milkway and rebooted.

Do I now click on the Driver Manager bullet button for nvidia-331?

Or, do I need to do this commend?

sudo apt-get install nvidia-modprobe

Thanks

Click on the button in the driver manager, then click on Apply or whatever it is. This will take some time. If you want to be entertained I "think" you can expand a "show details" type window.

If you are running 32-bit there is no need for modprobe. The 32-bit stuff was tested.

You need modprobe with 64-bit because the 64-bit stuff simply wasn't tested.

---

Roland - OK. The Projects now see the GPUs, but are not downloading any WUs.
Any idea what is wrong? Collatz is downloading to my Win7 crunchers.

If you rebooted after installing 331 you should _eventually_ get work units. I have given up trying to figure out which bag of dried chicken bones they are shaking and reading when it comes to assigning work units.

IF you don't have any compute restrictions AND you don't have any network restrictions, you should get work units.

Click on BOINC manager->tools->computing preferences (assuming you are already in the advanced view" and verify settings for processor usage, network usage, etc.

I have seen projects stack work on my 32-bit machine yet leave my Six Core idle. There is some method to the madness, but I know not what it is.

OH!!!!! When looking at Network Usage be certain you have a min work buffer of 1 day and max of at least 4. Sometimes you don't get WU because the guestimated time puts them beyond your max buffer.

Add/subscribe to at least 4 different projects and let things sit overnight. If everything is set correctly you will find at least one of them was not too proud to use your machine come morning.

---

Roland - Here is the Event Log. Maybe you can see something I missed.
I removed the WCG lines since WCG is working 5x5 ...

Thu 12 Mar 2015 10:00:26 PM PDT | | Starting BOINC client version 7.2.42 for i686-pc-linux-gnu
Thu 12 Mar 2015 10:00:26 PM PDT | | log flags: file_xfer, sched_ops, task, sched_op_debug
Thu 12 Mar 2015 10:00:26 PM PDT | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Thu 12 Mar 2015 10:00:26 PM PDT | | Data directory: /var/lib/boinc-client
Thu 12 Mar 2015 10:00:26 PM PDT | | CUDA: NVIDIA GPU 0: GeForce GTX 285 (driver version unknown, CUDA version 6.0, compute capability 1.3, 2047MB, 1864MB available, 1063 GFLOPS peak)
Thu 12 Mar 2015 10:00:26 PM PDT | | CUDA: NVIDIA GPU 1: GeForce GTX 285 (driver version unknown, CUDA version 6.0, compute capability 1.3, 1024MB, 969MB available, 1063 GFLOPS peak)
Thu 12 Mar 2015 10:00:26 PM PDT | | Host name: WhiteCastle
Thu 12 Mar 2015 10:00:26 PM PDT | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz [Family 6 Model 45 Stepping 7]
Thu 12 Mar 2015 10:00:26 PM PDT | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept vpid
Thu 12 Mar 2015 10:00:26 PM PDT | | OS: Linux: 3.13.0-37-generic
Thu 12 Mar 2015 10:00:26 PM PDT | | Memory: 11.84 GB physical, 11.97 GB virtual
Thu 12 Mar 2015 10:00:26 PM PDT | | Disk: 222.82 GB total, 205.48 GB free
Thu 12 Mar 2015 10:00:26 PM PDT | | Local time is UTC -7 hours
Thu 12 Mar 2015 10:00:26 PM PDT | | Config: use all coprocessors
Thu 12 Mar 2015 10:00:26 PM PDT | | Config: GUI RPCs allowed from:
Thu 12 Mar 2015 10:00:26 PM PDT | | 192.168.1.100
Thu 12 Mar 2015 10:00:26 PM PDT | | 192.168.1.101
Thu 12 Mar 2015 10:00:26 PM PDT | | 192.168.1.102
Thu 12 Mar 2015 10:00:26 PM PDT | | 192.168.1.103
Thu 12 Mar 2015 10:00:26 PM PDT | World Community Grid | URL http://www.worldcommunitygrid.org/; Computer ID 3264072; resource share 100
Thu 12 Mar 2015 10:00:26 PM PDT | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 165304; resource share 100
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 610823; resource share 100
Thu 12 Mar 2015 10:00:26 PM PDT | Collatz Conjecture | General prefs: from Collatz Conjecture (last modified 12-Mar-2015 04:09:38)
Thu 12 Mar 2015 10:00:26 PM PDT | Collatz Conjecture | Computer location: work
Thu 12 Mar 2015 10:00:26 PM PDT | | General prefs: using separate prefs for work
Thu 12 Mar 2015 10:00:26 PM PDT | | Reading preferences override file
Thu 12 Mar 2015 10:00:26 PM PDT | | Preferences:
Thu 12 Mar 2015 10:00:26 PM PDT | | max memory usage when active: 10909.13MB
Thu 12 Mar 2015 10:00:26 PM PDT | | max memory usage when idle: 10909.13MB
Thu 12 Mar 2015 10:00:26 PM PDT | | max disk usage: 100.00GB
Thu 12 Mar 2015 10:00:26 PM PDT | | (to change preferences, visit a project web site or select Preferences in the Manager)
Thu 12 Mar 2015 10:00:26 PM PDT | | Not using a proxy
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | [sched_op] Starting scheduler request
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | Sending scheduler request: To fetch work.
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | Requesting new tasks for NVIDIA
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
Thu 12 Mar 2015 10:00:26 PM PDT | Milkyway@Home | [sched_op] NVIDIA work request: 2419200.00 seconds; 2.00 devices
Thu 12 Mar 2015 10:00:27 PM PDT | Milkyway@Home | Scheduler request completed: got 0 new tasks
Thu 12 Mar 2015 10:00:27 PM PDT | Milkyway@Home | [sched_op] Server version 705
Thu 12 Mar 2015 10:00:27 PM PDT | Milkyway@Home | Project requested delay of 61 seconds
Thu 12 Mar 2015 10:00:27 PM PDT | Milkyway@Home | [sched_op] Deferring communication for 00:01:01
Thu 12 Mar 2015 10:00:27 PM PDT | Milkyway@Home | [sched_op] Reason: requested by project
Thu 12 Mar 2015 10:07:20 PM PDT | Milkyway@Home | update requested by user
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | sched RPC pending: Requested by user
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | [sched_op] Starting scheduler request
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | Sending scheduler request: Requested by user.
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | Requesting new tasks for NVIDIA
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
Thu 12 Mar 2015 10:07:24 PM PDT | Milkyway@Home | [sched_op] NVIDIA work request: 2419200.00 seconds; 2.00 devices
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | Scheduler request completed: got 0 new tasks
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | [sched_op] Server version 705
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | Project requested delay of 61 seconds
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | General prefs: from Milkyway@Home (last modified 12-Mar-2015 22:06:55)
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | Computer location: work
Thu 12 Mar 2015 10:07:26 PM PDT | | General prefs: using separate prefs for work
Thu 12 Mar 2015 10:07:26 PM PDT | | Reading preferences override file
Thu 12 Mar 2015 10:07:26 PM PDT | | Preferences:
Thu 12 Mar 2015 10:07:26 PM PDT | | max memory usage when active: 10909.13MB
Thu 12 Mar 2015 10:07:26 PM PDT | | max memory usage when idle: 10909.13MB
Thu 12 Mar 2015 10:07:26 PM PDT | | max disk usage: 100.00GB
Thu 12 Mar 2015 10:07:26 PM PDT | | (to change preferences, visit a project web site or select Preferences in the Manager)
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | [sched_op] Deferring communication for 00:01:01
Thu 12 Mar 2015 10:07:26 PM PDT | Milkyway@Home | [sched_op] Reason: requested by project
Thu 12 Mar 2015 10:08:31 PM PDT | Milkyway@Home | [sched_op] Starting scheduler request
Thu 12 Mar 2015 10:08:31 PM PDT | Milkyway@Home | Sending scheduler request: To fetch work.
Thu 12 Mar 2015 10:08:31 PM PDT | Milkyway@Home | Requesting new tasks for NVIDIA
Thu 12 Mar 2015 10:08:31 PM PDT | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
Thu 12 Mar 2015 10:08:31 PM PDT | Milkyway@Home | [sched_op] NVIDIA work request: 2419200.00 seconds; 2.00 devices
Thu 12 Mar 2015 10:08:33 PM PDT | Milkyway@Home | Scheduler request completed: got 0 new tasks
Thu 12 Mar 2015 10:08:33 PM PDT | Milkyway@Home | [sched_op] Server version 705
Thu 12 Mar 2015 10:08:33 PM PDT | Milkyway@Home | Project requested delay of 61 seconds
Thu 12 Mar 2015 10:08:33 PM PDT | Milkyway@Home | [sched_op] Deferring communication for 00:01:01
Thu 12 Mar 2015 10:08:33 PM PDT | Milkyway@Home | [sched_op] Reason: requested by project
Thu 12 Mar 2015 10:10:56 PM PDT | Milkyway@Home | update requested by user
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | sched RPC pending: Requested by user
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | [sched_op] Starting scheduler request
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | Sending scheduler request: Requested by user.
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | Requesting new tasks for NVIDIA
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
Thu 12 Mar 2015 10:10:57 PM PDT | Milkyway@Home | [sched_op] NVIDIA work request: 2419200.00 seconds; 2.00 devices
Thu 12 Mar 2015 10:10:58 PM PDT | Milkyway@Home | Scheduler request completed: got 0 new tasks
Thu 12 Mar 2015 10:10:58 PM PDT | Milkyway@Home | [sched_op] Server version 705
Thu 12 Mar 2015 10:10:58 PM PDT | Milkyway@Home | Project requested delay of 61 seconds
Thu 12 Mar 2015 10:10:58 PM PDT | Milkyway@Home | [sched_op] Deferring communication for 00:01:01
Thu 12 Mar 2015 10:10:58 PM PDT | Milkyway@Home | [sched_op] Reason: requested by project
Thu 12 Mar 2015 10:12:00 PM PDT | Milkyway@Home | [sched_op] Starting scheduler request
Thu 12 Mar 2015 10:12:00 PM PDT | Milkyway@Home | Sending scheduler request: To fetch work.
Thu 12 Mar 2015 10:12:00 PM PDT | Milkyway@Home | Requesting new tasks for NVIDIA
Thu 12 Mar 2015 10:12:00 PM PDT | Milkyway@Home | [sched_op] CPU work request: 0.00 seconds; 0.00 devices
Thu 12 Mar 2015 10:12:00 PM PDT | Milkyway@Home | [sched_op] NVIDIA work request: 2419200.00 seconds; 2.00 devices
Thu 12 Mar 2015 10:12:01 PM PDT | Milkyway@Home | Scheduler request completed: got 0 new tasks
Thu 12 Mar 2015 10:12:01 PM PDT | Milkyway@Home | [sched_op] Server version 705
Thu 12 Mar 2015 10:12:01 PM PDT | Milkyway@Home | Project requested delay of 61 seconds
Thu 12 Mar 2015 10:12:01 PM PDT | Milkyway@Home | [sched_op] Deferring communication for 00:01:01
Thu 12 Mar 2015 10:12:01 PM PDT | Milkyway@Home | [sched_op] Reason: requested by project

Subject to a reply from you, I will let the PC run overnight in the event Milkway decides to download WUs.
Collatz Conjecture is down again. No great surprise there.

Again, thanks for the help.

It's working!

I've been at this for years, trying to get Linux and GPUs working, and you did it, Roland. Give the man a cigar. Hell, give him the whole box.

Many thanks, you the man.

Gandalf

PS - If there is a way to mark this thread as solved, I don't know what it is.