Strange part is, as an end user you should never have to install the Cuda toolkit. Obviously it's not a suitable practical requirement for stock application deployment either.

For you, whatever works :), though this'll have to get slimmed down to the simplest possible steps with fewest downloads. I just don't recall it being that broken. Guess I'll have to have a look at mint soon.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Except your solution is leaving you without OpenCL. You will Not receive any GPU AstroPulse tasks until you get OpenCL working on your machine. From what I've experienced, getting OpenCL to work while using the 14.04 Repository drivers is not likely. If you want a working driver 352 then simply install the driver bundled with the 7.5 Toolkit After removing the Repository driver 352. The 352 driver bundled with the 7.5 Toolkit has been shown to work with OpenCL.

The next to last post at nVidia sums it up fairly well; Posted 04/23/2015 05:27 AM

...It sounds simple, but the way to get boinc communicating with your cards is to get rid of all of your previous driver software (if possible) and manually install the newest drivers from the website.
This requires downloading the setup.run file, making that file executable (right click, go to permissions or sudo chmod -x filename.run) and then booting into recovery mode and executing the setup (go to directory and ./filename.run) have it register with dkms but do not change your default desktop (unless you want to). I was unable to remove my previous drivers or install the new ones manually, something got corrupted with the ppa's and a fresh install was needed, which might be the case for you too if you have been trying to get it working with numerous setups.

After I installed the drivers I just installed boinc through package manager, nothing special, and it loaded and recognized the gpu's! I was good to go!

As noted in the post, usually by the time you arrive at the solution you have tried so many methods that your system has reached the FUBAR state. If you have a problem with the final solution, which Does Work, it would probably be best to reformat and make a clean install.

This is posted at nVidia, the People who make the drivers.

14.04 as in Ubuntu LTS like mine ? if so, yeah never got repo ones to work properly/completely, lol

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Yes, it would seem Linux Mint 17.3 Cinnamon is just Ubuntu 14.04.1 in another package. Look at the bottom of the page here, https://www.linuxmint.com/rel_rosa_cinnamon_whatsnew.php

Main components
Linux Mint 17.3 features Cinnamon 2.8, MDM 2.0, a Linux kernel 3.19 and an Ubuntu 14.04 package base.

If you open the Synaptic Package Manager and search for nvidia you will see the drivers are listed as 0ubuntu0.14.04.1. So, nothing special here, and I doubt the Ubuntu 14.04.1 package base is going to be changed anytime soon.

I think I've done about all I can with Mint, I really want to try Ubuntu 16.04. That means saying bye, bye to Mint.

Greetings All, looks like OP may have to do a clean install.

This is what I did and the result is below.

Clean install of MINT 17.3 on HDD
Once in desktop, do the Software Updates (it cant hurt)
Restart computer if it asks to, otherwise open up a web browser and navigate to the following websites to add the PPAs to your software resources list via the command line.

https://launchpad.net/~graphics-drivers/+archive/ubuntu/ppa

https://launchpad.net/~costamagnagianfranco/+archive/ubuntu/boinc

Open up the following link in your web browser and download the DEB package.

http://www.ubuntuupdates.org/package/core/xenial/multiverse/base/nvidia-modprobe

Open up Driver Manager (Menu - Administration - Driver Manager)
Select 361.42 from the list. Once complete it will ask to restart.

Upon restart and back in desktop, use GDEBI to install NVIDIA-MODPROBE.

use Synaptic Software Manager to install BOINC (select BOINC only). It will advise of the dependancies it will download to complete installation.

Restart and once back in desktop, run BOINC, then jump to event log to see your outcome, hopefully it will be the same as mine below.

Tue 26 Apr 2016 18:32:55 AWST | | Starting BOINC client version 7.6.2 for x86_64-pc-linux-gnu
Tue 26 Apr 2016 18:32:55 AWST | | log flags: file_xfer, sched_ops, task
Tue 26 Apr 2016 18:32:55 AWST | | Libraries: libcurl/7.35.0 OpenSSL/1.0.1f zlib/1.2.8 libidn/1.28 librtmp/2.3
Tue 26 Apr 2016 18:32:55 AWST | | Data directory: /var/lib/boinc-client
Tue 26 Apr 2016 18:32:55 AWST | | CUDA: NVIDIA GPU 0: GeForce GTX 670 (driver version 361.42, CUDA version 8.0, compute capability 3.0, 2047MB, 2006MB available, 2634 GFLOPS peak)
Tue 26 Apr 2016 18:32:55 AWST | | OpenCL: NVIDIA GPU 0: GeForce GTX 670 (driver version 361.42, device version OpenCL 1.2 CUDA, 2047MB, 2006MB available, 2634 GFLOPS peak)
Tue 26 Apr 2016 18:32:55 AWST | | Host name: ducati85mille-System-Product-Name
Tue 26 Apr 2016 18:32:55 AWST | | Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q9400 @ 2.66GHz [Family 6 Model 23 Stepping 10]
Tue 26 Apr 2016 18:32:55 AWST | | 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 syscall nx lm constant_tsc arch_perfmon pebs bts rep_good nopl aperfmperf pni dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 xsave lahf_lm dtherm tpr_shadow vnmi flexpriority
Tue 26 Apr 2016 18:32:55 AWST | | OS: Linux: 3.19.0-32-generic
Tue 26 Apr 2016 18:32:55 AWST | | Memory: 7.80 GB physical, 8.00 GB virtual
Tue 26 Apr 2016 18:32:55 AWST | | Disk: 450.45 GB total, 421.87 GB free
Tue 26 Apr 2016 18:32:55 AWST | | Local time is UTC +8 hours

Regards
Tazzduke

PS. I am using 361.42 drivers for my card as I have had no hiccups with that version, there are others to use which maybe more suitable.

Ran across an interesting tidbit while perusing the web this morning about MS and Lunix... Watch out, the 4 Horsemen of the Apocalypse must be on their way! Microsoft reveals all about Windows 10's Linux subsystem

---

Embrace-Extend-Extinguish has been going on for quite some time. It's not a huge step from virtualising a Linux Kernel, to the same for a Nix-BSD-OSX one. Might as well throw in an Xbox, Android JavaVM, and Doplhin and Mame emulator.

The problem I see coming with it is the difference between nixes Monolithic Kernel, OSX's Microkernel, and Windows' Hybrid. Suddenly security responsibilities (i.e. reliability, vulnerability, and efficiency) are no longer in the domain of one clear owner, but a hot potato in the hands of whoever is easiest to blame.

Switching to AmigaOS (functionally an ExoKernel ) is looking more appealing every day :D

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Greetings All, looks like OP may have to do a clean install.

This is what I did and the result is below...

Definitely something going on with his machine.
I'm running the cuda42 App on my Linux Mint with the 750Ti and driver 361.42 and it seems to be working the same way it did in Ubuntu; http://setiathome.berkeley.edu/results.php?hostid=7258715&offset=120
Now compare that with what the OP is getting; https://setiathome.berkeley.edu/results.php?hostid=7978786&offset=420
I've used Driver 361.42 on three different systems and it works fine on all of them. Right now I'm using the -poll command on the cuda42 App which makes it use a full cpu, but it runs a bit faster also.

I don't think the iGPU has anything to do with it either. The iGPU uses OpenCL, which the OP doesn't have, and needs an app_info entry for it even if it did have OpenCL. So, I'm not sure why his machines are running so Slow. You can see how the cuda42 App works on my GTS 250 & 8800 GT over here; http://setiweb.ssl.berkeley.edu/beta/results.php?hostid=72013&offset=280
Compare those times to the OPs GTS 250, https://setiathome.berkeley.edu/results.php?hostid=7904393&offset=280

Open up Driver Manager (Menu - Administration - Driver Manager)
Select 361.42 from the list. Once complete it will ask to restart.

Upon restart and back in desktop, use GDEBI to install NVIDIA-MODPROBE.

In my case, no 361.42 driver in list and no nvidia-modprobe in repository to Install.
So I have choosen other ways.

---

Greetings All

Is user Ralph Hyre (beta) Workunit ID# http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23347475
using a special app, as he is running a version of Linux and is getting what looks to be like OpenCL workunits.

I am a bit confused, or is it because he is a beta tester??????

Regards

Greetings Francesco,

That is correct it is not in the repository, you have to download the DEB package from the following link

http://www.ubuntuupdates.org/package/core/xenial/multiverse/base/nvidia-modprobe

Once downloaded, use GDEBI to install it.

Regards

Greetings All

Is user Ralph Hyre (beta) Workunit ID# http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=23347475
using a special app, as he is running a version of Linux and is getting what looks to be like OpenCL workunits.

I am a bit confused, or is it because he is a beta tester??????

Regards

SETI@home v8 v8.07 (opencl_nvidia_SoG) x86_64-pc-linux-gnu

whatever he is running meets the specs for that stock app. Or met them in march, he doesn't seem to be active any more.

---

A person who won't read has no advantage over one who can't read. (Mark Twain)

Okay, no worries.

Am off now to rebuild my 3rd PC. Replacing Ubuntu 16.04 with Mint 17.3.

2nd PC had Win 7, then Ubuntu, then Mate now has Mint 17.3 and is successfully crunching (Ran some SETI Beta units)

Okay, from here its nighty night lol
Tazzduke

Finally got Ubuntu 16.04 running. The Good news is if you have a GTX 750Ti installed, Driver 361.42 IS IN the Repository and it DOES WORK....with cuda.
The Bad news is after all this time the Repository 361.42 Still DOESN'T work with OpenCL. Kinda Unbelievable isn't it.
However, it is easy to fix by just making a link to /usr/lib/x86_64-linux-gnu/libOpenCL.so.1, moving it to /usr/lib, and renaming it libOpenCL.so or libOpenCL.so.1. I dunno which one it really needs, I made two links and named them libOpenCL.so and libOpenCL.so.1. Heck it might even work by just making a link to /usr/lib/x86_64-linux-gnu/libOpenCL.so.1 and naming the link libOpenCL.so without moving it. Anyway, you can use the Repository driver with minimum effort, which is much better than it was.

The biggest problem is getting a version of BOINC, which works, to use in your Home folder. I ended up using the old version 7.4.25, which doesn't update correctly. I suppose I'm going to have to build one, if I can find the source files.

After all the trouble, 16.04 seems to work about the same as 15.10, 15.04, 14.10, etc...
Oh, according to the Mint blog, the new version of Mint based on Ubuntu 16.04 should be out next month...or maybe the month after.

Greetings

That wasn't so painless for me, just rebuilt number 3 pc and have MINT 17.3.

Crunching away at this moment, still awaiting for some SETI, but seems to be a bit dry. This machine has a GTX 960 onboard.

Hey TBar, well thats strange, as the NVIDIA drivers and the BOINC package I am using recognise CUDA and OPENCL, and have crunched on both.

I do invoke NVIDIA-MODPROBE though and reboot and then start up BOINC and attach.

Regards
Tazzduke

Greetings

One more thing before I hit the hay, after a lot of trial and error, trust me there have been, I think I have just managed to find a way that works for my setup.
Things might work one way for one person, but cannot be guaranteed to work for the next person, it all comes down to computer setup, ie hardware, gpu's, os, drivers and kernel/software.

Thanks also to Tbar, Jason and all, for their bits of advice as well as directions to go in. It is appreciated.

Good Night All
Tazzduke

Yes, it looks as though Modprobe is back in the repository with 16.04, as we know, it's Not in the Mint repository. Strange they don't install it with the Repository nVidia driver. I prefer to use the Link instead, that way when I pull the nVidia card out and insert an ATI card the link should still be there and result in OpenCL working with the ATI card. I haven't tested it with an ATI card, but chances are those drivers will also not install the link just as they haven't since at least 2014. I don't have modprobe, and my driver now works with OpenCL, and I didn't have to install any other package. I try to avoid installing things I don't actually need.

Yes, it looks as though Modprobe is back in the repository with 16.04, as we know, it's Not in the Mint repository.

Just to be more exact, I have in this moment 4 host with mint 17.3 and in some of them I have nvidia mobprobe in the repository. I don't know how. May be some machine was mint 17 and then updated to 17.1, > 17.2 and finally 17.3. Max be something depend on Geforce GPU.

---

I believe I've mentioned how it depends on the hardware you have installed. If you change cards you will be offered different driver packages. My GTX 750Ti doesn't get offered the Modprobe in Mint 17.3 either. However, Ubuntu 16.04 does offer the modprobe with my 750 installed. The Next Release of Mint is supposed to be based on Ubuntu 16.04 verses the now 14.04.1, the Next Release of Mint should have the Modprobe offered with the 750. Back when 14.04.1 was released the 750Ti was a New card, I don't think it was supported very well with 14.04.1. Apparently it won't matter, I don't need Modprobe to get the 16.04 Repository 361.42 driver to work with my 750Ti. The same fix that has worked with the ATI cards for years also works with the new Ubuntu and the 750Ti, just place a link to the OpenCL library into /usr/lib. Strange how you don't need to create that link when using the Same driver from nVidia...not.

So, what's up with all the "finish file present too long" Errors?
https://setiathome.berkeley.edu/results.php?hostid=7974150&state=6

So, what's up with all the "finish file present too long" Errors?
https://setiathome.berkeley.edu/results.php?hostid=7974150&state=6

Probably implies that Linux *may* have moved to buffered IO/Multithreaded-C-Runtimes, which would probably make sense with more mobile targets these days wanting to offload to a dedicated IO sentinel core. (Lazy IO as has been normal in Windows since ~2005, as a desktop optimisation.)

Stock Boincapi and the client don't cope extraordinarily well with buffered IO and threading. You can try renicing to a more normal process priority (which may help).

As you would be using some standard boincapi as my Linux builds do, for a more permanent code fix, If updating your boincapi used in the build doesn't help (Rom switched at least some file IO to commit mode), then until the client is made less fussy, you probably need to shift the finished file creation from where it is in boinc_finish() (or whatever) into boinc_exit() as last call before exit().

At least those are the workaround/messy conclusions we reached on Windows. If similar symptoms are manifesting on Linux, and the workarounds are similar, then it points at proper fixes (for Boinc)

[If you need it, Can post a patch for current api after Cuda8 testing is over with]

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.