Message boards :
Number crunching :
Setting up Linux to crunch CUDA90 and above for Windows users
Message board moderation
Previous · 1 . . . 76 · 77 · 78 · 79 · 80 · 81 · 82 . . . 162 · Next
Author | Message |
---|---|
Sleepy Send message Joined: 21 May 99 Posts: 219 Credit: 98,947,784 RAC: 28,360 |
It can be done and works. Here: http://setiathome.berkeley.edu/forum_thread.php?id=81271&postid=1952070 You can skip the unplug/replug GPU step. But actually the most important part is the xorg.conf. You can just disable Nouveau drivers and install Nvidia even from device manager and cook the right .conf. For Intel, you may put "i915" as driver name (check on your system). Next step is overclocking. At the moment it does not work without Nvidia not running xorg. But I have ideas... Sleepy |
J. Mileski Send message Joined: 9 Jun 02 Posts: 632 Credit: 172,116,532 RAC: 572 |
It can be done and works. To be clear, my onboard graphics is an ASPEED chipset on the motherboard, not a gpu on cpu. I'm going to make it easy on myself and disable the onboard video when I install linux. |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
I'm going to make it easy on myself and disable the onboard video when I install linux.I think that is a good idea. The performance gain you gain from the special app will more than make up for the video usage ... as long as your card has enough memory. |
J. Mileski Send message Joined: 9 Jun 02 Posts: 632 Credit: 172,116,532 RAC: 572 |
|
Sleepy Send message Joined: 21 May 99 Posts: 219 Credit: 98,947,784 RAC: 28,360 |
It does not matter which onboard graphic you have. I assumed it was the Intel embedded in recent CPUs, but my suggestions are the same. You should just have to change the driver entry in xorg.conf In my experience, the problem was not what you loose or gain in Seti crunching, that was fine with my Nvdia running Xorg or not, no big difference. The difference was the usability of the PC. With Nvidia running Xorg, stuttering and lag were not acceptable for me while working. Disabling Seti while working would be overkill and probably really counterbalancing the advantages of the new application. In my current setting, I am really fine. Working and crunching do not interfere. Sleepy |
J. Mileski Send message Joined: 9 Jun 02 Posts: 632 Credit: 172,116,532 RAC: 572 |
I should be clear too, I only crunch seti on this box. I currently have 2 120GB SSD's in it, Windows is installed on one and linux will be installed on the other. Only time I'm on this computer is to see if it is still running, or if windows needs a reboot after an update. I have other computers for my internet needs. |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
I should be clear too, I only crunch seti on this box. I currently have 2 120GB SSD's in it, Windows is installed on one and linux will be installed on the other. Only time I'm on this computer is to see if it is still running, or if windows needs a reboot after an update. I have other computers for my internet needs. Many of us don't have an internal video card that can be run at the same time as discrete video card can. And most have not had any issues they have been posting about. Tom A proud member of the OFA (Old Farts Association). |
RickToTheMax Send message Joined: 22 May 99 Posts: 105 Credit: 7,958,297 RAC: 0 |
One of my host is not a dedicated cruncher, and sometimes i need to suspend/resume tasks... In the special app docs i can see this: "5) The App may give Incorrect results on a restarted task. One way to avoid restarted tasks is to set the checkpoint higher than the task's estimated run-time, and also avoid suspending/resuming a task." How do i set the checkpoint? is it the "Request tasks to checkpoint at most every X seconds" in Boinc Computing preference? If i set it higher than my task run-time, it would just discard the task if not finished, and when resuming it would just start over? and not generate incorrect result? Another problem i have since adding a 2nd card to my other host, i had to use the nvidia-xconfig --enable-all-gpus , and set coolbits to both device. But now it seems i have a ghost 2nd screen, when i move my mouse to the right side, it is like there is a 2nd monitor connected, my cursor vanish and when i move back to my screen the cursor is at a different position. (feels like that 2nd ghost screen is using a smaller resolution) |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
One of my host is not a dedicated cruncher, and sometimes i need to suspend/resume tasks... . . Hi Rick . . Yes! . . I can't remember if the default is 120, 180 or 300 seconds. I have used several different values there depending on the machine. But as long as you set it definitely higher than your normal run time you should have very few if any restarts. It does mean the running task will be discarded, but for one reason or another I have occassionally had a task restart and not had it crash so the problem may no longer be as severe as it had been. But it is still prudent to avoid restarts as much as possible. If I need to stop crunching at any time (with multiple GPUs) I check out the task status and wait for the most advanced task to end so I am only dumping one partially completed task. But that is me ... :) . . As for the screen issue it sounds to me like you have set the "output screen size" larger than your physical display can manage so it is a virtual screen you have to scroll around. Check out the maximum resolution your monitor can display and set the screen size in xorg to match it. It may be as simple as having a 4:3 monitor but selecting 16:9 screen size. Been there and done that :) Stephen :) |
RickToTheMax Send message Joined: 22 May 99 Posts: 105 Credit: 7,958,297 RAC: 0 |
Actually my working screen resolution is fine no problem there, the problem is, my system acts like there is a second display connected while there is none. Something is probably wrong with my xorg config |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Interesting symptom Rick. Zalster has been complaining of the exact same thing on a computer at work that he crunches with. Now I know it is not a one-off problem since you have it too. I think there is something wrong with your xorg.conf configuration too. Have you started from scratch with a --allow-empty-initial-configuration on nvidia-xconfig? Then follow up with a --no-separate-x-screens? Or a --only-one-x-screen? Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Kevin Olley Send message Joined: 3 Aug 99 Posts: 906 Credit: 261,085,289 RAC: 572 |
Just managed to make a mess of threadripper. Tried to update the Nvidia drivers and it has gone badly wrong. After I restarted it has failed to start - all I am getting is a blank screen with a message occasionally appear and disappear so fast that I cannot read it. I can get into the boot manager and the advanced options but cannot work out how to fix it. Kevin |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Just a wild guess since I was just dealing with this issue, but try moving your connected monitor to the other gpus in the system and see if the X server monitor output has moved to one of the other cards due to reenumeration of the BusID's in the xorg.conf file. The symptom I had was the last output on the screen was the enabling of my user id and then the output appears to freeze. I have removed the no splash in the kernel command line to see the boot process. But in reality that point is where the X server starts rendering the desktop and the output had moved to the bottom installed card. All I had to do was unplug the monitor from the top card and try the outputs of each other card until I found the normal desktop sitting there waiting for me. Then just had to figure out what the BusID of each card was with the nvidia X settings app and rewrite the xorg.conf file to properly identify the cards and get the desktop output back onto the top slot card which is the one I normally connect to the monitor. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
RickToTheMax Send message Joined: 22 May 99 Posts: 105 Credit: 7,958,297 RAC: 0 |
Thanks for the help! I will try your suggestions |
Kevin Olley Send message Joined: 3 Aug 99 Posts: 906 Credit: 261,085,289 RAC: 572 |
No, no signal on any other card, boot manager is on first card. Kevin |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
If you can boot to the recovery mode, then drop to the root console and enable read/write access to the drive that has the installation with a: mount -o rw,remount / then do a: apt-get remove --purge nvidia*followed up with a: apt-get autoremove Then do the installation again either with the nvidia .run file or with a: apt-get install nvidia-396or apt-get install nvidia-390 Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Kevin Olley Send message Joined: 3 Aug 99 Posts: 906 Credit: 261,085,289 RAC: 572 |
Ok will do Kevin |
Kevin Olley Send message Joined: 3 Aug 99 Posts: 906 Credit: 261,085,289 RAC: 572 |
Up and running again, Thanks Kevin |
Stephen "Heretic" Send message Joined: 20 Sep 12 Posts: 5557 Credit: 192,787,363 RAC: 628 |
If you can boot to the recovery mode, then drop to the root console and enable read/write access to the drive that has the installation with a: . . Hi Keith, . . Did I misunderstand TBar about using the recovery mode? I thought the reason for doing that was there is no video drivers loaded at that point and the install will run as a clean install so no need for stop lightdm or apt-get remove? Stephen ? ? |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
The drivers may not be loaded when in recovery mode, but they are still installed on the system you are wanting to update. |
©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.