Posts by mmonnin

1) Message boards : Number crunching : Constant computation errors. (Message 105460)
Posted 16 Mar 2022 by mmonnin
Post:
Error like this?

process exited with code 1 (0x1, -255)</message>
<stderr_txt>
command: ../../projects/boinc.bakerlab.org_rosetta/rosetta_4.20_x86_64-pc-linux-gnu @preetham_gen_38675_0001_0001_0.flags -nstruct 10000 -cpu_run_time 28800 -boinc:max_nstruct 20000 -checkpoint_interval 120 -mute all -database minirosetta_database -in::file::zip minirosetta_database.zip -boinc::watchdog -boinc::cpu_run_timeout 36000 -run::rng mt19937 -constant_seed -jran 1371728
Using database: database_357d5d93529_n_methyl/minirosetta_database

ERROR: Error in simple_cycpcp_predict app read_sequence() function! The minimum number of residues for a cyclic peptide is 4. (GenKIC requires three residues, plus a fourth to serve as an anchor).
ERROR:: Exit from: src/protocols/cyclic_peptide_predict/SimpleCycpepPredictApplication.cc line: 2264
BOINC:: Error reading and gzipping output datafile: default.out
16:47:07 (139426): called boinc_finish(1)
2) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 103692)
Posted 4 Dec 2021 by mmonnin
Post:
Python tasks failing

I don't see that you have VirtualBox installed.
https://boinc.bakerlab.org/rosetta/show_host_detail.php?hostid=6157362

But you are better off with VBox 5.2.44 anyway. Version 6.1 gives "Vm job unmanageable" suspensions.
https://www.virtualbox.org/wiki/Download_Old_Builds_5_2


I have 6.x and never have this issue with LHC but about half have these issues at Rosetta.. Plenty of place and memory. Rosetta has never had an efficient app.
3) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 103652)
Posted 2 Dec 2021 by mmonnin
Post:
Nothing but download errors now with the new vdi file

<core_client_version>7.16.6</core_client_version>
<![CDATA[
<message>
WU download error: couldn't get input files:
<file_xfer_error>
<file_name>AIMNet_minimization_python_project.py</file_name>
<error_code>-224 (permanent HTTP error)</error_code>
<error_message>permanent HTTP error</error_message>
</file_xfer_error>
</message>
]]>

No size on the file when it actually exists.
-rw-r--r-- 1 boinc boinc 0 Dec 1 18:56 AIMNet_minimization_python_project.py
4) Message boards : Number crunching : Not getting any python work (Message 103268)
Posted 14 Nov 2021 by mmonnin
Post:
Try:
Close BOINC client. Put a file called Sheduler projects.txt, in the boinc.bakerlab.org_rosetta folder. Write it:
Sheduling priority WUs: rah_make_work_rosetta_python_projects
Save and update.
Pre-emptively add BOINC space to the disk where the same is installed, via Options -> Processing preferences -> Disk and memory.
At least 30GB, otherwise the WUs won't go down (and could download Rosetta's wus).
If the space is insufficient the project will warn.
---------------------------------------------------------------------------------------------
I also tried to edit the app_config.xml file, for a Ryzen 3 3100 (4C / 8T), 16GB RAM as follows:

<app_config>
<app>
<name>rosetta_python_projects</name>
<avg_concurrent>3</avg_concurrent>
<max_concurrent>3</max_concurrent>
</app>
<app_version>
<app_name>rosetta_python_projects_v1.03</app_name>
<plan_class>vbox_64_mt</plan_class>
<avg_ncpus>2</avg_ncpus>
<max_ncpus>2</max_ncpus>
</app_version>
</app_config>

The purpose was to test the possibility of multithreading, but the file:
vboxwrapper_26203_windows_x86_64.exe (981 KB) does not allow this.
Perhaps because it is incomplete.
Another file from another project allows it and works: vboxwrapper_26200_windows_x86_64.exe (1452 KB).
I also tried to replace it, but it is automatically restored from the project.

However, on the Ryzen 3100, it processes a maximum of 2 wus at a time, while on the Ryzen 3600 (32GB RAM), the wus processed are 4.
Any changes I have tried to the app-config.xml file are totally ignored, even when restarting the BOINC client.

The amount of RAM is not a problem.

Byez


Your app_config doesn't work because that is the wrong plan_class
From the applications page: 1.03 (vbox64)
Replace vbox_64_mt with vbox64
Although that probably doesn't mean that the app will use 2 threads. BOINC may dedicate 2 threads but thats different from the app using 2 threads.
app_config.xml just needs to be reread. The client does not need restarted to pick up the changes.
The messages/log will show you what is incorrect with the app name and/or plan_class.
5) Message boards : Number crunching : Not getting any python work (Message 103266)
Posted 14 Nov 2021 by mmonnin
Post:
<?xml version="1.0"?>

-<app_config>


-<app>

<name>rosetta 4.20</name>

<max_concurrent>1</max_concurrent>

</app>

</app_config>

But BOINC sends out multiple tasks using as many cores as it wants to.
I have no restrictions set anymore.



I've read on these forums that there's a bug regarding the max concurrent option. Instead of limiting the number of tasks, it downloads too many or something like that.
Also, I think the Rosetta 4.20 app name is just "rosetta".


The bug, I thought it was in app_config, but that was many discussions ago
You might be on to something with the 4.2
I'll remove that number and see, because over in the BOINC Berkley someone had me use a fake name for RAH and it came back as just Rosetta.


There is no bug.
max concurrent only specifies how many to run, not how many to download.
W/o any project references to select apps there is no way to only get non-vbox tasks at rosetta. You can always abort then, Rosetta does not penalize downloads for high abort counts.
6) Message boards : Number crunching : New kind of app on Ralph (Message 102018)
Posted 5 Jun 2021 by mmonnin
Post:
... as well as being able to run more than once at a time. If the PC has memory.
But have you seen more than one running at a time?


For the 3rd time. YES. This is possible. I have seen it with my own eyes on my own PC. 64gb of memory on that PC and usage was not even 50%.
7) Message boards : Number crunching : New kind of app on Ralph (Message 101994)
Posted 1 Jun 2021 by mmonnin
Post:
By the way, I am running my first 1.02 Rosetta Python now, and BoincTasks is showing that it is using 7.6 GB of memory.

It completed successfully, but errored out on another machine that had only 8 GB. I think this will be a learning experience.
https://boinc.bakerlab.org/rosetta/result.php?resultid=1388548962

But maybe they will send you only one at a time? Even one will catch some people by surprise.
Some control at our end would be better.


I already mentioned the memory and disk space requirements as well as being able to run more than once at a time. If the PC has memory.
8) Message boards : Number crunching : New kind of app on Ralph (Message 101972)
Posted 31 May 2021 by mmonnin
Post:
Apparently, when i got one and another one is in waiting the CPU crunches one of the pythons ones and nothing else.

Same here. I have a 12 cores AMD Ryzen and even if i stop all other wus, the system crunchs only 1 python.


You're out of memory. 7.6GB required per task.
9) Message boards : Number crunching : New kind of app on Ralph (Message 101966)
Posted 31 May 2021 by mmonnin
Post:
I downloaded other 5 python wsus.
Some questions:
- is it possible to crunch more than 1 wu at once?
- will be possible to select this app in personal profile of the project?


I have several going at once on one PC.

Forever a wish here at Rosetta.

2 completed successfully here so far. Ralph had one exceed disk space.
Peak disk usage: 9,559.60 MB on good Rosetta task.
10) Message boards : News : Coronavirus update from David Baker. Thank you all for your contributions! (Message 99134)
Posted 26 Sep 2020 by mmonnin
Post:
WTF is going on here. I've never had a SPAM popup from BOINC before. Intentional or not, it is the definition of SPAM. Seems like the only likely solution is to detach Rosetta.
11) Message boards : Number crunching : CPU getting too hot (Message 91076)
Posted 31 Aug 2019 by mmonnin
Post:
The simplest way if you are not using the computer for other high CPU tasks is to reduce the CPU speed until the temp gets below your target. Use the power options/advanced and set the MAX CPU speed less than 100% and the Min cpu speed at 50%. On a stock PC this can be 75-95% on the MAX speed.


Check the timestamp. This thread is just about 10 years old.
12) Message boards : Number crunching : Unrealistic expectations (Message 91048)
Posted 20 Aug 2019 by mmonnin
Post:
The initial BOINC estimates are way off, about 4 hours that I recall (I run the 24-hour work units). So until they correct themselves, I always get too many when first attaching.

To speed up the correction, you need a cc_config.xml file placed in the BOINC data folder, with this entry:
<cc_config>
  <options>	
  	<rec_half_life_days>1.000000</rec_half_life_days>
  </options>
</cc_config>

You then need to activate it by reading it in, or else restarting BOINC.
It should correct the time estimates in a couple of days.


cc_config options require a client restart as it is only read at client startup. app_config can be read.
13) Message boards : Number crunching : Unrealistic expectations (Message 91043)
Posted 18 Aug 2019 by mmonnin
Post:
Your own BOINC Manager requests work from project servers. The server responds to that request. This is not the server/projects fault.

An 8 hour task is the default task length unless you change it. Other apps/OS using CPU resources can push that to ~9 hours.
14) Message boards : Number crunching : Computation errors (Message 90943)
Posted 26 Jul 2019 by mmonnin
Post:
I'd rather have the Rosetta mini tasks vs the Rosetta version that runs for 5h then has an error when the set run time is 1hr.
15) Message boards : Number crunching : Rosetta@home using AVX / AVX2 ? (Message 90841)
Posted 14 Jun 2019 by mmonnin
Post:
To clarify, Zen 2 will have AVX2 support in a single cycle. Zen/+ can do AVX2 but needs 2 cycles to complete it.
16) Message boards : Number crunching : Rosetta is not playing nice with my other projects. (Message 89698)
Posted 5 Oct 2018 by mmonnin
Post:
BOINC Manager controls how many tasks are downloaded and which ones run. Project's don't work either each other or your PC. This is all controlled by BM.
17) Message boards : Number crunching : Rosetta is not playing nice with my other projects. (Message 89684)
Posted 3 Oct 2018 by mmonnin
Post:
BOINC averages the number of WU over a period of time. When you add a new project, it will try to crunch the job mix to MEET the the WU mix defined. I think they balance on the number of WU and not on CREDITS. If you have 2 projects equally weighted, BOINC will try to match the number of WU.

A couple things you can use to get stable faster.
1. Select the shorter 1 hour Rosetta WU until it stabilizes. You will complete more WU more quickly.
2. Change the default 10 day window for calculating averages so the window is shorter.


<rec_half_life_days>X</rec_half_life_days>
A project's scheduling priority is determined by its estimated credit in the last X days. Default is 10; set it larger if you run long high-priority jobs.


BM should be scheduling by time, right? BM has no idea what credit will be. I would expect if R@H takes 1 hour and E@H takes 2 hours then 2 R@H tasks will run and 1 E@H will run as the resource share levels out. At that point a host's queue would have 2x R@H than E@H and run them 2:1 giving and equal resource share. Assuming the same share is given to each project.
18) Message boards : Number crunching : Keeping new volunteers (Message 89678)
Posted 2 Oct 2018 by mmonnin
Post:
Rosetta is not a new project so there has been plenty of time for users to come and go over the years. Many quit when they see their electrical bill.
19) Message boards : Number crunching : Rosetta is not playing nice with my other projects. (Message 89677)
Posted 2 Oct 2018 by mmonnin
Post:
Adding another project to a computer will make BM try to catch up and run the new project more until it reaches the given resource share ratio. You can let it go and it will work itself out in the long run or manually adjust R@H lower for now and gradually make it our desired ratio.
20) Message boards : Number crunching : For the betterment of BOINC (Message 89192)
Posted 29 Jun 2018 by mmonnin
Post:
FAH is benchmarked on one system and assigned a fixed base credit based on how it performs on that CPU and GPU. That system has changed over the years but it allows for much closer average PPD across it's types of work units. QRB has widened that range of PPD for faster GPUs due to its non-linear formula.

CreditNew seems to try use the Whetstone benchmark to get that even RAC/PPD across many systems vs benchmarking it on one system like FAH. Benchmarking on one system with many projects isn't very feasible.


Next 20



©2022 University of Washington
https://www.bakerlab.org