| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: Outsmart Ebola Together
Thread: OET performance on Android devices |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 48
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Neither BOINC for Android or HTC Power to Give does any crunching... it's the science apps, AND it so happened we went into a set of batches that dropped the runtimes to 20-30 minutes under Linux. ;P)
|
||
|
|
vlado101
Senior Cruncher Joined: Jul 23, 2013 Post Count: 226 Status: Offline Project Badges: 
|
Ah, well that explains it.
----------------------------------------Thanks for the update SekeRob. Hopefully the android work units in the future become more optimized. I will still keep crunching those work units so that hopefully at least a couple of them go through.  |
||
|
|
NixChix
Veteran Cruncher United States Joined: Apr 29, 2007 Post Count: 1187 Status: Offline Project Badges: 
|
Ok this is it for me. I just aborted OET1_ 0001204_ xZAGP-F_ rig_ 64427_ 2-- that has been stuck at 100% for days, but it is impossible to tell with the useless output log. Every time I look at it the CPU time is 9 hours and something. From the device log it looks like the client keeps restarting. I'm just wasting my time, electricity, and likely battery life.
----------------------------------------Since the BOINC community and WCG don't take the Android platform seriously, I'm just going to abandon it. It's not ready for use. Do give a shout when you're serious about it. Cheers    |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Ok this is it for me. I just aborted OET1_ 0001204_ xZAGP-F_ rig_ 64427_ 2-- that has been stuck at 100% for days, but it is impossible to tell with the useless output log. Every time I look at it the CPU time is 9 hours and something. From the device log it looks like the client keeps restarting. I'm just wasting my time, electricity, and likely battery life. Since the BOINC community and WCG don't take the Android platform seriously, I'm just going to abandon it. It's not ready for use. Do give a shout when you're serious about it. Cheers It works just fine on the FAAH project on WCG. I've had OET units that sit at 100% for quite sometime and if you disconnect from the charger, when you plug it back in later, it is now less than 100% on the progress. I have given up on OET on all platforms. If you want to crunch on your Android, why not FAAH? Just use a profile for the Android devices that only has listed FAAH listed and make sure the checkbox the allows work from other projects is not checked. |
||
|
|
NixChix
Veteran Cruncher United States Joined: Apr 29, 2007 Post Count: 1187 Status: Offline Project Badges:
|
It works just fine on the FAAH project on WCG. I've had OET units that sit at 100% for quite sometime and if you disconnect from the charger, when you plug it back in later, it is now less than 100% on the progress. I have given up on OET on all platforms. If you want to crunch on your Android, why not FAAH? Just use a profile for the Android devices that only has listed FAAH listed and make sure the checkbox the allows work from other projects is not checked. IanBrown, I initially was running FAAH on my Android devices, but I moved them to OET after completing my goal of 5 years. I did not notice any problems on my Android devices with FAAH, but I did not start using Android until Mid-June just before I finished FAAH. Given the way that the problems I have just experienced with OET do not show up in the logs, the same thing may have ocurred with FAAH as well and I just didn't notice. Since BOINC and WGC are apathetic about the problems with the Android platform I'm just going to uninstall. Maybe some day in the future it will be usable. Also I did have 2 profiles similar to your suggestion: 14 Android cores were set to recieve OET and 12 regular cores were set to recieve UGM. The prospect of doubling my processing capacity was exciting. I am equally saddened by the loss.  Cheers |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
It is not a BOINC issue nor a WCG issue; but that of the project itself. The scientists behind the project are the ones that would need to straighten it out. OET uses AD VINA just like FAAH; which doesn't have these issues. SO if they both use AD VINA, then the issue is not with the software.
There were issues with OET early on, like checkpointing. |
||
|
|
enels
Senior Cruncher Joined: Apr 25, 2008 Post Count: 286 Status: Offline Project Badges: 
|
You have identified a problem, NixChix, that I don't know anybody has a solution for. Other than to stop suspending. These work for me:
----------------------------------------Pause when screen is on [off] Pause at CPU usage above [100%]. RAM limit [100%] Crunch using battery for brief times unplugged. Edit: corrected misspelling of NixChix [Edit 1 times, last edit by enels at Sep 14, 2015 3:19:37 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
You have identified a problem, NxiChix, that I don't know anybody has a solution for. Other than to stop suspending. These work for me: Pause when screen is on [off] Pause at CPU usage above [100%]. RAM limit [100%] Crunch using battery for brief times unplugged. The problem is, when it is sitting at 100% for hours on end, not suspending is not an viable option. This is a mobile device and if the project cannot play nicely with mobile devices, then it shouldn't be available to mobile devices. FAAH doesn't have these issues. I had an OET run for 5 hours before it got to 100% and then ran for another 5 hours before it was available to upload. Was it really at 100% and if so, what could it possibly do for 5 hours? I had another that say at 100% for hours and then when I had to unplug it from the charger, when it was plugged back in, it went back to 98%, so when it finally hit 100% again, once again it sat there for hours before being uploaded. OET in the beginning has checkpoint issues; it seems like it still does. Within those hours a WU sits at 100%; multiple checkpoints should have occurred. |
||
|
|
NixChix
Veteran Cruncher United States Joined: Apr 29, 2007 Post Count: 1187 Status: Offline Project Badges:
|
You have identified a problem, NxiChix, that I don't know anybody has a solution for. Other than to stop suspending. These work for me: Thanks enels.Pause when screen is on [off] Pause at CPU usage above [100%]. RAM limit [100%] Crunch using battery for brief times unplugged. The only setting that you have recommended that I hadn't tried is to allow 100% RAM usage. I didn't think that would come to a good end, so I had left it at 50%, but I did bump it up to 80% on an old device that I was able to dedicate to WCG. I does have less problems - maybe the increased RAM allowance helped. Even though I have allowed that one to continue crunching, I have observed oddities with it that would appear to be BOINC client related rather than the app. Unfortunately I don't have the available time at the present to tinker and experiment with the app. Cheers |
||
|
|
supdood
Senior Cruncher USA Joined: Aug 6, 2015 Post Count: 333 Status: Offline Project Badges:
|
A lot of this is simply agreeing with what has already been said, but I want to reiterate for the WCG techs for emphasis.
----------------------------------------I just switched my Androids over from FAAH and have to agree with NixChix: what is the point of crunching OET on Android (where the whole idea is to crunch overnight while you aren't using your device) when the WUs take 8-15+ hours with only 8 checkpoints? The only solution I've found is to crunch 24/7 which means letting the device drain battery during commutes and then be plugged in all day at work. This really isn't a practical use of a mobile device and can cause heat issues during the commute and recharge. I'm not sure why this was allowed out of Beta in its current state. All future projects with Android crunching need either more frequent checkpointing or quicker WUs (basically, just like FAHV). |
||
|
|
|