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 . . . 153 · 154 · 155 · 156 · 157 · 158 · 159 . . . 162 · Next

AuthorMessage
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36774
Credit: 261,360,520
RAC: 489
Australia
Message 2032627 - Posted: 16 Feb 2020, 5:37:29 UTC

This is reminding of the days when someone running a Mac complained about not being sent enough AP work to keep his rig going because MB work didn't run on it at the time. :-(

Cheers.
ID: 2032627 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2032629 - Posted: 16 Feb 2020, 5:48:29 UTC - in response to Message 2032626.  

well if its occasionally dropping GPUs and randomly needing restarts, i'd say it's got some instability going on somewhere, especially compared to comparable systems that only reboot 3 or 4 times a year. so what do you think is the reason mine is more stable? i'd love to hear your thoughts since you keep dodging the question.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2032629 · 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 2032630 - Posted: 16 Feb 2020, 5:50:44 UTC - in response to Message 2032627.  

I fixed that...didn't I? Now the Mac runs the CUDA Special App, and Windows doesn't.
In fact, the Mac has now outscored All Windows machines at SETI, nice, https://www.boincstats.com/stats/0/host/list/0/
ID: 2032630 · 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 2032633 - Posted: 16 Feb 2020, 6:00:19 UTC - in response to Message 2032629.  

It's a simple matter of How Much money you're willing to spend, and I have mentioned that before. I'd say a number of people around here spent less on their House than you've spent on computers.
ID: 2032633 · Report as offensive     Reply Quote
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36774
Credit: 261,360,520
RAC: 489
Australia
Message 2032634 - Posted: 16 Feb 2020, 6:00:31 UTC

But it took so much time to convince you about things back back then too. :-(

Please just consider the fact that something may not likely be quite right with your system instead of carrying on like this. Go back to a basic setup if necessary with your rig first to rule out every other option. ;-)

Cheers.
ID: 2032634 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2032635 - Posted: 16 Feb 2020, 6:00:38 UTC - in response to Message 2032513.  

I've had the same problem with the machines using the Mining cable/setup, the normal machines don't have this problem. After nearly two years I've concluded the Mining setup just isn't as stable as it should be. You can move GPUs/Cables around and get it working nicely, but eventually a GPU will drop out at some point.


you called it unstable. your words, not mine.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2032635 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2032636 - Posted: 16 Feb 2020, 6:09:41 UTC - in response to Message 2032633.  
Last modified: 16 Feb 2020, 6:21:03 UTC

It's a simple matter of How Much money you're willing to spend, and I have mentioned that before. I'd say a number of people around here spent less on their House than you've spent on computers.


LOL. this is hobby money bro. i've actually spent a lot less than some other folks here, and certainly not house territory, unless you know a lot of people spending 10-15k on a house. not that anyone's finances have any relevance to anything. people can spend what they like without vilification.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2032636 · 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 2032638 - Posted: 16 Feb 2020, 6:26:11 UTC - in response to Message 2032634.  
Last modified: 16 Feb 2020, 6:37:34 UTC

But it took so much time to convince you about things back back then too. :-(

Please just consider the fact that something may not likely be quite right with your system instead of carrying on like this. Go back to a basic setup if necessary with your rig first to rule out every other option. ;-)

Cheers.
Which system are you referring to? No one has been able to get a Mining board to work better in Linux. Petri agrees there is a conflict with the Video driver on the Mac, the problem is, the Same conflict exists with a number of Linux machines. I have provided examples, I can provide more examples if necessary, they pop up in the computer lists often.
Oh, nVidia never responded to the Bug report on the Driver. But, the fact you can see the same bug in Linux makes it moot, Conflict with macOS 10.12 & 10.13 Web Drivers and SETI@Home CUDA App?
ID: 2032638 · Report as offensive     Reply Quote
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 36774
Credit: 261,360,520
RAC: 489
Australia
Message 2032639 - Posted: 16 Feb 2020, 6:43:31 UTC - in response to Message 2032638.  

But it took so much time to convince you about things back back then too. :-(

Please just consider the fact that something may not likely be quite right with your system instead of carrying on like this. Go back to a basic setup if necessary with your rig first to rule out every other option. ;-)

Cheers.
Which system are you referring to? No one has been able to get a Mining board to work better in Linux. Petri agrees there is a conflict with the Video driver on the Mac, the problem is, the Same conflict exists with a number of Linux machines. I have provided examples, I can provide more examples if necessary, they pop up in the computer lists often.
Oh, nVidia never responded to the Bug report on the Driver. But, the fact you can see the same bug in Linux makes it moot.
Arn't they just the same systems just dual booted?

Please stop trying to be evasive now (we can enough of that elsewhere) and do some actual fault finding at home on whichever rig/s is giving you these problems instead of your, "I'm right and everyone else is wrong", attitude that rings across to most here (and gets annoying, just like way back then).

I'll be 1 of the 1st to admit that you've come a long way and your work is much appreciated, but at times you just fall back into that bad attitude and won't listen to others.

I've said enough now.

Cheers.
ID: 2032639 · 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 2032641 - Posted: 16 Feb 2020, 6:53:00 UTC - in response to Message 2032639.  

Again which system are you referring to? I have two machines running Mining boards in Linux, and 2 machines running macOS. The other machines just run from time to time, the one with just 2 GPUs main purpose is to run a sound card loud enough my deaf 90 year old Dad can hear TV.
ID: 2032641 · 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 2032644 - Posted: 16 Feb 2020, 7:12:32 UTC - in response to Message 2032643.  
Last modified: 16 Feb 2020, 7:34:51 UTC

Yes, and the developer agrees there is conflict with the Driver, has agreed for over a Year.
Again from 12/28/2018, The developer claims this is due to a Bug with the macOS display driver and cannot be fixed.
The difference is, the Same bug can be seen often now in Linux.

Oh, there's also the New fact that the problem doesn't happen with a 1080Ti in the same machine.
ID: 2032644 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032649 - Posted: 16 Feb 2020, 10:13:39 UTC - in response to Message 2032626.  

Richard is presently running setiathome v8 enhanced x41p_V0.98b1, Cuda 10.1 special right now, https://setiathome.berkeley.edu/result.php?resultid=8550528263
I'd say He needs to explain just what he was talking about considering He and many others aren't having any trouble with it.
I'm also running a copy of

---------------------------------------------------------
SETI@home v8 enhanced x41p_V0.99b1p3, CUDA 10.2 special
-------------------------------------------------------------------------
Modifications done by petri33, Mutex by Oddbjornik. Compiled by Ian (^_^)
-------------------------------------------------------------------------
Oh, I see. Someone tried running the App as Stock and didn't include the API number in the app_config. That's why the SETI Server includes the API in the Stock configuration. Nothing other than someone didn't know what they were doing. Pull the API from the Apps running on the SETI Server and you will see chaos.
Again, a case of YOU NOT knowing what you are talking about.
The final line in that post, "It's built with API_VERSION_7.5.0 - which is old, but should be good enough."
It old, like the CUDA Code is much older.
Yes, I goofed. That was my very first experience of running Linux, and I was working it out as I went along. I probably used the modern API value from a more modern stock app - leaving the API line out will default to the ancient way of doing things. I'm pretty experienced in writing app_info.xml files, because there are dozens of them in the Lunatics Installer: once I saw the 'all tasks on one GPU' problem, it didn't take long to work out that the problem was very old code in a modern application (even though I had to source a hex editor for Linux to be certain). Since I'm spending more of my time working on improving BOINC through code these days, I'd simply forgotten that other people approach problems from a different perspective.

I've mentioned another possible problem with API 7.5.0 recently - the task which was killed 5 minutes after writing boinc_finish. There's not much point us fixing these things if you won't use them.
ID: 2032649 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032652 - Posted: 16 Feb 2020, 11:07:27 UTC - in response to Message 2032644.  

Yes, and the developer agrees there is conflict with the Driver, has agreed for over a Year.
Again from 12/28/2018, The developer claims this is due to a Bug with the macOS display driver and cannot be fixed.
The difference is, the Same bug can be seen often now in Linux.
Can you link to the actual developer statement, please? All I can see is a conversation between two SETI users.
ID: 2032652 · 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 2032663 - Posted: 16 Feb 2020, 12:42:15 UTC - in response to Message 2032652.  

As far as I know nVidia never responded. All I get when trying TimeLord04's link is page not found.
The developer I was referring to is Petri, he thinks it's Mac only...but the same thing happens in Linux.
ID: 2032663 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032664 - Posted: 16 Feb 2020, 12:45:50 UTC - in response to Message 2032663.  

All I get when trying TimeLord04's link is page not found.
Ah. As in?

Membership Required
You are not authorized to access this page.
ID: 2032664 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2032685 - Posted: 16 Feb 2020, 16:20:30 UTC - in response to Message 2032649.  
Last modified: 16 Feb 2020, 16:22:16 UTC


I've mentioned another possible problem with API 7.5.0 recently - the task which was killed 5 minutes after writing boinc_finish. There's not much point us fixing these things if you won't use them.


I thought the same thing, and thought I remembered you mentioning that, couldn’t find that post in the moment.

Also seems to be the issue Jim highlighted in his first example, which was the basis for my recommendation to try a different app compiled on a newer build.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2032685 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032688 - Posted: 16 Feb 2020, 16:41:08 UTC - in response to Message 2032685.  

Yesterday, message 2032508 - replying to Jimbocus.
ID: 2032688 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032691 - Posted: 16 Feb 2020, 17:02:10 UTC

Going back to the API_version question, I think the question may simply be one of consistency.

Looking at my machines now, they are running:

A current (v7.16) version of BOINC
Ian's SETI build
An app_info which doesn't mention API_version

I originally downloaded the All-In-One package:

An older version of BOINC
TBar's SETI build against API v7.5.0
An app_info which also doesn't mention API_version

I'm assuming (without testing further) that each grouping is internally self-consistent, but you can't mix'n'match.

My problem probably arose when I upgraded BOINC, kept TBar's SETI build, but didn't add API_version to app_info.xml at that point.
ID: 2032691 · Report as offensive     Reply Quote
Ian&Steve C.
Avatar

Send message
Joined: 28 Sep 99
Posts: 4267
Credit: 1,282,604,591
RAC: 6,640
United States
Message 2032692 - Posted: 16 Feb 2020, 17:07:55 UTC - in response to Message 2032691.  
Last modified: 16 Feb 2020, 17:12:27 UTC

When you say it doesn’t mention API version, where are you looking? Do you want it in a readme file or in the file name or something?
Edit - Sorry disregard I misread what you had wrote.

You had previously stated that my compile was 7.15.0, so I assumed you had some other way to determine it.

I’m not sure why it matters in reference to the app_config or app_info anyway. I never found it necessary to add that info into the file.
Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

ID: 2032692 · Report as offensive     Reply Quote
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 2032693 - Posted: 16 Feb 2020, 17:12:07 UTC - in response to Message 2032692.  
Last modified: 16 Feb 2020, 17:23:38 UTC

I’m not sure why it matters in reference to the app_config anyway. I never found it necessary to add that info into the file.
No, doesn't go there - it's specified for app_info. Long story, relating to gridcoin - I'll dig it out.

Edit - worse than that, Bitcoin Utopia. The old way for BOINC to tell the science app which GPU to use was a device number on the command line. It didn't distinguish very well between CAL / CUDA / OpenCL, so a new init_data.xml structure was created to convey all that (and much more). The two systems ran side-by-side for some years.

But Bitcoin Utopia couldn't handle the unexpected command line (??!!), so the already outdated command line method was eventually abolished. Perhaps it only matters if BOINC and the science app are on opposite sides of the v7.5.0 boundary.
ID: 2032693 · Report as offensive     Reply Quote
Previous · 1 . . . 153 · 154 · 155 · 156 · 157 · 158 · 159 . . . 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.