Help ! With Linux Mint 18.3

[biodoc]

It'll be interesting to how that works.

If you want to try a different display manager like gdm then:

sudo apt-get install gdm
sudo dpkg-reconfigure gdm
sudo service lightdm stop
sudo service gdm start

 

[Markfw]

So what I did was exclude everything after it started folding successfully. If it reboots it will be interesting to see if that fixed the problem.

 

[biodoc]

So what I did was exclude everything after it started folding successfully. If it reboots it will be interesting to see if that fixed the problem.

See my post above if you'd like to switch the display manager to gdm.

 

[biodoc]

OK, new try. Below is the list I did NOT run, its running 2.5 million, lets let it run and see if it does not reboot, or if it does, if it still works: (the menu install for drivers includes the openCL and cuda)

sudo apt-get install nvidia-opencl-icd-387

sudo apt-get install libcuda1-387

sudo apt-get -y install python-gnome2 mesa-common-dev freeglut3-dev nvidia-settings gedit

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

cd ~

gedit fancontrols

and add and then save (for 2 video cards)

nvidia-settings -a [gpu:0]/GPUFanControlState=1
nvidia-settings -a [fan:0]/GPUTargetFanSpeed=100
nvidia-settings -a [gpu:1]/GPUFanControlState=1
nvidia-settings -a [fan:1]/GPUTargetFanSpeed=100

I noticed that your protocol doesn't include a system update after the linux install and before the nvidia driver installation. Your rig contains the latest and greatest hardware so I think a system upgrade is key.

sudo apt-get update
sudo apt-get dist-upgrade

 

[Markfw]

I noticed that your protocol doesn't include a system update after the linux install and before the nvidia driver installation. Your rig contains the latest and greatest hardware so I think a system upgrade is key.

sudo apt-get update
sudo apt-get dist-upgrade

It has the update, but not the dist-upgrade. Next thing to try if this doesn't work.

 

[biodoc]

It has the update, but not the dist-upgrade. Next thing to try if this doesn't work.

The update just identifies which packages need updating. The dist-upgrade command downloads and installs those packages.

Then reboot

 

[TennesseeTony]

BumbleBee is having some issue(s). Normally I walk by the main cluster, look at the display on the UPS, see 1.79-1.83 kw on the display, assume everything is running and walk away. Now I see 1.63-1.79 kw, and it's always BumbleBee, running the GPU rather poorly (GTX 1080).

This is a 28 thread box, with 2-3 free threads just for Folding. Previously it has been working at 800k ppd, just like the other linux 1080's, but now it fluctuates wildly between 300k-550k ppd. And it's not just the estimate, it is actually not putting out the points. The "Estimated TPF" (whatever that is) goes from mid 40 seconds to 1m41s. Now here is the kicker...it goes right back to 800k ppd if I disable (suspend) BOINC (messes up with both WCG [both SCC and MIP] and Universe, haven't tried any other tasks yet).

I am going to finish current running tasks while I do a bit of internet self-help. I've tried newer drivers already. Any thoughts while I search for answers and eat some dinner?

 

[biodoc]

My 3930K with 2x1080s seems to work fine with Boinc (primegrid) and F@H.

I have 12 logical cores with 8 dedicated to SOB (multithreaded with 1 WU), 2 free cores and 2 cores for the 2 1080s.

If I run "top", I see 2 x FahCore_21 with each using 100% of a logical core (%CPU column) and primegrid_llr using 728% CPU (nearly 8 logical cores). The 2 free cores are for other apps since this is my main rig.

 

[TennesseeTony]

And now we know your first name.

I'll run top in a moment....but until then...

More info:

Nvidia Xserver reports:

• Card is running at a cool 56C
• Card is running at BOOST clock for both chip and memory
• Card is utilized at up to 97%, but then drops to 64%, then drops to zero for a few seconds, then back to 97%
• Running on 8x PCIe 3.0

I have tried setting all the FAH processes (5, 6 of them?) at higher priority, no luck there. Tried reducing the number of BOINC tasks running, even at 1 task, performance of the GPU is handicapped.

 

[TennesseeTony]

Ran "top" and the results add to the confusion, same as the process tab on System Monitor. Both show the CPU has maybe 10 processes that are using up to a whopping 0.4%. I restarted the tasks in BOINC, and nothing changed, still only a few tasks running at 0.1-0.4% . Using System Monitor, and one of the other tabs, it does show activity on the various threads, but not on the processes tab.

 

[Markfw]

I ran top, and now I am real curious. Your stuff is all running under your login name, mark. Mine runs as user boinc or fahclient. Hardly anything as me.

Here:

 

[TennesseeTony]

I may have cured the GPU slowdown......by adding a 2nd card. ??? Anyway, first boot after new card was awesome, added 2nd GPU Folding Slot, all is well. Made adjustments to manual fanspeed control start-up file, restarted, and.........I got Mark's blank screen issue. BUT, it did show up on TeamViewer. BUT, Cinnamon has crashed, would you like to restart Cinnamon? No thanks, I'll just reboot and give it another try. Nope, blank screen again.

BUT, and here's the good stuff, I fixed my blank screen problem rather quickly and easily....I just plugged the monitor into the 2nd GPU (in this case, the one furthest from the CPU. Had the same issue with the 2P system yesterday, had to plug into the 'bottom' card, furthest from the CPUs).

 

[Markfw]

I may have cured the GPU slowdown......by adding a 2nd card. ??? Anyway, first boot after new card was awesome, added 2nd GPU Folding Slot, all is well. Made adjustments to manual fanspeed control start-up file, restarted, and.........I got Mark's blank screen issue. BUT, it did show up on TeamViewer. BUT, Cinnamon has crashed, would you like to restart Cinnamon? No thanks, I'll just reboot and give it another try. Nope, blank screen again.

BUT, and here's the good stuff, I fixed my blank screen problem rather quickly and easily....I just plugged the monitor into the 2nd GPU (in this case, the one furthest from the CPU. Had the same issue with the 2P system yesterday, had to plug into the 'bottom' card, furthest from the CPUs).

Oh, cool, not sure it I ever tried that ! !! But my Intel 6700k system crashes when I try to boot into it after a shutdown. Idd error, and its a hard crash

 

[StefanR5R]

I ran top, and now I am real curious. Your stuff is all running under your login name, mark. Mine runs as user boinc or fahclient.

boinc-client and FAHClient can be started in three ways:

1. as a service, started automatically at boot,
2. as a service, started manually,
3. started directly in the command line.

Neither one of these options affects performance, per se. Options 1. and 2. generally result in the processes being run with those pseudo-users (by default boinc and fahclient on Mint, but it can be reconfigured); 3. results in running as the user who is logged in into the command line (unless doing something like sudo -u boinc -g boinc boinc [...]).

I use 1. for boinc on some but not all of my machines, 2. for boinc on others, and 3. if I run multiple boinc instances on the same machine. I use 2. for FAHClient on my GPU hosts, and 3. for FAHClient on my CPU hosts (out of habit, not for a specific reason).

Re 1. vs. 2., some distros have a GUI to control that, vaguely similar to how we are used to control services on Windows. On Mint, the command line needs to be used with some obscure commands. I didn't find Mint-specific documentation for that, and followed Ubuntu documentation.

Edit:
E.g. here is some discussion of starting/ stopping/ enabling/ disabling services on various Ubuntu versions:
https://askubuntu.com/questions/19320/how-to-enable-or-disable-services
On Mint 18.3, the older as well as the newer commands appear to work (the older being wrappers for the newer), and I don't know which set of commands will be the preferred ones on Mint, going forward.

Somewhere I read that F@H should be started after the desktop, but the default service configs don't ensure this. Don't know if either of this is true.

 

[biodoc]

I may have cured the GPU slowdown......by adding a 2nd card. ??? Anyway, first boot after new card was awesome, added 2nd GPU Folding Slot, all is well. Made adjustments to manual fanspeed control start-up file, restarted, and.........I got Mark's blank screen issue. BUT, it did show up on TeamViewer. BUT, Cinnamon has crashed, would you like to restart Cinnamon? No thanks, I'll just reboot and give it another try. Nope, blank screen again.

BUT, and here's the good stuff, I fixed my blank screen problem rather quickly and easily....I just plugged the monitor into the 2nd GPU (in this case, the one furthest from the CPU. Had the same issue with the 2P system yesterday, had to plug into the 'bottom' card, furthest from the CPUs).

Wow, I don't have an explanation for why that works but that's cool that it does.

Regarding cinnamon crashing, I think that it has to do with the switch (in bold) in the following command in the protocol.

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

That switch might be important in recognizing multiple gpus in your system but when I applied it on my 17.3 system and rebooted, cinnamon crashed. When I edited /etc/X11/xorg.conf and removed the offending lines and rebooted, cinnamon did not crash. Both my GPUS were still recognized by FAHClient after I removed those lines.

I highlighted the lines in /etc/X11/xorg.conf that need to be removed to allow cinnamon to run without crashing.

Section "Screen"
Identifier "Screen0"
Device "Device0"
Monitor "Monitor0"
DefaultDepth 24
Option "AllowEmptyInitialConfiguration" "True"
SubSection "Display"
Depth 24
EndSubSection
EndSection

Section "Screen"
Identifier "Screen1"
Device "Device1"
Monitor "Monitor1"
DefaultDepth 24
Option "AllowEmptyInitialConfiguration" "True"
SubSection "Display"
Depth 24
EndSubSection

 

[StefanR5R]

Regarding cinnamon crashing, I think that it has to do with the switch (in bold) in the following command in the protocol.

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

Interesting. I mentioned that my systems did not exhibit cinnamon crashes after (re)boot. This may be related to the fact that I did not perform the coolbits step* on any of my two** Mint installations yet. That supports your finding that this config option may be the cause of the crashes.

--------
*) All my GPUs, except one in a remaining Windows-only PC, are watercooled, hence no need for GPU fan control. I may play with the coolbits thing eventually though, to attempt to optimize perf/W.
**) The first Mint installation, which I bootstrapped with the nouveau driver before installation of the nvidia driver, and which I cloned to more PCs, does not have an /etc/X11/xorg.conf file in the first place. The other installation, which I needed to boot into compatibility mode before being able to install the nvidia driver, does contain a minimal /etc/X11/xorg.conf file which refers to the vesa driver.

 

[biodoc]

The more I read, the more confused I am. My problem is that I never have done a clean install with Mint 18.3 and I've never installed the nvidia drivers from this alternative source so my experience with the protocol @brony is null. I've always downloaded drivers directly from nvidia and used a installation protocol that is similar to this from the folding forum.

 

[Markfw]

The more I read, the more confused I am. My problem is that I never have done a clean install with Mint 18.3 and I've never installed the nvidia drivers from this alternative source so my experience with the protocol @brony is null. I've always downloaded drivers directly from nvidia and used a installation protocol that is similar to this from the folding forum.

Here is my entire tested procedure. It works perfectly (the first boot). Some people said to stop right before the line that has the allow empty config. I tied it once, and still working so far.

First, disable secure boot. The easiest way I found was to delete the keys in key management.

Next, after booting with a linux mint USB or CD, install it. Use an empty disk, or one you don't care if its wiped clean.

Then open a terminal window and do the below commands

wget http://launchpadlibrarian.net/109052632/python-support_1.0.15_all.deb

sudo dpkg -i python-support_1.0.15_all.deb

sudo add-apt-repository ppa:graphics-drivers/ppa

sudo apt-get update

sudo apt-get install nvidia-387 (or if it appears on the linix home page drivers list, you can do it from there.

sudo apt-get install nvidia-opencl-icd-387

sudo apt-get install libcuda1-387

sudo apt-get -y install python-gnome2 mesa-common-dev freeglut3-dev nvidia-settings gedit

Install fah client and console using these links:
Folding at home client (background service)  https://folding.stanford.edu/releases/public/release/fahclient/debian-testing-64bit/v7.4/fahclient_7.4.4_amd64.deb
Folding at home control panel (user interface) https://folding.stanford.edu/releases/public/release/fahcontrol/debian-testing-64bit/v7.4/fahcontrol_7.4.4-1_all.deb

cd /var/lib/fahclient

sudo wget http://fah-web.stanford.edu/file-releases/public/GPUs.txt

Then fire up fahcontrol and add the GPU client(will show under education in the menu), and maybe also the remote access by adding ,192.168.1.1-192.168.1.255 to the two IP places

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

cd ~

gedit fancontrols

and add and then save (for 2 video cards)

nvidia-settings -a [gpu:0]/GPUFanControlState=1
nvidia-settings -a [fan:0]/GPUTargetFanSpeed=100
nvidia-settings -a [gpu:1]/GPUFanControlState=1
nvidia-settings -a [fan:1]/GPUTargetFanSpeed=100

 

[biodoc]

Looks good except for one thing.

After the command:

sudo apt-get update

(this command checks the system repositories and compares with your system to see which packages need to be upgraded)

then run:

sudo apt-get dist-upgrade

(this command downloads and installs the packages on your system. This could take an hour or so but be patient)

then reboot the system.

Unlike windows 10 where installation of system files is automatic, on linux, the user is expected to take care of this.

On linux Mint, you can tell if your system needs updates to be installed by running the 2 commands above or by looking at the "shield icon" on the bottom right next to the clock. If the shield icon has a check mark in it then your system is up to date. If the shield icon has an exclamation point in it, then there are system files that need to be downloaded and installed.

At least in my mind, the idea is to:

1) install the system from the distro CD or USB stick.
2) make sure it's fully upgraded to the latest kernel and other system packages (reboot)
3) install the nvidia drivers (reboot may a couple of times to see if everything is ok)
4) install F@H

 

[Markfw]

Ok, new info. So I had to reboot a linux box today, and I got the black screen with the mouse. So I tried Tony's trick with the other video card. Did not work, so then I moved it back to the primary, and BAM, video ! Really odd problem, and I lost a few hundred k points, but all is back now.