Posts by Michael H.W. Weber

21) Questions and Answers : Web site : Up-/download & website availability issues (Message 1359)
Posted 19 Aug 2020 by Profile Michael H.W. Weber
Post:
I noticed that Rakesearch has become significantly less reliable in terms of up-/download of tasks: during the past weeks, several tasks were sitting in the upload queue for hours or even days withuot being transferred to the server and no new work is retrieved, too. This is very problematic as our machines are sitting idle.
Moreover, even the Rakesearch website sometimes is not reachable.

Michael.
22) Message boards : Number crunching : Badges for SAT contributions (Message 1349)
Posted 7 Aug 2020 by Profile Michael H.W. Weber
Post:
Are there any plans for badges for the SAT contributions of the last month?

Michael.
23) Message boards : News : First workunits of new search will be generated soon (Message 1342)
Posted 28 Jul 2020 by Profile Michael H.W. Weber
Post:
So, what is the plan for this project?

Michael.
24) Message boards : News : Thank you for R10 search! (Message 1321)
Posted 24 Jun 2020 by Profile Michael H.W. Weber
Post:
Thank you for this overview.
What fraction of the entire R10 search space has been covered so far?
Given the fact that the optimized app has only been released very recently and allows for a significant improvement in computational throughput, wouldn't it make sense to keep Rakesearch running with this app to further expand the search space?

Michael.
25) Message boards : News : The last bunch of tasks were generated (Message 1304)
Posted 1 Jun 2020 by Profile Michael H.W. Weber
Post:
I am a bit baffled. If I remember correctly, it was said before that the R10 search space would never be finished completely and now the project is over entirely? What happened?

Michael.
26) Message boards : Number crunching : Same WUs reach 100% but staying active - No upload (Message 1251)
Posted 11 Feb 2020 by Profile Michael H.W. Weber
Post:
Maybe related to this issue?
Strangely, only starting today, tasks error out with "given up" status and no error report in the relevant tasks.

Michael.
27) Message boards : Number crunching : Rakesearch appears as duplicate projects in BOINC manager (Message 1250)
Posted 11 Feb 2020 by Profile Michael H.W. Weber
Post:
This morning I found Rakesearch listed twice in the BOINC managers of all my machines.

I noticed that when clicking on the referrer link to the project website, one forwards to the old http website the other to the https website.
I have set the project to complete the current tasks and then to autoremove.
Unfortunately, when trying to remove rakesearch using BAM! only the https project version will be removed. The outdated one remains!
There appears to be no way to get rid of the broken http project version.

What is happening?

Michael.

[edit]: Strangely, only starting today, tasks error out with "given up" status and no error report in the relevant tasks.
28) Message boards : News : New run for RakeSearch R10 (Message 1120)
Posted 19 Jul 2019 by Profile Michael H.W. Weber
Post:
Hello Rudy! I placed the answer in your thread Question about 10 X 10 squares.

Could you please also move this post to the other discussion thread to have everything in one place?

Michael.
29) Message boards : News : New run for RakeSearch R10 (Message 1108)
Posted 17 Jul 2019 by Profile Michael H.W. Weber
Post:
Is it possible that something is not yet properly adjusted with the virtual credits system?

First of all it appears to me that the credits are very low given run times of sometimes over 23 hrs where for similar tasks in the rank 9 scheme for a given credit run times of 20 to 60 minutes were normal (factor of approx. 23 to 69).
Second, I have a case where almost identical credits are given for tasks which differ by a doubled run time:

https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24534598
https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24534603

Michael.
30) Message boards : News : R10 search temporary stopped! (Message 1079)
Posted 25 Jun 2019 by Profile Michael H.W. Weber
Post:
Ehm, didn't you guys get the message? The app is bad, hence you can delete the useless results.

Michael.
31) Message boards : News : Future of the RakeSearch project (Message 1074)
Posted 25 Jun 2019 by Profile Michael H.W. Weber
Post:
How many tasks for R10 have been completed so far?

The server status page currently reports approx. 14% completion for rank 10 already.
I guess that's a mistake?

Michael.
32) Message boards : News : Future of the RakeSearch project (Message 1030)
Posted 6 Jun 2019 by Profile Michael H.W. Weber
Post:
Why would you assume a separate data set will complete in the same time?

I did not assume this at all. I was just interested to know whether additional optimizations may be expected for the rank 10 app. Running the original rank 9 client vs. the SSE/AVX optimized one is simply a waste of energy ressources. So, in case further improvements could be expected, I would prefer to wait until these have been implemented before I blow out energy as hot air.

Michael.
33) Message boards : News : Future of the RakeSearch project (Message 1024)
Posted 5 Jun 2019 by Profile Michael H.W. Weber
Post:

For search in rank 10, we made a new application. And of course, we implemented the most powerful optimizations. (Due to changes of structure, the need for some is gone). But we hope that this application can be additionally optimized - it's another very interesting problem!

For rank 9 a third-party client was released which utilizes SSE/AVX etc. variants and is significantly faster compared to the original client.
Are you telling me that the current rank 10 app release (which requires run times of up to 4 hrs for some of the tasks I completed - compared to 20 min. for rank 9 tasks on the same machine) already employs the same optimizations incl. an autodetection module to select the appropriate SSE/AVX code?

Michael.
34) Message boards : News : Future of the RakeSearch project (Message 1018)
Posted 4 Jun 2019 by Profile Michael H.W. Weber
Post:
I have completed about 700 of the new rank 10 WUs without issue on my various Threadripper and Intel systems under Windows 10. Have any ODLS pairs been found yet in the rank 10 search?

Thank you for reports! Now completed ~11000 of workunits and no pairs with ODLS found. For rank 9 on average, ratio is ~1/270.

Optimize the app for rank 10 and continue the search for a while.

Michael.
35) Message boards : Number crunching : Processing both R9 & R10 tasks on the same machine (Message 1005)
Posted 3 Jun 2019 by Profile Michael H.W. Weber
Post:
I applied these setting to exactly 14 test tasks of the R9 type. So far, two of these validated successfully (against clients with the original app):

https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24010823
https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24010823

Oh, my mistake - so far only vs. clients with the OPTIMIZED app. Now it's three successful validations.

Michael.
36) Message boards : Number crunching : Processing both R9 & R10 tasks on the same machine (Message 1004)
Posted 3 Jun 2019 by Profile Michael H.W. Weber
Post:
Since yesterday around 1 pm the Rakesearch server does not reliably deliver any R9 tasks to may machines anymore. As long as the app_info.xml of the optimized Rakesearch app is located in the project directory, none of the novel R10 tasks can be processed. In order to test whether the same machine can be used to process both, the old R9 tasks usind the optimized app and the new R10 tasks using the non-optimized original app, I did the followning test:

I deleted the app_info.xml of the optimized app from the project directory and simply renamed the optimized app from
rakesearch.exe
to
rakesearch_7.02_windows_x86_64.exe
(which is the name of the original non-optimized app for R9 tasks).
As a result, the machine now processes both, R9 tasks (using the re-named optimized app) and the novel R10 tasks (usign the original app).

I applied these setting to exactly 14 test tasks of the R9 type. So far, two of these validated successfully (against clients with the original app):

https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24010823
https://rake.boincfast.ru/rakesearch/workunit.php?wuid=24010823

Of the remaining 12 tasks, 10 are sitting in the validation queue (status: pending) while two more are processing. It should be noted that the server records the returned results as processed with the original client (although in fact I used the optimized one).

Because I do not know whether this client renaming procedure might cause any issues to the project I am unaware of, I limited this to 14 known tasks and have stopped receiving further R9 tasks on the test machine (computer ID: 3819) until someone of the project lead confirms that this approach is OK (I also have a full list of the tasks I used for testing).

Michael.

P.S.: It would be very nice, if an optimized client could be released which allows to process the novel R10 tasks because with the R9 tasks, the original client is rather slow and therefore requires significantly more electricity to yield results compared to the optimized one.
37) Message boards : Number crunching : Update frequency of the individual ODLS findings page (Message 634)
Posted 9 Nov 2018 by Profile Michael H.W. Weber
Post:
How often is this page updated?

http://rake.boincfast.ru/rakesearch/user_odls_results.php

Although I successfully returned a few thousand tasks already in November, the bottom of the page shows a misleading information line:

Your last result received at: 2018.10.31 01:05:25 | The last result of entire list as a whole received at: 2018.10.31 23:55:35

That information, at least the first part, is certainly wrong and not consistent with the rest of the data from the user page (where the returned tasks are listed).
Could you please fix this?

Michael.
38) Message boards : News : Crunchers poll about Distributed Computing projects (Message 629)
Posted 1 Nov 2018 by Profile Michael H.W. Weber
Post:
Unfortunately, the same over here...

Michael.
39) Message boards : News : Crunchers poll about Distributed Computing projects (Message 628)
Posted 1 Nov 2018 by Profile Michael H.W. Weber
Post:
Unfortunately, the same over here...

Michael.
40) Message boards : News : Second bunch of April 2018 totals (Message 384)
Posted 27 Apr 2018 by Profile Michael H.W. Weber
Post:
Thank you, very nice!
Would it maybe also be possible to autogenerate the corresponding graphical plots for the individual findings?

Michael.


Previous 20 · Next 20

©2024 The searchers team, Karelian Research Center of the Russian Academy of Sciences