The Threadripped system something really wrong. I pulled it offline and will try to see whats wrong...I cant install Linux on it because I use it for Rendering sometimes during the day. (ofcourse these times-I turn off boinc) but past 2 weeks i did no rendering.

You can if you just do a Ubuntu USB bootable, reduce the size (shrink in diskmanager) on your bootdisk (if you're not using softraid or anything weird) by about 25 GB, then you can install Ubuntu Desktop easily and dualboot your machine.
I do the same with my 2080Ti host. I use it for games and when i'm done a just reboot it because first bootup line is to start Ubuntu and seti@home autostarts in Ubuntu, so i can turn my back when i'm done with Windows and don't need to babysit the reboot process! :)

If i boot with windows i need to select that option in the startboot menu. Picture below is example when booted from Ubuntu desktop install along with the recognition of Windows already installed.


Linux is the best way now to take advantage of the Nvidia gpu's full potentional. You can increase a gpus throughput (RAC) with about 400% when running a single task only on the gpu. Look at the top hosts as they all run Linux and Petris/TBars executable offerings.

Youtube guide as example below but i use Rufus boot manager instead:

https://www.youtube.com/watch?v=u5QyjHIYwTQ
https://rufus.ie

---

_________________________________________________________________________
Addicted to SETI crunching!
Founder of GPU Users Group

I wanted to update you guys, I did monitor my rac and noticed it going up:

11/09/2019
Recent average credit 72,134.85
VIPER 12,955.28
[2] NVIDIA GeForce GTX 980 (4095MB)
Emily 17,060.45
[4] NVIDIA GeForce GTX 690 (2048MB)

11/13/2018
Recent average credit 86,327.17
VIPER 17,245.60
[2] NVIDIA GeForce GTX 980 (4095MB)
Emily 22,363.04
[4] NVIDIA GeForce GTX 690 (2048MB)

Once I have the time and confidance, I will try to get these machines on Linux and see if I can maximize my RAC. I am just not familiar with Linux and I need some time to read/learn more about it and how to run the opmitized apps on Linux.

Thank you all for your help and support on this issue, I really appreciate it

---

Stephen made a thread showing the simple steps to installing Linux. Then it is very simple to go to Crunchers Anonymous and download TBar's All-In-One BOINC package which is already pre-configured to run the special CUDA9 app. As simple as unpacking the downloaded file onto your desktop and double clicking the Manager executable.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Famous last words.....

Stephen made a thread showing the simple steps to installing Linux. Then it is very simple to go to Crunchers Anonymous and download TBar's All-In-One BOINC package which is already pre-configured to run the special CUDA9 app. As simple as unpacking the downloaded file onto your desktop and double clicking the Manager executable.

I so stipulate. But you apparently still have to install p7zip, webkit*, and curl* from the command line or I can't get it to work :)

Tom

---

A proud member of the OFA (Old Farts Association).

But is any of that information hidden? No, it is right there in the download instructions on what is necessary to use the package. Just three Terminal commands and you are ready to double click the Manager to start BOINC.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

But is any of that information hidden? No, it is right there in the download instructions on what is necessary to use the package. Just three Terminal commands and you are ready to double click the Manager to start BOINC.

Must be in some docs I missed. Didn't see them on the directions inside the archive. And you can't get inside the archive unless you have installed the p7zip. And the last page I downloaded the "all in one from" didn't have any download directions about the 3 items I mentioned. So need to mention them frequently for the poor readers like me :)

Tom

---

A proud member of the OFA (Old Farts Association).

Must be right in front of your face from the post that contains the download link at CA.
https://arkayn.us/forum/index.php?msg=4515

This is a Package to install the Updated Berkeley BOINC version 7.8.3 with the CUDA 9.0 Special App Preinstalled. You should check the Dependencies and Execute bits before running. Most distros will need to install libwebkitgtk, some libcurl3. Mint users should probably use the repository version of BOINC. Read the README.md and the README_x41p_V0.97.txt in the docs folders. See 'The Berkeley Installer' for outdated instructions. To 'install', simply expand the download to your Home folder, make sure Dependencies are installed, open the BOINC folder, and double click boincmgr. You will need to add the Project SETI@Home to BOINC.

To Install the Dependencies, In the terminal;
To install 7zip run, sudo apt-get install p7zip
To install libwebkitgtk run, sudo apt-get install libwebkitgtk-1.0-0
To install libcurl3 run, sudo apt-get install libcurl3

The CUDA 9.0 Special App Requires at least Ubuntu 14.04.1 and a CUDA 9.0 driver installed. The Repository Driver 384.x should work.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Do you still require the special command line for each GPU type in linux?

Second: any hope for these GpUz, I know they pretty low end:

NVIDIA GeForce GTX 560 Ti (1024MB) driver: 388.13 OpenCL: 1.1
AMD A8-3870 APU with Radeon(tm) HD Graphics [Family 18 Model 1 Stepping 0]


NVIDIA GeForce GTX 460 (1024MB) driver: 388.13 OpenCL: 1.1
Intel(R) Core(TM) i3-4130 CPU @ 3.40GHz [Family 6 Model 60 Stepping 3]

---

I think I know what you are asking. You want to know if the OpenCL command line for SoG tasks is required for Linux?? Yes you can run the stock SoG applications in Linux and use the command line parameters we have been discussing in this thread.

But the whole point of running Linux is to run the special app CUDA9 or CUDA10 applications which don't use OpenCL and therefore don't require the OpenCL parameters. The special app CUDA applications use an internal command line that automatically optimizes the task parameters for each card type in the host without requiring any user intervention.

The special app has its own requirements for minimal CC capability of the Nvidia card. The CUDA 9.0 Apps requires at least CC=5.0. The CUDA 9.2 App requires at least CC=6.1

None of those older Nvidia cards have that capability. The 560Ti was/is notorious for generating tons of invalids and inconclusives. I would stay away from using that card. It just runs too hot for the stock clocks/fan profiles. The 460 card is just wasteful of energy consumed for the amount of output. I'm not familiar with any of the AMD APU chips so have no opinion. Mike would have an opinion on use of that host cpu/gpu combo.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

If you have a GTX 780 or 780 Ti, there is a CC 3.5 special app for them. But you would be better off power usage wise with a later family model GTX card.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

. I'm not familiar with any of the AMD APU chips so have no opinion. Mike would have an opinion on use of that host cpu/gpu combo.

I have a little familiarity with some of the older APU's (I have an A4 and an A6, had an A-10?) and we have been "watching" the newer generation of Ryzen 3 2200G, Ryzen 5 2400G's. From a pure Seti production point of view, the short answer is don't go out and buy any of the APU systems. Or if you do, plan on adding a discrete gpu card like a 750Ti or 1050 or higher in the future. Otherwise it appears you are mired in the sub 5,000 RAC ranks, even the sub-2,000 RAC ranks.

We discuss various ideas for speeding them up here based on the research threads I have been pointed to by other helpful Seti-Citizens (eg. on a 4 cpu system, running 3 cpus gains you more total production when running the gpu than running 4 cpus and no gpu).

https://setiathome.berkeley.edu/forum_thread.php?id=83390

HTH,
Tom

---

A proud member of the OFA (Old Farts Association).

If you have a GTX 780 or 780 Ti, there is a CC 3.5 special app for them. But you would be better off power usage wise with a later family model GTX card.

I am running a GTX 750Ti on Linux/CUDA91 here https://setiathome.berkeley.edu/show_host_detail.php?hostid=8565979 and the Gtx 750Ti gpu is ticking along at around 4 minutes per Seti gpu task. That is almost twice as fast as a Windows gtx 1060 3GB running the standard Seti SOG gpu task that I had.

The system is running a heavy mixed BOINC load on the cpu side so the results are less than a pure Seti box would show.

Tom

---

A proud member of the OFA (Old Farts Association).

My numbers went up >>> I am very pleased so far. I need to do some cleaning up and remove any useless machines from now account, and perhaps do some GPU upgrading if I can afford it soon. I will also find time this weekend or next to try Linux and go from there. Once again I appreciate all the feedback and support from everyone here! Lets find ET!

---

Once again I appreciate all the feedback and support from everyone here! Lets find ET!

RIGHT ON!

---

A proud member of the OFA (Old Farts Association).