Posts by Michael H.W. Weber
|
1)
(Message 5954)
Posted 14 Sep 2018 by 
Michael H.W. Weber
Post: No comment? Michael. |
|
2)
(Message 5948)
Posted 10 Sep 2018 by
Michael H.W. Weber
Post: I had suggested this to a number of other BOINC projects and in many cases this was implemented very nicely (see the Rakesearch or Numberfields projects for the latest examples): When opening the individual user page on the Asteroids@home website, would it be possible to see an overview of the asteroid models to which a participant has contributed? Also, there is this nice DAMIT database - why not include the Asteroids@home user names to each of the models to which this project user has contributed? I believe this would make contributions to the project even more attractive. :) Michael. |
|
3)
(Message 4955)
Posted 1 Aug 2016 by
Michael H.W. Weber
Post: is the code in question public? *bump* Michael. |
|
4)
(Message 1334)
Posted 2 Jun 2013 by
Michael H.W. Weber
Post: What application version do you run on Odroid? I am running your official version of Period Search Application v100.00. Please check these results of my i7 system which are marked as invalid: http://asteroidsathome.net/boinc/results.php?hostid=30648&offset=0&show_names=1&state=5&appid= I highly doubt that these are invalid. Michael. |
|
5)
(Message 1330)
Posted 1 Jun 2013 by
Michael H.W. Weber
Post: Interestingly, my ODROID-U2 has only two inconclusive results, but for the following WU all three results differ (are incinclusive): http://asteroidsathome.net/boinc/workunit.php?wuid=2436444 So, there is a fundamental problem with validation. What I do not understand is that there are results that DO validate. This should not happen if it is a rounding issue or a carriage return problem. Michael. |
|
6)
(Message 1329)
Posted 1 Jun 2013 by
Michael H.W. Weber
Post: Well, I can tell you for sure that ODROID-U2 and ODROID-X2 do both run with your currently supplied Asteroid@home clients. In general, the more recent ODROID systems fall into two classes that differ by using two types of CPUs: Samsung Exynos 4412 Prime: 1.7 GHz CPU clock frequency ODROID-U2 ODROID-X2 ODROID-Q2 and the older Samsung Exynos 4412: 1.4 GHz CPU clock frequency ODROID-U ODROID-X ODROID-Q I suspect that all should run with the same client, but as said above, I (and a friend) have tested only two of them in practice. Of course there are other ODROID families, as you can see here: http://www.hardkernel.com/renewal_2011/products/prdt_info.php Best, Michael. |
|
7)
(Message 1327)
Posted 1 Jun 2013 by
Michael H.W. Weber
Post: Still, I am posting the same issue. Is it possible that you have a fundamental problem by trying to cross-validate results over different CPU types and operating systems? My machine has ID 30648 (i7 system, 8 GB RAM, Windows 7 Prof, 64 Bit) and since yesterday it has processed around 30 results which all are listed as "validation inconclusive". Among the other machies returning the inconclusive results there are Darwin, Windows 7, Windows Vista, Windows Server 2008 R2, Windows 8 Pro, etc. (32 and/or 64 bit). So fa no Linux or ARM machine. By the way: I run an ODROID-U2 (ARM Cortex-A9) under Ubuntu 12.04 LTS and it works fine with Asteroids@home - maybe you want to add that in your projects applications page (ARM support not only for Raspberry Pi or ARM Android). With our RNA World project, we figured that validation is giving a carriage return problem when the output data is text file (Linux doed not validate Windows results). Another isssue is a fundamental rounding differences between AMD and Intel (AMD will not validate Intel results and vice versa) - possibly Apple? Just a few ideas, you might probably also have thought of already... :) Michael. |