Posts by Link

1) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 109200) Posted 9 days ago by Link Post: Given panic-mode means Boinc realises tasks can't be completed within deadline, preventing Panic mode occurring is the entire solution. Eliminating the reason for the panic mode is the entire solution, everything else is a workaround, which might fail as soon as something changes (new WU type, new project, whatever) or even before. The root cause reason for panic mode is holding too large an offline cache. The root cause for the panic mode is highy misconfugured client, too large cache is just a small part of it. This isn't a problem, because Adrian (in this case) said both projects are important to him. Than he should configure BOINC properly so it can coexist with Folding without any issues, currently it seems he doesn't really care if BOINC works properly or not.
2) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 109175) Posted 11 days ago by Link Post: Given panic-mode means Boinc realises tasks can't be completed within deadline, preventing Panic mode occurring is the entire solution. Eliminating the reason for the panic mode is the entire solution, everything else is a workaround, which might fail as soon as something changes (new WU type, new project, whatever) or even before. It may not be pretty in that processes are sharing cores, but imo no-one in their right mind cares which bit of a process of which task runs at what time as long as 1) the CPUs are being fully utilised and 2) tasks complete within deadline without further manual intervention. It's not just not pretty, highly overcommiting the system might slow down the overall production, in particular with hyperthreading CPUs many people leave 1-2 theads for non-BOINC stuff.
3) Message boards : Number crunching : More threads on VM tasks possible? (Message 109170) Posted 11 days ago by Link Post: You can limit the amount of VM tasks using app_config.xml and run other tasks on the other cores, something like: <app_config> <app> <name>Application_Name</name> <max_concurrent>1</max_concurrent> </app> </app_config> Application_Name you will find in your client_state.xml.
4) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 109036) Posted 25 Mar 2024 by Link Post: I've been very critical of WCG privately, but it seems that if it wasn't taken on by the current people it would have disappeared altogether, so while not being happy at all with it, and very frustrated, even a badly running WCG is better than no WCG at all Sure, it's better, but still very frustrating and I don't see how they could possibly offer any new projects in the foreseeable future, their setup is breaking down with MCM running alone and what's worse, they don't know, why something is breaking and how to get it running again. I mean, they have now been working two weeks (!) on getting MCM work generators to work properly again and they still didn't succeed and the plans on how to resolve the issue posted by TigerLily read like "we don't know why something, that worked for several weeks, suddenly stopped working but we will be poking in the dark and hopefully hit the right thing".
5) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 109031) Posted 25 Mar 2024 by Link Post: Trouble is that none of them have any work to supply and my machine is running only 4 tasks. 2 Einstein and 2 Rosetta. Einstein has lots of work, both FGRP5 and BRP4G, and should easily use all your cores. WCG should have lots of MCM1 work and some OPNG work sometimes (the other projects are suspended), but they simply do not have staff capable of running a BOINC project properly, the reliability of their cloud environment is disastrous and the communication with the so called "tech team" is painfully slow. The fact, that they got highly fu... modified BOINC environment from IBM propably doesn't help either. There are however far more projects than those four you mentioned doing a real science, so perhaps you should try to find one or two more as basically you have only one reliable project in your list (Einstein). Milkyway, which you have crunched for in the past, is doing pretty well recently if you don't mind multithreaded WUs.
6) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 109029) Posted 25 Mar 2024 by Link Post: 0.1 and 0.5 results in the cache running down to less than 0.1 days of work, before then refilling itself to 0.1 plus the additional 0.5 days worth. The amount of work in the cache will vary between 0.6 days & less than 0.1. Well, that depends on how long the WUs are running, so basically on the projects you run. Current versions of BOINC report completed tasks latest one hour after uploading the result files, so after about 0.042 days. So if you run a project with short tasks (<1h), with this cache settings BOINC will contact the project servers about once an hour to report the completed tasks and will piggyback a request for new tasks on that scheduler request. So (assuming a project with steady work supply) your chache will be somewhere between around 0.55 and a bit (max. the length of one WU) over 0.6 days. With modern CPUs with many cores (i.e. 8+), even if running longer tasks, there's nearly always a task, that has been finshed in the past hour and that's where the issues start when running more than 1 or 2 projects simultaneously, in particular if the run times on their WUs are different enough. The project with shortest tasks will always have something to report and will always fill up the cache when reporting, projects with longer tasks will have simply less chances to ask for new work. So over time the project with shortest tasks might push out all other projects, in particular if they don't have an absolutely steady WU supply. That's why Richard could never get automatically any work for Rosetta, his 0 days cache was filled with up to 3 days of work from other projects, so why should BOINC ask for work from Rosetta? So if he wants to run 4 or more projects simultaneously, which is in general nothing wrong, I think the best results he will get with 0 additional days (once everything settles down), as that eliminates any piggybacking of requests for new work on scheduler requests to report completed tasks, so BOINC will always make separate request for work and when doing that, choose exactly the project that should get more work to respect the ressource setttings.
7) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 109026) Posted 23 Mar 2024 by Link Post: I have a question. I have cut down to four projects. Suppose I want to swap out a project and add a different project. How do I be sure to not let a new project screw things up? By keeping the WU cache small. There were already some suggestions, my (for your case) would be something in the range of "Store at least 0.05-0.15 days of work" and "Store up to an additional 0 days of work". Start low and increase by 0.01-0.02 if needed. Setting the last one to 0 is important IMHO as it prevents a project reporting a task to request new work on that scheduler request and fill up the cache even if another project should actually request new work to respect your ressource share settings. Also you don't need to remove any projects in the future if the client fetches too much work for them "by accident", just set them to "no new tasks" to fix the issue.
8) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 108971) Posted 10 Mar 2024 by Link Post: folk who cannot deal with all of the complexities of BOINC There's nothing complex about changing some numbers or ticking/unticking some boxes and the adjustments suggested by Grant (SSSF) could have been done in half the time needed to write your post. That's how you configure any software including your web browser. No one told you to modify or write some XML files. You have also managed to set up BAM! (which is btw. completely unnecessary to manage one single computer except for some very special cases), so apparently you can handle some complexity. I attached to a number of projects using the default configurations to keep things simple. After installation I did not touch the configuration, but I am told that somehow there were many changes to the projects. I have no idea how that happened. I can't tell you how that happened, but I can tell from the log where and when it happened: on GPUGRID, 16-Feb-2024 23:38:58, so not really long ago. No idea if you did some changes on other projects before, but that's where you last time updated or at least saved the settings. Not in the client itself, since it's using global and not local preferences. I am also participating in Folding@home. With this set of projects there is absolutely noting to do after the install. It just runs, even if the browser is closed. Sure, it's just one project, of course it can be done a lot simpler than a client made to handle dozens of projects simultaneously, some of them using virtual box, some using wrapper applications and some just running their applications and all of it according to the wishes of the user. Like the old SETI client, that was also just running.
9) Message boards : Number crunching : Why no work on my computer since Feb 17, 2024 (Message 108969) Posted 10 Mar 2024 by Link Post: 3/9/2024 7:57:31 AM \| \| - Use at most 50% of the CPU time Should be "Use at most 100 % of CPU time" (for a desktop, current laptops with their limited cooling may need that set to 90% or less in order to prevent thermal throttling). Considering he's running Milkyway: no, must be 100% or Milkyway's N-Body might get stuck, that might be the explanation for the task which is past the deadline. So use less cores if the system cooling is insufficient. @Richard Mitnick: If you are not going to change anything, what's the point of asking for help? You changed the not so good default settings to complete BS and when people tell you what you need to change to get it to work properly again, you refuse to do it. Your cache settings prevent any project, from which you have no tasks running from getting any new work, because your 0 days cache is filled up with additional 3 days everytime BOINC is reporting a completed task. That's why you got new work when you reset the project, because that forced BOINC to contact the servers and ask for minimal amount of work. Otherwise it only contacts servers to report completed tasks and any requests for new work are added to those by the "additional 3 days". No dedicated scheduler requests for new work.
10) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 108843) Posted 13 Feb 2024 by Link Post: Contributed decades ago (all Windows), recently trying to get back into it (Linux.) Unfortunately, BOINC has forgotten all of my past contributions; guess that's my fault for letting it pasture. No, it's just your fault to not login with your old email adress and password. BOINC does not forget anything, perhaps except on WCG, where everything gets deleted 3 years after your last activity (does not have to be crunching, logging in is enough).
11) Message boards : Number crunching : New jobs in Rosetta Python projects (Message 108800) Posted 23 Jan 2024 by Link Post: I know Rosetta hasn't ever done this but these new Python tasks take 7.45gb of memory for EACH task on my Windows pc's, i have pc's with anything from 6 to 32 cpu cores but not a one of them has enough memory to fill every cpu core with a Python task and not lock up the pc to the point only a reboot and reducing the physical number of tasks I can run at Rosetta. Normally it's not a problem with anything EXCEPT the Python tasks Than limit the amount of concurrent python tasks using an app_confil.xml. This can be even different for each of your computers, depending on the amount of memory.
12) Message boards : Rosetta@home Science : Unable to get any work (Message 108751) Posted 7 Dec 2023 by Link Post: Asteroids also has cpu and gpu tasks but I don't know if they can use your Intel Arc gpu or not No, no IntelGPU app on Asteroids. But SRBase has one, apparently special one only for the Arc GPUs, not all Intel GPUs.
13) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 108718) Posted 24 Nov 2023 by Link Post: One problem with dark pages is that I need to have a light on in the room to see the keyboard to type (or anyone else as crap as me that has to look at it) I'd never sit in a dark room and read something on light pages, definitely not good for eyes. For office type of work I like office type of light, i.e. lots of light. Like the "Aug2001 vintage 17" CRT monitor" I have just sold on ebay for £80 I was more thinking about something like OLED or LCD with background light, which adjusts itself to what's needed.
14) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 108712) Posted 22 Nov 2023 by Link Post: Or does it use less electric to have dark webpages ? On some types of screens, yes. And it's a lot better for the eyes, some people even use add-ons like DarkReader to have all pages dark. Some pages are automatically light or dark, depanding on your OS or browser settings.
15) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 108710) Posted 21 Nov 2023 by Link Post: I don't even try to get work from here, there are more than enough computers doing the available work for Rosetta, so my don't need to "steal" WUs from those, which might not be attached to any other projects and will sit idle if they don't get anything from here. The job get done, that's all what matters, not whose computers do it.
16) Message boards : Number crunching : 2023 BOINC Census is here, please take 5 minutes to fill it out and make your voice heard! (Message 108706) Posted 18 Nov 2023 by Link Post: Done. :-)
17) Message boards : Number crunching : Rosetta Beta 6.00 (Message 108608) Posted 22 Sep 2023 by Link Post: 0 should mean zero. -1 should mean no limit. I agree, either that or a dedicated tags in the app_config.xml file, one to disable the entire app and one to disable specific app version: <app_config> [<app> <name>Application_Name</name> <max_concurrent>1</max_concurrent> [<report_results_immediately/>] [<fraction_done_exact/>] [<disable_app/>] <gpu_versions> <gpu_usage>.5</gpu_usage> <cpu_usage>.4</cpu_usage> </gpu_versions> </app>] ... [<app_version> <app_name>Application_Name</app_name> [<plan_class>mt</plan_class>] [<avg_ncpus>x</avg_ncpus>] [<ngpus>x</ngpus>] [<cmdline>--nthreads 7</cmdline>] [<disable_app_version/>] </app_version>] ... [<project_max_concurrent>N</project_max_concurrent>] [<report_results_immediately/>] </app_config> This would help also in many other cases, like for example to disable the CUDA app on not ancient Nvidia cards on Moo! or in general disable inefficient or problematic app versions without that the servers retest them every now and than.
18) Message boards : Number crunching : Wu's ignoring my target time. (Message 108579) Posted 9 Sep 2023 by Link Post: It would still be running at full speed, but if "Use at most xxx % of CPU time" is set to something other than 100%, then the CPU will be running at 100% speed, but the processing will run for a bit, then stop, run for a bit, then stop, run for a bit, then stop, run for a bit, then stop... Running not at 100% of CPU time however does not lead to such difference between run time and CPU time, BOINC stops the counter for run time when it tells the app to pause. I was trying that too on my laptop and the result was actually a CPU time, which was higher than the run time, there was apparently some small delay between BOINC telling the app to pause and pausing the run time counter and the app actually pausing, so it was always running always a bit without the run time counter.
19) Message boards : Number crunching : Wu's ignoring my target time. (Message 108574) Posted 8 Sep 2023 by Link Post: I think it displays everything correctly... on an undercommited 8-core system Rosetta should be using 12.5% of your CPU and FAH probably 25% (2 cores), both get a lot less than that. I recall reading something about some kind of annoying energy saving crap in Windows 11 for applications that use lot of CPU time, but you'll have to google something like that yourself, I don't find it anymore and I'm still on Win10, so no experience with it. Is your CPU even running at full speed? It should be constantly at at least 3.6GHz, boost up to 3.9GHz. The only time I've seen BOINC applications not using a full core was when I limited the CPU speed on my laptop using Windows energy options for it, than I've seen that type of values in task manager as apparently it didn't understand, that that is the current max speed and they are using actually 100% of a core. So my 1st guess is your CPU is running way below it's max speed.
20) Message boards : Number crunching : Wu's ignoring my target time. (Message 108568) Posted 7 Sep 2023 by Link Post: Windows 11 has a bug in which it misreports cpu time. For example task is running for 12:36:00 but cpu time is 7:20:00 The first is the run time, the 2nd the CPU time (and that's the target, not the run time, even if it's written like that in the preferences). Nothing different from other Windows version, but perhaps you should check what is using so much CPU time on your computer and if it might be a good idea to leave an empty core or two for it.

Next 20