Rosetta and Android

Message boards : Number crunching : Rosetta and Android

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 . . . 6 · Next

AuthorMessage
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77525 - Posted: 30 Sep 2014, 14:06:16 UTC - in response to Message 77522.  

For now, I am not going to spend much time on our android version until they fix this issue.


What's next? Update server side? Avx/Avx2? :-)
ID: 77525 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 77529 - Posted: 30 Sep 2014, 21:20:21 UTC - in response to Message 77525.  

For now, I am not going to spend much time on our android version until they fix this issue.


What's next? Update server side? Avx/Avx2? :-)


Probably server updates including software and hardware. Also, there's been some recent large scale code changes/refactoring of Rosetta so our next application update may not be trivial.
ID: 77529 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Coleslaw
Avatar

Send message
Joined: 24 Mar 07
Posts: 5
Credit: 4,255,355
RAC: 2,871
Message 77531 - Posted: 1 Oct 2014, 14:11:05 UTC
Last modified: 1 Oct 2014, 14:12:22 UTC

It is disappointing that SETI is the only project with an x86 Android app. I am hoping to see more project get these because it is only a matter of time before Google offers a true 64bit version. Being a Linux variant, it should see many of the same gains as other Linux distros at various projects. Now, I know many will say to just use the other Linux distros. However, Android is the only OS to have a version of BOINC that can auto update and do so easily. This is a huge feature in my eyes as I like to keep my clients somewhat up to date. With AMD starting to offer ARM cores in their upcoming servers, it will just be a matter of time before more people consider this as an option. I would run it from a VM as I have done in the past.

Just my $.02
ID: 77531 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77533 - Posted: 2 Oct 2014, 6:34:40 UTC - in response to Message 77529.  

Probably server updates including software and hardware.


That's great!

P.S. Please, stop android task on ralph and start new "standard" cpu test....

ID: 77533 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
nick n
Avatar

Send message
Joined: 26 Aug 07
Posts: 49
Credit: 219,102
RAC: 0
Message 77540 - Posted: 4 Oct 2014, 3:00:31 UTC
Last modified: 4 Oct 2014, 3:01:53 UTC

This is more of a Rosetta issue than an Android issue. I have had my old quad core Nexus 7 on the simap project for several weeks with no problems at all. I have had 0 work unit failures with over 100+ successes, so again don't blame android or the android version of boinc.

http://boincsimap.org/boincsimap/show_host_detail.php?hostid=938423
ID: 77540 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77548 - Posted: 6 Oct 2014, 6:40:19 UTC - in response to Message 77540.  

so again don't blame android or the android version of boinc.


I don't blame use of android version of boinc in general. For some protocols (some projects) is good enough, but not for rosetta. Android version has great potential, but up to now is only marginal for computational power.

ID: 77548 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 77550 - Posted: 6 Oct 2014, 17:56:05 UTC - in response to Message 77540.  

This is more of a Rosetta issue than an Android issue. I have had my old quad core Nexus 7 on the simap project for several weeks with no problems at all. I have had 0 work unit failures with over 100+ successes, so again don't blame android or the android version of boinc.

http://boincsimap.org/boincsimap/show_host_detail.php?hostid=938423


There are some older and various other versions that are working well with Rosetta, even with less RAM. But the majority are not. Can you try the Rosetta project on your phone? It may be one of those that work ok. Not all android versions/devices are equal. I have the same issue running Seti@h on my test device also. On my test devices, the background tasks (Rosetta and Seti) go to sleep and do not use much CPU when it should. This is a feature of Android to save power, i.e. not to use power/cpu when the user is not actively interacting with the task. I truly believe the BOINC client should take this into account and make sure the tasks use as much CPU as possible (keeping the user preference settings in mind). I am not familiar enough with Android to even know if this is possible (but it must be, like playing music).

I do agree that the majority of errors are likely due to large arrays used by Rosetta, but they are not that big (<10MB) and should not be hitting a heap space limit IMO. These errors are occurring somewhat randomly and only occur with the boinc client api on my test device. And I believe the CPU sleep issue is compounding this heap space issue.
ID: 77550 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 77552 - Posted: 6 Oct 2014, 18:04:18 UTC - in response to Message 77548.  

so again don't blame android or the android version of boinc.


I don't blame use of android version of boinc in general. For some protocols (some projects) is good enough, but not for rosetta. Android version has great potential, but up to now is only marginal for computational power.


I agree, android is not yet up-to-par when running Rosetta. There's a 11x performance loss when running the standard abinitio-relax folding protocol on a 96 residue protein sequence, compared to our in house machines. The average float and int ops BOINC cpu benchmark values of active hosts compared to Windows is around 2.4x and 1.7x less, respectively. I'm guessing the rosetta performance loss is mainly from memory bandwidth but we'll have to profile to see what is so slow.

But the true metric, IMO should be not just CPU throughput but CPU throughput/energy consumption. I don't really have an idea of what that value might be.
ID: 77552 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
cristipurdel

Send message
Joined: 1 Oct 06
Posts: 2
Credit: 56,335
RAC: 0
Message 77568 - Posted: 9 Oct 2014, 17:49:01 UTC - in response to Message 77552.  
Last modified: 9 Oct 2014, 17:49:40 UTC

so again don't blame android or the android version of boinc.


I don't blame use of android version of boinc in general. For some protocols (some projects) is good enough, but not for rosetta. Android version has great potential, but up to now is only marginal for computational power.


I agree, android is not yet up-to-par when running Rosetta. There's a 11x performance loss when running the standard abinitio-relax folding protocol on a 96 residue protein sequence, compared to our in house machines. The average float and int ops BOINC cpu benchmark values of active hosts compared to Windows is around 2.4x and 1.7x less, respectively. I'm guessing the rosetta performance loss is mainly from memory bandwidth but we'll have to profile to see what is so slow.

But the true metric, IMO should be not just CPU throughput but CPU throughput/energy consumption. I don't really have an idea of what that value might be.


Glad to found an answer as to why some WUs are failing on my androids.

I agree that the performance/power is the true metric to watch out for.

I have around 6-7 androids running boinc, and probably I will add 4-5 more, since I find the crunching during charging... priceless ...
ID: 77568 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77576 - Posted: 14 Oct 2014, 9:13:01 UTC - in response to Message 77529.  

Probably server updates including software and hardware. Also, there's been some recent large scale code changes/refactoring of Rosetta so our next application update may not be trivial.


Any roadmap??
ID: 77576 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile David E K
Volunteer moderator
Project administrator
Project developer
Project scientist

Send message
Joined: 1 Jul 05
Posts: 1018
Credit: 4,334,829
RAC: 0
Message 77586 - Posted: 17 Oct 2014, 18:45:43 UTC - in response to Message 77576.  

Probably server updates including software and hardware. Also, there's been some recent large scale code changes/refactoring of Rosetta so our next application update may not be trivial.


Any roadmap??


Not specifically. I heard we will get some funding for hardware upgrades. Not sure about the timeline but once we get the hardware, we'll look into the software and coordinate the update.
ID: 77586 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77636 - Posted: 10 Nov 2014, 16:47:03 UTC - in response to Message 77586.  

Not sure about the timeline but once we get the hardware, we'll look into the software and coordinate the update.


http://boinc.berkeley.edu/trac/wiki/SourceCodeGit
This can help you??
Staying Current

The local git repository which you created when you cloned the remote repository has a complete snapshot of development at the time it was cloned. You can use this to build the BOINC software, and you can even make local changes to your local copy. Meanwhile the BOINC developers will be making changes to the remote repository as they continue working on the software. To see what changes have been made there since you cloned the repository you must first "fetch" the changes from that repository. These will go into a branch called "origin/master", which is not the current working branch of your repository (which by default will be the "master" branch). So to compare your local copy with the remote, follow this example:

$ cd boinc
$ git fetch
$ git diff --name-status master..origin/master
M client/app.cpp
M client/app_config.cpp
M client/client_types.h
M client/cpu_sched.cpp
M client/cs_prefs.cpp
M client/cs_scheduler.cpp
M client/result.cpp
M client/result.h
M client/rr_sim.cpp
M client/sim.cpp
M clientgui/MainDocument.cpp
M doc/versions.inc
M html/user/get_passwd.php
M html/user/server_status.php
M lib/coproc.h

The "--name-status" flag causes only the name and status of differing files to be displayed. The "master..origin/master" specification causes the difference to be displayed between your local repository (the "master" branch) and the remote repository (as fetched onto the local "origin/master" branch).

If you wish to update your local branch to match the remote repository then simply use the git "pull" command, as in this example:

$ git pull
Updating 6790085..2bb3e74
Fast-forward
client/app.cpp | 2 +-
client/app_config.cpp | 12 ++
client/client_types.h | 2 +-
client/cpu_sched.cpp | 12 +-
client/cs_prefs.cpp | 5 +-
client/cs_scheduler.cpp | 15 +-
client/result.cpp | 7 +-
client/result.h | 3 +
client/rr_sim.cpp | 4 +-
client/sim.cpp | 6 +-
clientgui/MainDocument.cpp | 12 +-
doc/versions.inc | 380 ++++++++++++++++++++++----------------------
html/user/get_passwd.php | 101 ++++++------
html/user/server_status.php | 38 ++++-
lib/coproc.h | 5 +
15 files changed, 332 insertions(+), 272 deletions(-)

You could also use the "git merge" command, which would merge the changes from your local copy of the "origin/master" branch into your local "master" branch. The "git pull" command is shorthand for "git fetch" followed immediately by "git merge".

ID: 77636 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
cristipurdel

Send message
Joined: 1 Oct 06
Posts: 2
Credit: 56,335
RAC: 0
Message 77733 - Posted: 14 Dec 2014, 12:22:16 UTC

After reading the previous threads, it is still not clear to me how much RAM the current app is requesting.

Now my Galaxy 1 is saying that it needs 488.28 MB RAM and 808.70 MB disk space.
Are these numbers accurate or is just my phone requesting too much?

When I am checking http://wuprop.boinc-af.org/results/ram.py?plateforme=all&tri=2&sort=desc it says that Rosetta Mini should require 62.4 MB RAM and 338.1 MB disk space.

Why is the discrepancy so high?


ID: 77733 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 77736 - Posted: 15 Dec 2014, 18:07:28 UTC

Any figures on a phone app would be highly subject to change, as the development efforts there were limited to a few specific types of tasks etc. So, I would suspect that the figures they observed when making the table have now changed. The phone doesn't structure the request, BOINC Manager does this based upon information sent with the tasks.
Rosetta Moderator: Mod.Sense
ID: 77736 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sid Celery

Send message
Joined: 11 Feb 08
Posts: 2042
Credit: 40,322,419
RAC: 19,654
Message 77742 - Posted: 21 Dec 2014, 2:26:13 UTC

Now that Simap has ended, I've added Rosetta to my Samsung Galaxy S4. Awaiting tasks to come through. Nothing over the last week, so only WCG going through now.
ID: 77742 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Chilean
Avatar

Send message
Joined: 16 Oct 05
Posts: 711
Credit: 26,694,507
RAC: 0
Message 77748 - Posted: 22 Dec 2014, 5:34:56 UTC - in response to Message 77586.  

Probably server updates including software and hardware. Also, there's been some recent large scale code changes/refactoring of Rosetta so our next application update may not be trivial.


Any roadmap??


Not specifically. I heard we will get some funding for hardware upgrades. Not sure about the timeline but once we get the hardware, we'll look into the software and coordinate the update.


Any news on this?
ID: 77748 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
alex

Send message
Joined: 21 Dec 14
Posts: 8
Credit: 2,668,966
RAC: 0
Message 77749 - Posted: 22 Dec 2014, 7:46:48 UTC - in response to Message 77742.  

Now that Simap has ended, I've added Rosetta to my Samsung Galaxy S4. Awaiting tasks to come through. Nothing over the last week, so only WCG going through now.


Hi,
I've joined Rosetta yesterday, mainly because SIMAP ends and I was looking for a replacement project.
I have added 2 devices, not getting any wu until now.
The messages say: No work sent, Rosetta is not available for your type of cpu.
The application page here says, there is an android version.
Can anyone explain that?

Alexander
ID: 77749 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Mod.Sense
Volunteer moderator

Send message
Joined: 22 Aug 06
Posts: 4018
Credit: 0
RAC: 0
Message 77752 - Posted: 22 Dec 2014, 20:05:42 UTC

There is an android version, but it is only used for very limited numbers of tasks. So, at present, there must not be any tasks created for android.
Rosetta Moderator: Mod.Sense
ID: 77752 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
alex

Send message
Joined: 21 Dec 14
Posts: 8
Credit: 2,668,966
RAC: 0
Message 77753 - Posted: 22 Dec 2014, 20:26:19 UTC - in response to Message 77752.  

There is an android version, but it is only used for very limited numbers of tasks. So, at present, there must not be any tasks created for android.


THX,
I will keep waiting for wu's!

Alexander
ID: 77753 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile [VENETO] boboviz

Send message
Joined: 1 Dec 05
Posts: 1924
Credit: 9,040,283
RAC: 9,082
Message 77833 - Posted: 16 Jan 2015, 19:09:11 UTC - in response to Message 77529.  
Last modified: 16 Jan 2015, 19:10:21 UTC

Probably server updates including software and hardware. Also, there's been some recent large scale code changes/refactoring of Rosetta so our next application update may not be trivial.


Not specifically. I heard we will get some funding for hardware upgrades. Not sure about the timeline but once we get the hardware, we'll look into the software and coordinate the update.


Any news about server upgrade??
ID: 77833 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Previous · 1 · 2 · 3 · 4 · 5 . . . 6 · Next

Message boards : Number crunching : Rosetta and Android



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