Message boards :
Number crunching :
Unable to get any cpu work anymore on ARM64
Message board moderation
Author | Message |
---|---|
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
I just realized I haven't had any cpu work from the project for a week now or more. I ask for work but never get any. The other strange thing is that the cpu app goes missing every time I restart BOINC. I put it back in and it stays there until BOINC is restarted. So why is the scheduler deleting the app? Has the app been deprecated? The host is my Nvidia Nano. https://setiathome.berkeley.edu/show_host_detail.php?hostid=8707387 The appname is setiathome_8.02_aarch64-unknown-linux-gnu and is correctly defined in the app_info. <app_info> <app> <name>setiathome_v8</name> </app> <file_info> <name>setiathome_8.02_aarch64-unknown-linux-gnu</name> <executable/> </file_info> <app_version> <app_name>setiathome_v8</app_name> <platform>aarch64-unknown-linux-gnu</platform> <version_num>802</version_num> <file_ref> <file_name>setiathome_8.02_aarch64-unknown-linux-gnu</file_name> <main_program/> </file_ref> </app_version> <app> <name>setiathome_v8</name> </app> <file_info> <name>setiathome_x41zc_aarch64-unknown-linux-gnu_cuda100</name> <executable/> </file_info> <app_version> <app_name>setiathome_v8</app_name> <platform>aarch64-unknown-linux-gnu</platform> <version_num>987</version_num> <cmdline>-nobs</cmdline> <coproc> <type>NVIDIA</type> <count>1</count> </coproc> <avg_ncpus>1.0</avg_ncpus> <max_ncpus>1.0</max_ncpus> <file_ref> <file_name>setiathome_x41zc_aarch64-unknown-linux-gnu_cuda100</file_name> <main_program/> </file_ref> </app_version> </app_info> I also request seconds of cpu work but never get any. Get regular gpu work just fine. 13-Oct-2019 10:40:43 [SETI@home] Sending scheduler request: Requested by user. 13-Oct-2019 10:40:43 [SETI@home] Requesting new tasks for CPU and NVIDIA GPU 13-Oct-2019 10:40:43 [SETI@home] [sched_op] CPU work request: 38016.00 seconds; 4.00 devices 13-Oct-2019 10:40:43 [SETI@home] [sched_op] NVIDIA GPU work request: 89.76 seconds; 0.00 devices 13-Oct-2019 10:40:44 [SETI@home] Scheduler request completed: got 1 new tasks 13-Oct-2019 10:40:44 [SETI@home] [sched_op] Server version 709 13-Oct-2019 10:40:44 [SETI@home] Project requested delay of 303 seconds 13-Oct-2019 10:40:44 [SETI@home] [sched_op] estimated total CPU task duration: 0 seconds 13-Oct-2019 10:40:44 [SETI@home] [sched_op] estimated total NVIDIA GPU task duration: 2959 seconds Anyone have any information about this app? Why do I never get cpu work anymore? Why does the scheduler delete the app when restarted? Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Tom M Send message Joined: 28 Nov 02 Posts: 5124 Credit: 276,046,078 RAC: 462 |
Here is a shot in the dark. Assume there is something garbled in the string that names the app that keeps going missing in the app_info.xml. And it might not be a displayable character. Do a clean, hand re-typing of the file name in another file. Then copy and paste over top of every place it is showing up. A more extreme step would be re-type the entire file in another file, by hand. Like I said, a shot in the dark. Sometimes I can't see the problem but when I replace it with "clean code" from a known working file, the problem goes away... I suppose it might qualify as voodoo tech support? Tom A proud member of the OFA (Old Farts Association). |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
The app_info.xml passes the XML validator which is good for finding extraneous characters that are not visible. I think the app has been deprecated and no one told anybody. Until somebody else running the app tells me they have no issues, that is my gut feeling. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
The extreme case would be to reset the host and see if the scheduler resends the app and work and it functions correctly. Lightbulb moment. I just realized I can create a new host using the WOW contest bunkering protocol to make a new host and run it stock and prove one way or the other if the app is sent and work is processed. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Gary Charpentier Send message Joined: 25 Dec 00 Posts: 30608 Credit: 53,134,872 RAC: 32 |
I just realized I haven't had any cpu work from the project for a week now or more. I ask for work but never get any. The other strange thing is that the cpu app goes missing every time I restart BOINC. I put it back in and it stays there until BOINC is restarted. So why is the scheduler deleting the app? Has the app been deprecated? Maybe an old app. My Pi https://setiathome.berkeley.edu/show_host_detail.php?hostid=8173315 had SETI@home v8 8.02 arm-unknown-linux-gnueabihf , but now runs SETI@home v8 8.06 arm-unknown-linux-gnueabihf My guess is there is no 8.02 ARM work being made and only 8.06 ARM work. You may need to go stock so BOINC will download the new CPU app. Then you can put back your App Info pointing to the new app. Or you could sign up for Beta and let it download the latest and greatest and then copy it to main and point the App Info at it. |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
No I have the correct app for ARMv8. My Nano is not a Pi. It is a 64 bit machine. My Pi runs the 8.06 app you listed. The 8.02 app is the only ARM64 bit app for Linux. I just remembered that CyborgSam runs only the cpu app now. He is getting work and processing it fine right now. So the app is still valid and working. Now I just have to figure out why my host is not working correctly on the app. https://setiathome.berkeley.edu/show_host_detail.php?hostid=8719681 Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
Well I tried resetting the project and that did nothing. I didn't think through my idea of using the AIO. Wrong platform. Would have to install the standard service mode BOINC to test the theory. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Jord Send message Joined: 9 Jun 99 Posts: 15184 Credit: 4,362,181 RAC: 3 |
Do know that when you run an app_info.xml and reset the project without removing that file, that it stays and survives the project reset. If you manually removed it prior to resetting, ignore me. 😠|
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
Do know that when you run an app_info.xml and reset the project without removing that file, that it stays and survives the project reset. Yes, I forgot that little bit of info that is in the docs. I will remove the app_info's and the gpu apps so I can preserve them. I don't want to have to compile them all over again. I think I am going to have to erase the distro and reinstall it to sort things out. But then I will lose all of my work setting up the correct compile environment. The gpu apps work for both Einstein and Seti. And of course they are a lot more efficient than the cpu apps. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13161 Credit: 1,160,866,277 RAC: 1,873 |
OK, I have finally sorted things out. Don't know what was wrong. Renamed app_info so it wouldn't get picked up and started the client without it. It said the existing cpu app was already present so it skipped downloading and then downloaded 3 cpu tasks and started running two of them along with a Einstein gpu app. So I knew that the existing app was fine and nothing wrong with it. Then I re-copied the app_config and and app_info and restarted the client with those in place and it continues to run the two cpu tasks and the Einstein gpu task. I have 3 of 4 cpus limited with a full cpu core to support a gpu task. So something glitched somewhere along the way. I made things worse with the attempted BOINC reinstall so now I can't launch the client by launching the Manager anymore. Now I have to manually start both the client and Manager in their own terminals with an alternate port address. Normally I have been just starting the client with the Manager, doing any configurations necessary and then stopping the Manager and leaving the client running since the Nano is run headless and accessed by RPC over the network via BoincTasks. So just a little more work to get things running than before. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
©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.