Posts by Mod.Sense

21) Message boards : Cafe Rosetta : something broke (Message 97644)
Posted 26 Jun 2020 by Mod.Sense
Post:
Presently, there are no R@h work units available for download. More coming soon (as is always the case).
22) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 97423)
Posted 16 Jun 2020 by Mod.Sense
Post:
Yawn.... move it then.


Please heed the advice to carry on the conversation where more appropriate. All of it is off topic, and it is easier to delete than to move. Far better for all to simply do more than yawn in the first place.
23) Message boards : News : Rosetta Graphics released for Windows (Message 97365)
Posted 13 Jun 2020 by Mod.Sense
Post:
When the conformation of a given protein is known, it is very useful to see if your computational algorithms can come up with the same answer. Because if they don't, you've got more work to do.

In very general terms, any time you come up with a computer model of something that occurs in nature, you want to confirm your model by blindly applying it to a natural occurrence to see if it predicts the outcome you actually observe in nature. If your model does not predict what actually occurs, then there is room to improve the model.

For example, if you have a computer model that predicts changes to world climate, you want to enter all of the data that you have for the year 1900 and see if you accurately predict what was observed in 1910 and 1920.

Rosetta is really about developing the best computational model of how proteins work.
24) Message boards : Cafe Rosetta : Personal Milestones (Message 97345)
Posted 12 Jun 2020 by Mod.Sense
Post:
Rosetta crosses the 100 Billion credits issued mark!
25) Message boards : Number crunching : Waiting for Memory (Message 97247)
Posted 5 Jun 2020 by Mod.Sense
Post:
The project already makes an effort to ensure the host machine has sufficient memory to run the WU. Unfortunately, BOINC Manager does not seem to handle it well, and you have users willing to drop the entire project when they happen to notice BOINC Manager accommodating larger WUs.

If you created distinct applications for various memory footprints, you would also create an expectation for a continuous workflow for each memory footprint. But that isn't how the research works. If you are running a study of HIV, or COVID, these are large proteins that typically have a large runtime memory requirement. You, in that specific study, have zero need for any low memory WUs. The BakerLab and RosettaCommons researchers have many studies advancing concurrently. But, much of the area where new innovations and study are required are with larger proteins.
26) Message boards : Number crunching : Why losing average credit. (Message 97246)
Posted 5 Jun 2020 by Mod.Sense
Post:
My understanding is that the credits received are dependant on the number of decoys completed and is a measure of the work done.


Yes, the number of decoys is the basis for granting credit. However, in your extremely rare case where two machines both completed the same WU, all are granted the same credit. In essence there is a BOINC rule superseding the project's method of computing credit.

R@h does not request the WU verifications that some other BOINC projects do. It is better to spend that time exploring new models.

It is because:
a deadline was passed
and a new task for the WU was issued
and both the expired task and the reissued task were reported back
... that you got the result you see. If any of those three things had not happened, things would have been normal.
27) Questions and Answers : Windows : Computation error (Message 97220)
Posted 4 Jun 2020 by Mod.Sense
Post:
The tasks says it got an out of memory error. If you look in Windows task manager, how much memory are the active tasks using? You can sort by memory used by clicking the column heading.
28) Questions and Answers : Windows : Rosetta eating up my computer memory (Message 96858)
Posted 30 May 2020 by Mod.Sense
Post:
There are a number of dials to control things. If you still find the machine sluggish or swapping, I would suggest posting on the Number Crunching board with both your objectives (such as "I need to use the machine for other work during the business day") and your observations (such as how many tasks your are running concurrently, how much memory they are consuming, etc.)
29) Questions and Answers : Getting started : Tasks lost through hardware failure (Message 96839)
Posted 29 May 2020 by Mod.Sense
Post:
They will be reassigned automatically. Thanks.
30) Message boards : Number crunching : Less is more? (Message 96837)
Posted 29 May 2020 by Mod.Sense
Post:
There are many potential bottlenecks is processing work on a computer. If you need to write to disk, is the disk busy writting data for other tasks? If you need to examine the contents of memory, is the memory page out in the swap space? Is the page available in the on-processor memory cache? Etc.

R@h tasks are typically fairly high memory usage. As such, there is a lot of interaction between the active process and memory. As such, processors with larger L2/L3 caches often do better than systems with smaller caches. It improves the odds of any given memory access to be serviced immediately.

By running all of your 4 cores at a fractional percentage, you have 4 active tasks with memory footprints and contention for L2/L3 cache. By splitting the work across 4 tasks, you diminish the odds of finding the active things in the cache. If instead you run 2 tasks at 100% of CPU, there will only be two tasks worth of contention for cache rather than 4 tasks worth. Thus improving the odds of getting higher performance cache hits. It will also help the system be more responsive to higher priority tasks because of the reduced memory/cache footprint of the BOINC tasks.

So yes, there are certainly cases where less is more. It seems the thing you are most focused on is fan speed/noise. You would have to try it and see, but running 2 cores at 100% would have some other efficiency gains over running 4 cores at 36%, if the resulting fan speed allows it. If fan kicks back to higher speed, then I would suggest running 2 cores (50% of CPUs), for something less than 100% of the time, rather than all 4.
31) Questions and Answers : Unix/Linux : boincmgr with rosetta downloaded lots of data and when I rebooted it seemed to start over (Message 96810)
Posted 27 May 2020 by Mod.Sense
Post:
Looks like you are getting download errors:

<core_client_version>7.16.6</core_client_version>
<![CDATA[
<message>
app_version download error: couldn't get input files:
<file_xfer_error>
  <file_name>database_357d5d93529_n_methyl.zip</file_name>
  <error_code>-120 (RSA key check failed for file)</error_code>
  <error_message>signature verification failed</error_message>
</file_xfer_error>
</message>
]]>


Perhaps you have an anti-virus that is blocking the zip file from downloading?
32) Message boards : Number crunching : Problems and Technical Issues with Rosetta@home (Message 96747)
Posted 23 May 2020 by Mod.Sense
Post:
The databases are all in the root of the downloads directory
https://boinc.bakerlab.org/rosetta/download/database_357d5d93529_n_methyl.zip
33) Questions and Answers : Macintosh : Client can't connect to internet (Message 96726)
Posted 22 May 2020 by Mod.Sense
Post:
Which VPN do you use? Perhaps others have experience with using it.
34) Questions and Answers : Macintosh : new URL (Message 96725)
Posted 22 May 2020 by Mod.Sense
Post:
The project has enabled https, secure network connections, so the only change in the URL is the "S".

Lots of info. here.

Important points:
attach with same profile and password as your original account. i.e. select the option to connect with an existing account, not a new one.
You must first detach the project under the existing URL without the "S", and this will abort all of the tasks you currently have, hence the discussion of setting "no new tasks" (NNT) until existing work completes and is reported back.
I would keep it simple and ignore the manual editing of files that some describe in the thread.
35) Message boards : Number crunching : Problems with web site (Message 96714)
Posted 22 May 2020 by Mod.Sense
Post:
Stats are generally exported once a day. The timestamps on the files all look up-to-date.
https://boinc.bakerlab.org/rosetta/stats/
36) Message boards : Number crunching : Rosetta 4.1+ and 4.2+ (Message 96696)
Posted 21 May 2020 by Mod.Sense
Post:
There have been a few changes recently.

The change to Rosetta version 4.20. If you had explicitly whitelisted a prior version in an anti-virus application, it may need to be updated for the new version.

The change in project URL to be https. Have you reconnected to the project under the new URL? If so, perhaps wget needs support for SSL? I don't know why R@h would use wget. I believe all network activity is handled via BOINC APIs.
37) Questions and Answers : Unix/Linux : Connection Closed error using Fold for Covid on Raspberry Pi 4 (Message 96682)
Posted 20 May 2020 by Mod.Sense
Post:
BOINC does schedule retries of failed file transfers. You'd have to get more specific about the sequence of events. The event log shows transfer failures and the pending transfers list shows when it plans to schedule a retry. In general, the more failed transfer attempts, the longer it waits before trying again.
38) Questions and Answers : Macintosh : Client can't connect to internet (Message 96666)
Posted 20 May 2020 by Mod.Sense
Post:
Have you set the preferences so that BOINC only uses the network at specific times of day? BOINC also has activity settings that allow you to disable network access, or enable network access based on a schedule that you setup. ...are you using those? Or turning network access on and off from the operating system?

BOINC does not require a network connection that is always on. But, if it trips upon an event that causes it to want to do network access (such as trying to upload the file shown in your message), it will try to access the network and throw an error like that when you have the network down.
39) Questions and Answers : Unix/Linux : boincmgr with rosetta downloaded lots of data and when I rebooted it seemed to start over (Message 96662)
Posted 20 May 2020 by Mod.Sense
Post:
The BOINC Manager will take care of retrying downloads that get interrupted. BOINC also has settings where you can limit bandwidth usage if you like. "Avg. work done" is over the last 10 days, and during most of those, it sounds like you did no work because you were not attached to the project, so sort of a meaningless number hours after you signup.
40) Questions and Answers : Unix/Linux : Jobs seem to complete OK but have status 'abandoned' (Message 96659)
Posted 20 May 2020 by Mod.Sense
Post:
@loris, sorry Loris, we seem to be talking about two different things. So, it sounds like you are indeed in the right place. You just (jokingly) have to put up with all of the questions about how you went about settings this up and are trying to run it. The simplest way would be to install each machine and let them each do their own connections to the project for work. In that sense, the project never sees a cluster.


Previous 20 · Next 20



©2024 University of Washington
https://www.bakerlab.org