Releasing Android application for ARM processors
log in

Advanced search

Message boards : News : Releasing Android application for ARM processors

1 · 2 · 3 · Next
Author Message
Profile Kyong
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 9 Jun 12
Posts: 548
Credit: 52,283,184
RAC: 7,147
Message 953 - Posted: 10 Mar 2013, 18:55:37 UTC

There is a new application for ARM devices running Android.

Radim Vančo (Kyong)

Profile ritterm
Avatar
Send message
Joined: 22 Jun 12
Posts: 18
Credit: 5,142,840
RAC: 34
Message 955 - Posted: 10 Mar 2013, 21:16:05 UTC - in response to Message 953.

There is a new application for ARM devices running Android...

Attached, downloaded, and running... Well done, Radim!
____________

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 956 - Posted: 11 Mar 2013, 0:00:51 UTC

Same here, and ditto the kudos. I've just attached, gotten first WU; only problem: I'll use the search engine to see if there's any relevant news aside from this thread. However, is anyone else who runs multiple ARM applications getting a comparatively astonishingly long running time for their WUs? Most other projects (aside from PG and MW which are fairly long), mine usually complete (on different machines) in 1-4 hours, with the lengthy ones going anywhere from 8-14 hours (I aborted one which took in excesss of 15 hours and had an incredibly long time left, but that was an aberration for the project, subsequently fixed). My first Asteroids WU is listing 139 hours and counting to completion, at 0% completion w/15 mins now elapsed.

1) How often does the WU checkpoint, by default?

2) Any idea on how long WU actually takes to complete?

3) How accurate is that status bar, etc.?

Excited that Asteroids is joining the ARMs race ;) I love the platform. But if it tuly takes 350 hours, that's prolly not going to work for me, for now.\

Any other feedback on this exciting development? Echo the props - well done, Radim! Can you answer how long I can expect to run these, is the credit fixed per WU, etc.? I got the Period Search 100.


Best regards,
StandardbredHorse/Mic Morose, Founder of
Richard Dawkins Foundation for Reason and Science.

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 957 - Posted: 11 Mar 2013, 0:09:31 UTC - in response to Message 956.

Questions also directed @Kyong if you're able to give estimates as well - thx!

S. Horse (Mic)

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 958 - Posted: 11 Mar 2013, 0:28:12 UTC - in response to Message 953.

There is a new application for ARM devices running Android.

Radim Vančo (Kyong)

Update: 47:05 elapsed
0% complete
Last checkpoint (cpu time) 0.00.00
status running
to completion: 138:51:36

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 959 - Posted: 11 Mar 2013, 0:31:46 UTC - in response to Message 958.

There is a new application for ARM devices running Android.

Radim Vančo (Kyong)

Update: 47:05 elapsed
0% complete
Last checkpoint (cpu time) 0.00.00
status running
to completion: 138:51:36


here r my WUs:


http://asteroidsathome.net/boinc/results.php?userid=4862

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 960 - Posted: 11 Mar 2013, 0:33:41 UTC - in response to Message 959.

There is a new application for ARM devices running Android.

Radim Vančo (Kyong)

Update: 47:05 elapsed
0% complete
Last checkpoint (cpu time) 0.00.00
status running
to completion: 138:51:36


here r my WUs:


http://asteroidsathome.net/boinc/results.php?userid=4862


(edit): comp # 11643 pf vourse

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 961 - Posted: 11 Mar 2013, 1:43:47 UTC - in response to Message 960.

Update: 1:26:51 elapsed time
progress 1.456%
elapsed timeP: 1:26:51
To completion: 138:11:44
CPU time: 01:20:18
last checkpoint 01:14:44
deadline: 2013-033-10

Feels like I'm quoting bible verses, but is it time to abort 1st ARM test?

Will prob abort if there's no reply.interest, no offense, will wait for feedback.

Anyone else?

S. Horse

Profile ritterm
Avatar
Send message
Joined: 22 Jun 12
Posts: 18
Credit: 5,142,840
RAC: 34
Message 962 - Posted: 11 Mar 2013, 2:25:45 UTC

Will prob abort if there's no reply.interest, no offense, will wait for feedback.

Anyone else?


I would be patient and let it run, at aleast a little longer. :-)

Task 2328683 is my first WU. At the last checkpoint (4:24 hrs), it was 23.4% complete, but showed 77:42 hrs to completion. My guess is it will finish in about 18 hrs.

____________

SaintLaser
Send message
Joined: 3 Feb 13
Posts: 3
Credit: 4,885,920
RAC: 10,535
Message 964 - Posted: 11 Mar 2013, 7:59:38 UTC

I will start crunching after work.
To completion 132:08:32.

Profile Kyong
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 9 Jun 12
Posts: 548
Credit: 52,283,184
RAC: 7,147
Message 965 - Posted: 11 Mar 2013, 9:28:43 UTC - in response to Message 961.

You could have edit the first post, and I was sleeping, so I couldn't immediately reply.
Now the answers:
The computation take about 22 hours on HTC Desire Z (700 MHz) and 14 hours on HTC Sensation (1600 MHz (tested by matszpk)). I didn't check the checkpoints so I can't say now how oftem did NativeBOINC checkpointed it. The estimated time is from the same template like other workunits so I suppose that too big estimated time is problem by the actual BOINC client version core. The status bar shows well but there is some precomputation so it take a while before starting of computing.

matszpk
Send message
Joined: 21 Jul 12
Posts: 7
Credit: 2,066,300
RAC: 0
Message 966 - Posted: 11 Mar 2013, 9:44:40 UTC
Last modified: 11 Mar 2013, 9:45:15 UTC

My BOINC client has support for checkpointing (you can check it). This is not problem. It tested on other project apps (succesfully).

matszpk
Send message
Joined: 21 Jul 12
Posts: 7
Credit: 2,066,300
RAC: 0
Message 967 - Posted: 11 Mar 2013, 10:09:45 UTC
Last modified: 11 Mar 2013, 10:10:02 UTC

Yes. My first workunit was validated successfully! link:
http://asteroidsathome.net/boinc/result.php?resultid=2328248

matszpk
Send message
Joined: 21 Jul 12
Posts: 7
Credit: 2,066,300
RAC: 0
Message 968 - Posted: 11 Mar 2013, 11:20:55 UTC
Last modified: 11 Mar 2013, 11:21:07 UTC

Checkpointed workunit also validated succesfully. link http://asteroidsathome.net/boinc/result.php?resultid=2328219

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 969 - Posted: 11 Mar 2013, 13:08:06 UTC - in response to Message 965.

You could have edit the first post, and I was sleeping, so I couldn't immediately reply.
Now the answers:
The computation take about 22 hours on HTC Desire Z (700 MHz) and 14 hours on HTC Sensation (1600 MHz (tested by matszpk)). I didn't check the checkpoints so I can't say now how oftem did NativeBOINC checkpointed it. The estimated time is from the same template like other workunits so I suppose that too big estimated time is problem by the actual BOINC client version core. The status bar shows well but there is some precomputation so it take a while before starting of computing.


Kyong: Please accept my apologies. I can tell you are, at the very least, a bit annoyed, and I know I deserve some of that subtly-delivered terseness but nevertheless patient tone that was conveyed within your reply, and also appreciate its brevity and thoroughly informative nature- duly noted, and thank you.

I am not new to BOINC, but I am (relatively) new to this project, and although I have crunched several (non-ARM thus far) WUs successfully, and a fellow team member has crunched quite a few, I have not made any posts here before yesterday; the number of posts I've made to all BOINC projects combined over almost a decade of crunching on BOINC is probably less than 50, so I am unfamiliar with this particular style of message board, what functions are allowed/enabled here, etc. etc., but I am learning. This is the first time I have been (among) the initial group of volunteers to kind of (I qualify this statement because I am not sure what your process or policy is before releasing a new platform on which to complete work) test (or beta-test, if applicable) new work or WUs, and though 'test' may not apply because I'm sure they went through rigorous testing before public release, I'm proud to be among the first participants nonetheless.

Although you were sleeping, thank you for letting me be among the first to test the ARM WUs on the project. I am not sure what time zone everyone resides in, but I took Ritterm's advice and did not abort the WU; I too was curious to see how it would run, and was willing to volunteer the time to complete the WU, which on one of my update posts I did say had checkpointed, and after that, allowed it to run. I too fell asleep not long afterwards. I just, literally, woke up to find that the WU had progressed further, and to find your post.

Please excuse the separate postings, but I did not edit for the following reasons:

1) Since I have not posted to this project before, and, being relatively unfamiliar with this message board format, I do not see an 'edit' button once I have posted something. I am fairly bright and technically-oriented, as I'm sure most on this thread are, being fellow BOINC'ers, so I will probably figure it out after this post (hopefully!), but I see the message numbers highlighted, I looked at the smaller icons such as the 'rating' where it shows the ability to +/- a post; initially I thought the red 'x' was to edit, but found that it was to 'report post as offensive'. Aside from that, I see our names highlighted along with our titles and avatars listed (with a 'send message' button, for PM'ing I presume) below that. I still do not see a button for editing, though I know I can hit the 'back' button on my browser and edit that way. However, on a similarly-styled BOINC message board (they are all different, and they can be customized; recently I posted a full HTML-encoded reply to my friend on another project, and a moderator/admin had to edit for me when I reported my own post since editing was not enabled there for some reason). I'm the founder of the RDFRS team which has over a billion credits and several hundred members amongst all the projects combined, and I recently updated our team's profile on almost every BOINC project (not all of them use BOINC-wide team updating, so I had to check them all and update many styles of message board manually; some take HTML, some use limited tags, some take message-board style coding), and as I was tired, I did not figure out how to edit some of what I wanted to edit.

2) I thought that by *not* editing, and by giving you (and others) a blow-by-blow update to how the ARM-project was running could help be beneficial in case there were still bugs that needed to be ironed out, and I manually copied the messages from my ARM-enabled smart phone to my PC here to post, and hoped to hear back from some other participants and solicit some other feedback so we could compare; had I edited the post, I would probably have chosen not to overwrite the data which may have been helpful had we not experienced success in running the ARM platform. I would have edited other parts, but not the actual data, and several short posts were preferable to one long edited post, I was thinking.

3) I am sorry if I came across as a bit cranky. I was tired, hadn't slept for quite a while and had worked all weekend. On a personal level, I got a phone call in between posts with some bad (not terrible, just upsetting) family news. You, or anyone else on the thread, could not have known that, so I'll re-read and if I see things that are inappropriate or not representative of me as an individual, will definitely edit them using the edit function when I figure out where it is. Please accept my apologies; I do not even remember what I posted aside from expressing a bit of doubt and being perhaps a bit insistent in soliciting answers from others regarding their WUs and any Asteroid-affiliated members for answers that I felt were unaddressed in the initial announcement of the release. I hope that you won't hold an inauspicious beginning to my posting history define my personality as a poster. Thank you for your feedback as well as the answers, and thank you to all of those who posted followups while I was on the phone and then fell asleep without updating.

4) As you may/may not have seen in my BOINC-wide team description which was written only a couple of days before this project implemented it, I single out ARM as the biggest BOINC development since GPU computing, and give special shout-outs to the projects as they implement ARM-coding and make ARM-enabled WUs available to process on diverse hosts. Thank you for being among the first to be open-minded, and thank you to everyone, including the developer that I echoed the thank yous to in my first post.

Those are basically the reasons I did not edit, and why I expressed uncertainty; I let the posts stand as magnet posts for others to post their experience(s) re: running the WUs and the checkpointing, status, completion, and verification phases of computing. For the more sharply- or impatiently-worded posts, I apologise for the impatience; I thought more would post right away as the news had just been released an hour or a bit more before I began to run it.

I'm happy for the Asteroid project, and plan to continue to crunch it, and there are team members who have already begun to gravitate to this project (maybe one or two so far, but that's how our team does things, it seems). Congrats on what seems to be a successful implementation of a new platform, and with relatively few to no problems thus far. Thank you for taking the time to reply with an answer re: checkpointing and estimated time being from the BOINC client version core. I've also experienced that problem (WUprop comes to mind, though now that I know that, it's no longer an issue for me), and also have crunched for projects using ARM that have a system that checkpoints faster/more frequently as well as gives a better idea re: completion status, so it's possible there is a solution as they continue to get feedback - it's open-source code, isn't it? I now realise your project takes some pre computation, so won't be accurate at first. If I'd have gotten a bum WU, I wouldn't have wanted to continue it, wasting your project's time as well as mine, when timely feedback could have saved that and prevented others from having an issue. Please trust that I had only the best intentions in posting that, and my style of post could have been better; I'll be better in the attitude department. I'm happy to donate my time to this project and this sub-project. You're a responsive and obviously hands-on PA/dev/tester/scientist, and I thank you for the personal and quick reply. I definitely had the project's best interests in mind, as well as the interests of its volunteers and potential new ones and could have expressed myself with a bit more patience and perhaps better wording - I'll read the posts again when I hit 'post' here.

Thank you for your response, thanks also to *everyone* who posted on this thread, and I'm glad to hear of the success! I took a phone call and fell asleep without plugging my phone back in (forgot), so my WU is not abandoned or aborted; I just ran out of juice until now. So...in a funny sort of circular fashion, I am the beneficiary of my own post (and of your initial post/resultant replies).

Thanks, Kyong. Respect and props where due for being a pioneering project who realised that your A@H fit the ARM platform well (and vice-versa). I will recommend it to my team, as I'm sure others will, and hopefully you see a (manageable) large increase in crunchers and much success :o) Great work, all!

Best regards, and take care
StandardbredHorse / Mic
(founder)
Richard Dawkins Foundation for Reason and Science

StandardbredHorse
Send message
Joined: 14 Jan 13
Posts: 8
Credit: 960
RAC: 0
Message 970 - Posted: 11 Mar 2013, 14:16:04 UTC - in response to Message 966.

My BOINC client has support for checkpointing (you can check it). This is not problem. It tested on other project apps (succesfully).


Hello matszpk, and thanks for the reply/update.

Out of curiosity, what BOINC client do you run? I know of a few; I'm running (on some of my devices not all) NativeBOINC. As part of my (just for personal knowledge, and to help inform others, whether they are part of my team or not) research, are you running on a smart phone as well? I know ANDROINC was popular for a while, but NB may have supplanted it. I took part of yesterday to do some (admittedly surreptitious and interrupted) personal research, but only found those couple of clients - NativeBOINC has been just about all I've needed, though it shows up as 'anonymous platform' on most projects. My WU checkpointed at the time I posted, close to 1:30:00 into it and that's when I knew NB was OK with this project (it doesn't handle others very well, not necessarily the client's fault, but the way some dev's on other projects configure their WUs and perhaps some user-defined settings).

Thanks for anything you're willing to share, and congrats on your success! It was exactly the type of post I was looking for.

Best regards and take care for now,
StandardbredHorse / Mic
(founder)
Richard Dawkins Foundation for Reason and Science

matszpk
Send message
Joined: 21 Jul 12
Posts: 7
Credit: 2,066,300
RAC: 0
Message 971 - Posted: 11 Mar 2013, 14:43:46 UTC
Last modified: 11 Mar 2013, 14:45:52 UTC

For question about BOINC client: This is native (compiled to native code) BOINC client compiled from modified (by me) original sources from BOINC site. I added client monitor, update apps mechanism (for nativeBOINC) and I extended local prefs (power management, battery temperature options). sources codes available at
https://github.com/matszpk/native-boinc-for-android/. the Most project apps has been ported by me and published on the my site (instead on the project servers). This is reason why 'anonymous platform' is visible.

Profile Kyong
Project administrator
Project developer
Project tester
Project scientist
Avatar
Send message
Joined: 9 Jun 12
Posts: 548
Credit: 52,283,184
RAC: 7,147
Message 973 - Posted: 11 Mar 2013, 17:47:20 UTC - in response to Message 969.

StandardbredHorse: Don't worry, I didn't mean it bad. I just added the platform and app and went sleep. :-D I am also curious how fast will be the computation on some newest mobile ARM based devices.

peeticek_LubosKrutek
Send message
Joined: 8 Jan 13
Posts: 31
Credit: 48,803,160
RAC: 20,737
Message 974 - Posted: 11 Mar 2013, 19:52:11 UTC

Congrats Kyong - i think, this is time for a small donation again :)

This is the first time, that im sad iOS user :)
- ehm, hello - any plans to create NativeBOINC for iOS? :):):):):)

p.

Profile ritterm
Avatar
Send message
Joined: 22 Jun 12
Posts: 18
Credit: 5,142,840
RAC: 34
Message 976 - Posted: 11 Mar 2013, 20:47:28 UTC - in response to Message 962.

Task 2328683 is my first WU...My guess is it will finish in about 18 hrs.

Well, I was close...Task 2328683 finished in just under 19 hrs. It's awaiting its wingman for verification and should get a whopping 6 cr/hr! :D

____________

1 · 2 · 3 · Next
Post to thread

Message boards : News : Releasing Android application for ARM processors


Main page · Your account · Message boards


Copyright © 2017 Asteroids@home