Posts by pututu

1) Message boards : News : Target # 23: Ebola GP1 (Message 2287)
Posted 11 Jul 2024 by pututu
Post:
Edit 2: If gpu application requires a lot of coding resources perhaps explore the multi-threaded application within cpu. It's quite common nowadays such as again in folding@home where the project offers cpu multi-threaded application (you can specify how many cores/threads that you want to run), in the area of astrophysics such as milkyway@home or in mathematics such as primegrid.com & amicable_numbers@home. Just throwing out ideas here.
2) Message boards : News : Target # 23: Ebola GP1 (Message 2286)
Posted 11 Jul 2024 by pututu
Post:
We will employ high-resolution structures to conduct virtual screening experiments coupled to molecular dynamics simulations to ultimately identify potential GP inhibitors/modulators.


Does the portion of the molecular dynamics simulations use GPU or CPU or combination of both? From what I understand in another molecular dynamics simulation of protein folding project (e.g. folding@home), it has both cpu and gpu applications.

If the task has a long run time using cpu, crunching with gpu will speed up the computation time. Current target #22 tasks already taken a very long time already. If target #23 deals with small atoms molecules size, cpu will probably be ok (again at least from what I understand in folding@home).

Edit: World community grid (WCG) open pandemics project used to have both cpu and gpu applications but later dropped the cpu application because gpu can do more work much faster.
3) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2254)
Posted 21 Apr 2024 by pututu
Post:
with the setting above, my 8-core computer just got 28 long sidock tasks while there was a limit to cache 0.8 days :(

i'm going to set this project no more work until admin would implement a max jobs limit setting to avoid losing the work in other projects

UPDATE: i realized that this was a linuz computer without boinc 8.0.0 . For now i set "nomorework" on it for Sidock, and will experiment later

The other computer with boinc 8.0.0 still crunches through 2 sidock tasks and i set the app_config setting to keep it 2. I will report once it will finish crunching and get new work


What boinc version did you install for the PC with more work cache that you specified? I would suggest letting the PC with version 8.0.0 runs for a while to see if the prior bug is fully resolved. I'm still using a range of boinc 7.x versions but I use multiple boinc instances to control work cache if I decide to run more than one boinc project per PC like you are currently doing.
4) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2250)
Posted 19 Apr 2024 by pututu
Post:
Note that the <max_concurrent> tag in the app_config.xml file sets the maximum number of tasks of a specific application to run at a given time. It does not control the work cache size, meaning if you set the max_concurrent to "2" say, it will download tasks to fill whatever cache size and the cpu thread to use that you specified. It will only run 2 tasks irrespective of your cache size. However there is a bug in the boinc manager (not sure what version that is) that could potentially load more tasks that you need and override the cache size. Just fyi.


The bug has been fixed in the latest versions of the Boinc Client. Even before that, I have been running max_concurrent with a small cache size for several years without a problem - each project (including SiDock and CPDN) working according to their project weighting. (If anything it’s WCG that overfills the cache with respect to its project weighting).


Last year I had PCs running older boinc version and they still have the bug, similar to the discussion here with version 7.16: https://github.com/BOINC/boinc/issues/4322.
Have not tested latest version 8.0. Edit: looks like they have fix the bug according to the github post above.

Anyway one of OP main concerns is to limit the number of downloaded tasks other than running a fix number of tasks per project.
5) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2246)
Posted 19 Apr 2024 by pututu
Post:
Agree. That would be the easiest approach. Not sure if this will be the project admin top priority or working on the research project. Most of us here look for alternative solution such as running multiple boinc clients. BTW, this is commonly used during boinc competition.
6) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2244)
Posted 19 Apr 2024 by pututu
Post:
Note that the <max_concurrent> tag in the app_config.xml file sets the maximum number of tasks of a specific application to run at a given time. It does not control the work cache size, meaning if you set the max_concurrent to "2" say, it will download tasks to fill whatever cache size and the cpu thread to use that you specified. It will only run 2 tasks irrespective of your cache size. However there is a bug in the boinc manager (not sure what version that is) that could potentially load more tasks that you need and override the cache size. Just fyi.
7) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2242)
Posted 18 Apr 2024 by pututu
Post:
If you have multiple active projects, one way to control the number of tasks per project is to setup multiple boinc clients (aka boinc instances). See the guide here on how to set it up: https://www.overclock.net/threads/guide-setting-up-multiple-boinc-instances.1628924/. For each boinc clients or boinc instances, you only run one active projects. For each clients or instances, you can set up your own "Computing Preferences...".

The screenshot below shows that I'm running two boinc instances. One is dedicated for Denis@home and the other is Latinsquares. Both running on a single computure. You have more control over each project like how many days of work you want to fill your work cache and how much cores/threads you want to run for each project. Note that for Denis I'm running 3 tasks and for Latinsquares, I'm running 4 tasks. So there are a total of 7 tasks running on that single computer. Of course optimally, you don't want to run more tasks than your total available cpu cores/threads. If you have 8 threads, then best to run a total of 8 tasks or less in total on two separate boinc instances as shown.

If Denis runs out of work for example, then instead of letting that Denis boinc instance stays idle, you can add Latinsquares as a backup project but you need to set the Latinsquare resource share from 100% to 0% in Latinsquare account preferences (you need to log into Latinsquare website to do this). Since Latinsquare has work available almost 24/7, setting resource share to 0 means there will be 0 day of work cache stored in both boinc instances but that's ok.

8) Message boards : Number crunching : Is it possible to turn on the setting "Max # of simultaneous tasks" ? (Message 2239)
Posted 18 Apr 2024 by pututu
Post:
Set "Store at least ___ days of work" to 0.00 or 0.01 in your "computing preferences..." under "Options" from the boinc manager window. Also make sure that "Store up to an additional ___ days of works" is set to 0.

This will ensure that if you are running 100% of your cpus (say 8 threads as in the screen shot below), it will only download 8 tasks assuming you only have one active project running from your boinc manager.

9) Message boards : Number crunching : corona_PLpro_v4 errors (Message 1626)
Posted 18 Apr 2022 by pututu
Post:
That's a very good news to speed up the search. Thanks.
10) Message boards : Number crunching : corona_PLpro_v4 errors (Message 1623)
Posted 17 Apr 2022 by pututu
Post:
No more invalid results anymore. Thanks!

BTW, I notice the new tasks for this target 14 search has quorum of one. Is that the plan for future new target search or just this one? With quorum of one (assuming this still provide reliable results), the total computational tasks will almost be double and will cut the search time by half.
11) Message boards : Number crunching : corona_PLpro_v4 errors (Message 1621)
Posted 15 Apr 2022 by pututu
Post:
Seeing errors on PLpro_v4 project on my windows clients. Linux clients appear to be ok for now. Edit: started to see invalids on my linux clients :(

https://www.sidock.si/sidock/workunit.php?wuid=30457978

<core_client_version>7.16.20</core_client_version>
<![CDATA[
<stderr_txt>
15:17:26 (5616): wrapper (7.17.26016): starting
15:17:26 (5616): wrapper: running build\cmdock.exe (-c -j 1 -r target.prm -p "C:\ProgramData\BOINC\slots\24\data\scripts\dock.prm" -f htvs.ptc -i ligands.sdf -o docking_out)
16:00:00 (5616): build\cmdock.exe exited; CPU time 2533.937500
16:00:00 (5616): called boinc_finish(0)

</stderr_txt>
]]>
12) Message boards : News : Half a year, the BOINC Workshop and a move to the new server (Message 824)
Posted 25 Apr 2021 by pututu
Post:
Just a suggestion: Might be a good idea to increase the projection communication update from less than 10 secs currently to maybe 2 to 5 mins between updates. This will minimize "DDOS" like event esp during the beginning of the race and may potentially crash the server if the server and/or bandwidth can't handle the sudden surge in traffic.




©2025 SiDock@home Team