WARNING: This website is obsolete! Please follow this link to get to the new Albert@Home website!
Posts by Bikeman (Heinz-Bernd Eggenstein) |
21)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112586)
Posted 3 Jun 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: First result validated, nice! Testing here is also used to adjust the credits for Einstein. The variations in run time of different "kinds" of tasks (frequency, etc) is much more pronounced in this run, e.g. older Core2 CPUs show a greater "run time scatter" than modern Sandy Bridge or Ivy Bridge CPUs. We need to understand these variations first and then find a compromise credit policy that is more or less fair to everyone. Cheers HB |
22)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112585)
Posted 3 Jun 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Apropos, why are the auto-format buttons like [url] so far off-screen to the right here? I'm using Chrome on Windows XP. Hi! There will be a mojor overhaul of the Web presence in the near future, so forgive us for not investing any significant time in fixing stuff in the legacy web front-end now. Cheers HB |
23)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112584)
Posted 3 Jun 2013 by Bikeman (Heinz-Bernd Eggenstein) Post:
No idea yet :-(. Since this seems to affect other projects as well, I guess it's most likely not a science app problem. I also can't think of a way how a BOinc server side problem could trigger this. Strange.. HB |
24)
Message boards :
Wish List :
App for Intel HD graphics?
(Message 112579)
Posted 31 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Hi Thanks guys, that's a plan. OK, so I put an item on my TODO list, expect something next week. Feel free to nag me if I should not deliver :-) Cheers HB |
25)
Message boards :
Wish List :
App for Intel HD graphics?
(Message 112575)
Posted 30 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Any progress on this? I have successfully crunched GPU tasks for both collatz and SETI (astropulse), on my intel HD 4000. Hi! No, unfortunately no progress yet. Our BOINC server side code is not yet able to handle Intel GPUs consistently (certainly not on Einstein and not even on Albert IMHO). We (the E@H admins and devs) have quite a lot of other stuff on our desks at the moment (to name just a few: launch of S6directed on Einstein, transition of BRP4 to a CPU only run, evaluating the early BRP5 results, a complete overhaul of the Web Homepage and forum platform, postprocessing of finished GW runs, ....). Yeah, it's a shame really, especially with the next Intel CPU generation to be launched soon (Haswell) which is said to feature some relatively powerful GPUs. Sigh. What would be possible at the moment: we provide (in the timeframe of a few days) an app to download that someone could test using an app_info.xml file that would pretend the app is a CPU app. So the Boinc client would ask for work for a CPU anonymous platform task, but the task would then in fact be calculated on the Intel GPU, so one would at least get an idea about the runtime. Is anyone who is experienced in app_info.xml writing interested in such an experiment? Cheers HB |
26)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112573)
Posted 27 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: I've attached my HTC One S to Albert@home, not had any validations yet, eithier the app gets a SIGSEGV: segmentation violation, or if it completes, it then gets a Validate error, running the Official Android 7.0.58 Boinc, Thanks for the report! I see one, maybe two other devices in the DB that might have similar issues. I'll have to investigate. Thanks for reporting this!! Cheers HB |
27)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112569)
Posted 24 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Just happened to open the Windows task manager and noticed that each of the three running S6 Directed Search tasks was using ~600 MB of RAM each, compared to the S6 LineVeto search over at Einstein where each task use ~110-115 MB of RAM. One of the (many) differences between this and the previous run is that work units will vary a lot more, both in memory consumption and in the data volume that is required for a specific task. Among the currently generated test work units will be some of the most demanding in this respect, kind of a worst-case test. BOINC allows to configure limits both for RAM and disk usage by BOINC, so it's possible with some care to make sure that BOINC will not take more ressources that the user is prepared to contribute. Cheers HB |
28)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112563)
Posted 23 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: I notice the Raspberry Pi app has updated to 1.06 now. Were there problems with 1.05 or some further improvements? There was a small bug fix, and the new app was compiled with a different toolchain (cross compiled on an Intel x86 Linux host like the Android (and Windows) app versions). Before that, it was a bit cumbersome to build the app on the Raspberry Pi itself, and it didn't integrate well into our automated continuous integration solution (Jenkins). I expect no significant performance change. Cheers HB |
29)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112560)
Posted 22 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Hi Thanks for the feedback. We are still fine tuning and fixing things, so expect some hickups. We are aware of the app selection problem, and we fixed some work generation problem. Please keep testing this, at this point in testing we *want* to see errors and problems (better now than later in production :-) ). Cheers HB |
30)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112555)
Posted 20 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Hi! I think it's too early for funeral eulogies wrt. Intel Atom. There are (more than) rumors that Samsung will use Atoms in some future tablet and possibly even smart phones. I guess they know what they are doing. We'll see. Cheers HB |
31)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112554)
Posted 20 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post:
Good!!
Indeed. The code is more or less the same, but the parameters of the tasks which tell the code 'where' to look for what kind of signals, is radically different. In fact, there has probably never been so much difference between a E@H run to the previous run in this respect, so we are now verifying and possibly tweaking the set-up of this run, and that's what the test here on Albert is all about. The final run parameters on Einstein might be different, but hopefully we got it about right in the first attempt already :-). Thanks for testing and feedback, Cheers HBE |
32)
Message boards :
Problems and Bug Reports :
FGRP application v 1.07 (OPENCL) feedback thread
(Message 112545)
Posted 19 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Nice!!! Thanks for the feedback. Cheers HB |
33)
Message boards :
Problems and Bug Reports :
S6 Directed Search (CasA) Feedback thread
(Message 112543)
Posted 19 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Year, sorry, there seems to be a general problem. I'm not sure we'll be able to fix this before Tuesday (Monday is a public holiday here), so for now I have "deprecated" the new S6Directed apps. Feel free to suspend or cancel any that you received alraedy. Cheers HBE |
34)
Message boards :
Problems and Bug Reports :
FGRP application v 1.07 (OPENCL) feedback thread
(Message 112542)
Posted 19 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: since actual GPU use is so low (~20%) can you allow us to change the GPU utilization factor in settings? I guess eventually we'll do that, probably with a separate setting as this app consumes more RAM on the graphics card than the BRP4 (or future BRP5) tasks. So those who are able to execute (say) 4 tasks of BRP4 in parallel might only be able to run 3 of the FGRP2 tasks in parallel, hence separate settings. Cheers HB |
35)
Message boards :
Problems and Bug Reports :
FGRP application v 1.07 (OPENCL) feedback thread
(Message 112538)
Posted 18 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Interesting, thanks for additional data. We'll have to look deeper into this. Cheers HB |
36)
Message boards :
Problems and Bug Reports :
FGRP application v 1.07 (OPENCL) feedback thread
(Message 112535)
Posted 17 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post:
Hmm.. that stuff looks odd. The card with the highest load according to the diagram over a timeline has a low temperature, while another GPU with only intermittent load has a very high temperature, and the second GPU doesn't have a temperature reading at all, even tho all GPUs are very similar (Cypress GPU). So I'm not quite sure which part of the data from this tool can be trusted, because there are obvious inconsistencies. Cheers HB |
37)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112524)
Posted 12 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: It still hasn't quite finished the 1st one yet (its on 96% after 30 hours). I will provide links once it completes. Cool, validated against an Android (NEON) task that took a bit less than half as long to finish. Cheers HB |
38)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112522)
Posted 12 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post:
Excellent, thanks! The obvious question is: will it be possible to improve the Android app versions in a similar way? That remains to be seen(I'll need to make a wrapper app first that produces and dumps the FFTW3 'wisdom' on an Android device. Stay tuned. Cheers HBE |
39)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112520)
Posted 10 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: OK, next attempt .... I just released version 1.05 for Raspberry Pi . If all goes well this should be 20 - 25 % faster than the previous version. Feel free to cancel task that are/will be running under version 1.04, this will help to get 1.05 results in soon. Thanks for testing!! Cheers HB |
40)
Message boards :
Problems and Bug Reports :
BRP4U (Raspberry Pi , single DM tasks) feedback thread
(Message 112519)
Posted 10 May 2013 by Bikeman (Heinz-Bernd Eggenstein) Post: Hi! yeah, I think you are right. Doesn't seem faster at all. Maybe I didn't turn off the FFT wisdom file when I tested this one....(including a wisdom file is another thing I'm working on and this WOULD speed up things indeed). So stay tuned. Thanks for testing Cheers HB |