Setting up Linux to crunch CUDA90 and above for Windows users

Message boards : Number crunching : Setting up Linux to crunch CUDA90 and above for Windows users
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 77 · 78 · 79 · 80 · 81 · 82 · 83 . . . 162 · Next

AuthorMessage
Profile Keith Myers Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 29 Apr 01
Posts: 13164
Credit: 1,160,866,277
RAC: 1,873
United States
Message 1957432 - Posted: 27 Sep 2018, 7:18:35 UTC - in response to Message 1957431.  

Yes, exactly correct. You need to remove all vestiges of the previous drivers for a good working reinstall. That is where I goofed Sunday when I just purged and didn't autoremove too when I tried installing the CUDA10 drivers.

Good to hear you are up and running again Kevin.
Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)
ID: 1957432 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957445 - Posted: 27 Sep 2018, 11:39:26 UTC

I am starting to think my issue with the disabled "second monitor" is probably a problem with KDE and how it handles multi monitors..
I am pretty sure my xconf is set correctly.

When i boot the computer, when logging in, I have a splash screen with KDE logo/loading, at this point i have my whole screen, but there is a mirrored smaller screen on the top left corner,
the resolution of this small screen seems to be 640x480 so it match the "X Screen 1" properties. When at the desktop it vanish (i think it is moved to where xconf set it up (right of "screen 1") by default.
And it looks like KDE doesn't like this fake non-connected display much and let me move my cursor into it, even if it is shown as "displays: none" for this screen.

I could try normal Ubuntu, but i am not much of a gnome fan, so i might just leave it like this. Maybe it will get fixed some days..
ID: 1957445 · Report as offensive     Reply Quote
Profile Brent Norman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 2786
Credit: 685,657,289
RAC: 835
Canada
Message 1957452 - Posted: 27 Sep 2018, 13:50:36 UTC - in response to Message 1957445.  

Can you see 2 monitors in X Server Settings?
You should be able to disable 1 there, or even set them to be clones. So you have only 1 screen.
ID: 1957452 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957466 - Posted: 27 Sep 2018, 17:03:20 UTC - in response to Message 1957452.  

Can you see 2 monitors in X Server Settings?
You should be able to disable 1 there, or even set them to be clones. So you have only 1 screen.


No, when in the X server display configuration, i only see one screen i can move around..
Same thing if i go into KDE monitor settings, no trace of a second screen.

The only control i have over this "fake" non connected screen is in the nvidia x server display configuration, on the dropdown box
if i select X screen 1, i can select its color depth, and position/offset (absolute, right of, left of... above or below)
I am guessing that is the normal behavior even if nothing is connected.

If i use the --only-one-x-screen, my second gpu vanish from xconf, so coolbits won't be enabled on it
ID: 1957466 · Report as offensive     Reply Quote
juan BFP Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 16 Mar 07
Posts: 9786
Credit: 572,710,851
RAC: 3,799
Panama
Message 1957470 - Posted: 27 Sep 2018, 17:34:21 UTC
Last modified: 27 Sep 2018, 17:34:56 UTC

How many active displays shows nvidia-smi ?
ID: 1957470 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957495 - Posted: 27 Sep 2018, 21:05:24 UTC - in response to Message 1957470.  

How many active displays shows nvidia-smi ?

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 396.54                 Driver Version: 396.54                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 106...  Off  | 00000000:26:00.0  On |                  N/A |
| 60%   60C    P2   107W / 156W |   2150MiB /  6075MiB |     99%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 106...  Off  | 00000000:27:00.0 Off |                  N/A |
| 80%   65C    P2    84W / 120W |   1966MiB /  6078MiB |     99%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1323      G   /usr/lib/xorg/Xorg                           135MiB |
|    0      1939      G   /usr/bin/kwin_x11                             38MiB |
|    0      1943      G   /usr/bin/krunner                               1MiB |
|    0      1945      G   /usr/bin/plasmashell                          59MiB |
|    0     14606      C   ...thome_x41p_V0.97b2_Linux-Pascal+_cuda92  1911MiB |
|    1      1323      G   /usr/lib/xorg/Xorg                            20MiB |
|    1      1954      G   /usr/bin/kwin_x11                             30MiB |
|    1     14621      C   ...thome_x41p_V0.97b2_Linux-Pascal+_cuda92  1911MiB |
+-----------------------------------------------------------------------------+


Not sure how to find it? is it the (Disp.A) next to the BusID?
ID: 1957495 · Report as offensive     Reply Quote
Kevin Olley

Send message
Joined: 3 Aug 99
Posts: 906
Credit: 261,085,289
RAC: 572
United Kingdom
Message 1957496 - Posted: 27 Sep 2018, 21:13:04 UTC - in response to Message 1957432.  


Good to hear you are up and running again Kevin.


Just given it another prod, now got TBar's latest running, just getting my first valids coming through.
Kevin


ID: 1957496 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1957510 - Posted: 27 Sep 2018, 22:41:55 UTC - in response to Message 1957495.  
Last modified: 27 Sep 2018, 22:51:59 UTC

How many active displays shows nvidia-smi ?

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 396.54                 Driver Version: 396.54                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 106...  Off  | 00000000:26:00.0  On |                  N/A |
| 60%   60C    P2   107W / 156W |   2150MiB /  6075MiB |     99%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 106...  Off  | 00000000:27:00.0 Off |                  N/A |
| 80%   65C    P2    84W / 120W |   1966MiB /  6078MiB |     99%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      1323      G   /usr/lib/xorg/Xorg                           135MiB |
|    0      1939      G   /usr/bin/kwin_x11                             38MiB |
|    0      1943      G   /usr/bin/krunner                               1MiB |
|    0      1945      G   /usr/bin/plasmashell                          59MiB |
|    0     14606      C   ...thome_x41p_V0.97b2_Linux-Pascal+_cuda92  1911MiB |
|    1      1323      G   /usr/lib/xorg/Xorg                            20MiB |
|    1      1954      G   /usr/bin/kwin_x11                             30MiB |
|    1     14621      C   ...thome_x41p_V0.97b2_Linux-Pascal+_cuda92  1911MiB |
+-----------------------------------------------------------------------------+


Not sure how to find it? is it the (Disp.A) next to the BusID?


. . Yes that is the Display Active indicator. But I am surprised about a couple of things. Firstly, that while you have two GTX1060-6GB cards nvidia-smi shows two different TDP values. And while the unit with the display active (I am presuming it is the top card) shows the higher TDP and current power consumption it is showing a lower temperature and fan speed than the less 'power hungry' card. Maybe you need to try a different version of the video drivers or maybe your overclocking settings are confusing something?

. . Also if you look in the active processes section you will see that the 'upper' card is running the /runner and /plasma shell functions which are your display. As well the second card is running its own copies of /xorg and /kwin_x11. My rig does not show that, they are only running on the first card.

Stephen

??
ID: 1957510 · Report as offensive     Reply Quote
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1957512 - Posted: 27 Sep 2018, 22:47:49 UTC - in response to Message 1957510.  
Last modified: 27 Sep 2018, 22:48:08 UTC

But I am surprised about a couple of things. Firstly, that while you have two GTX1060-6GB cards nvidia-smi shows two different TDP values. And while the unit with the display active (I am presuming it is the top card) shows the higher TDP and current power consumption it is showing a lower temperature and fan speed than the less 'power hungry' card. Maybe you need to try a different version of the video drivers or maybe your overclocking settings are confusing something?

. . Also if you look in the active processes section you will see that the 'upper' card is running the /runner and /plasma shell functions which are your display.

Stephen

??


Probably different manufacturers of the same card.
ID: 1957512 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957513 - Posted: 27 Sep 2018, 22:50:08 UTC - in response to Message 1957510.  
Last modified: 27 Sep 2018, 22:59:20 UTC

No it is quite normal, the top one is a factory OC'ed with higher tdp limits, 3fan model with huge heatsink (Asus ROG Strix 1060 6G)
the second one is a normal clocked card with a very tiny heatsink and only a single fan. (EVGA Gaming 1060 6G)
On a cool day the Asus will run 2164mhz and probably more, The 1fan card would probably catch fire at those frequency hehe

I am only using coolbits on the 2nd card to control the fan speed
ID: 1957513 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1957516 - Posted: 27 Sep 2018, 22:59:32 UTC - in response to Message 1957513.  
Last modified: 27 Sep 2018, 23:00:31 UTC

No it is quite normal, the top one is a factory OC'ed with higher tdp limits, 3fan model with huge heatsink (Asus ROG Strix 1060 6G)
On a cool day the Asus will run 2164mhz and probably more, The 1fan card would probably catch fire at those frequency hehe


. . That speaks volumes for the Strix card then eh? I guess there IS an advantage in running the top end version cards for a given GPU.

Stephen

:)
ID: 1957516 · Report as offensive     Reply Quote
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1957517 - Posted: 27 Sep 2018, 23:03:38 UTC - in response to Message 1957516.  

No it is quite normal, the top one is a factory OC'ed with higher tdp limits, 3fan model with huge heatsink (Asus ROG Strix 1060 6G)
On a cool day the Asus will run 2164mhz and probably more, The 1fan card would probably catch fire at those frequency hehe


. . That speaks volumes for the Strix card then eh? I guess there IS an advantage in running the top end version cards for a given GPU.

Stephen

:)


If it will fit in your case....
ID: 1957517 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957518 - Posted: 27 Sep 2018, 23:07:46 UTC

Yes i guess there is one, i bought this card used tho, which makes more sense.
Because the price of this card when new was a bit ridiculous considering the price of a 1070 was rather close.
ID: 1957518 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1957519 - Posted: 27 Sep 2018, 23:09:14 UTC - in response to Message 1957513.  

I am only using coolbits on the 2nd card to control the fan speed


. . What script are you using for your overclocking and can you remember the command line you used to set coolbits?

Stephen

?
ID: 1957519 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1957521 - Posted: 27 Sep 2018, 23:10:47 UTC - in response to Message 1957517.  

. . That speaks volumes for the Strix card then eh? I guess there IS an advantage in running the top end version cards for a given GPU.
Stephen

If it will fit in your case....


. . Yes, those 3 fan units get very very long ...

Stephen

:)
ID: 1957521 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957524 - Posted: 27 Sep 2018, 23:15:58 UTC - in response to Message 1957519.  

I am only using coolbits on the 2nd card to control the fan speed


. . What script are you using for your overclocking and can you remember the command line you used to set coolbits?

Stephen

?


I am not using a script at the moment to set clocks, i only do it in nvidia-settings yet, i plan to set it all soon, and possibly use keepP2
I've used sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

(-a is a short for --enable-all-gpus)
ID: 1957524 · Report as offensive     Reply Quote
Stephen "Heretic" Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Sep 12
Posts: 5557
Credit: 192,787,363
RAC: 628
Australia
Message 1957526 - Posted: 27 Sep 2018, 23:32:07 UTC - in response to Message 1957524.  

. . What script are you using for your overclocking and can you remember the command line you used to set coolbits?

I am not using a script at the moment to set clocks, i only do it in nvidia-settings yet, i plan to set it all soon, and possibly use keepP2
I've used sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

(-a is a short for --enable-all-gpus)

. . Ca va! I have learned something new, I did not know about that -a shortcut :)

Stephen

:)
ID: 1957526 · Report as offensive     Reply Quote
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1957527 - Posted: 27 Sep 2018, 23:41:57 UTC - in response to Message 1957466.  
Last modified: 27 Sep 2018, 23:45:48 UTC

Can you see 2 monitors in X Server Settings?
You should be able to disable 1 there, or even set them to be clones. So you have only 1 screen.


No, when in the X server display configuration, i only see one screen i can move around..
Same thing if i go into KDE monitor settings, no trace of a second screen.

The only control i have over this "fake" non connected screen is in the nvidia x server display configuration, on the dropdown box
if i select X screen 1, i can select its color depth, and position/offset (absolute, right of, left of... above or below)
I am guessing that is the normal behavior even if nothing is connected.

If i use the --only-one-x-screen, my second gpu vanish from xconf, so coolbits won't be enabled on it

What does your xorg.conf have for the entry ServerLayout. It should be the first section in etc/x11/xorg.conf, just post the section here. My three GPU machine has this;
Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0" 0 0
    Screen      1  "Screen1" 1920 0
    Screen      2  "Screen2" RightOf "Screen1"
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"
    Option         "Xinerama" "0"
EndSection
I don't have any trouble with ghost screens, and coolbits works on all 3 GPUs.
ID: 1957527 · Report as offensive     Reply Quote
RickToTheMax

Send message
Joined: 22 May 99
Posts: 105
Credit: 7,958,297
RAC: 0
Canada
Message 1957529 - Posted: 27 Sep 2018, 23:57:50 UTC

Section "ServerLayout"
    Identifier     "Layout0"
    Screen      0  "Screen0" 0 0
    Screen      1  "Screen1" RightOf "Screen0"
    InputDevice    "Keyboard0" "CoreKeyboard"
    InputDevice    "Mouse0" "CorePointer"
EndSection


Ill add the xinerama 0 see if helps at all
ID: 1957529 · Report as offensive     Reply Quote
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1957531 - Posted: 28 Sep 2018, 0:05:00 UTC - in response to Message 1957529.  

You might want to add the screen width to screen 1 as you see in mine;
Screen      1  "Screen1" 1920 0

My screen is 1920 wide, so it knows where the screen ends.
ID: 1957531 · Report as offensive     Reply Quote
Previous · 1 . . . 77 · 78 · 79 · 80 · 81 · 82 · 83 . . . 162 · Next

Message boards : Number crunching : Setting up Linux to crunch CUDA90 and above for Windows users


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.