SETI@home v7 7.06 released for ARMv7 Android

Message boards : News : SETI@home v7 7.06 released for ARMv7 Android
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

AuthorMessage
Nvgnte
Volunteer tester
Avatar

Send message
Joined: 19 Oct 08
Posts: 6
Credit: 160,888
RAC: 0
Spain
Message 46651 - Posted: 21 Aug 2013, 8:17:39 UTC - in response to Message 46649.  

I've just sent out v7.08 for Android-armv6

Great! It seems to be working now :)
It checkpoints OK, and it finds no problem so far... but it's gonna take loooong time to finish it (0,5% done @ 2 hours running)
Task 14523568
Thanks!
Descarga gratuitamente y comparte mi primer eBook: Amaneceres
ID: 46651 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,440,339
RAC: 5
United Kingdom
Message 46652 - Posted: 22 Aug 2013, 5:57:08 UTC - in response to Message 46649.  

I've just sent out v7.08 for Android-armv6. I've rebuilt every library, so hopefully it is fully v6 compatible now. In addition, I've made it friendlier to small memory devices. On devices with less than 512MB, it will use the CPU capabilities to decide which functions to use rather than timing them. (Timing was memory intensive).

An added enhancement... the chosen functions on all platforms will be stored in the state file, so on restart it won't have to benchmark the functions again.

First run of the 7.08 app on my probablamatic HTC One S ended after 0 seconds and a SIGILL: illegal instruction, normally it runs for a little while before erroring.

All tasks for computer 64283

Claggy
ID: 46652 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46653 - Posted: 22 Aug 2013, 11:28:49 UTC - in response to Message 46622.  

No rush, my Droid 1 will still be working on its first result at that point.

Downloaded 7.08 on Samsing Galaxy mini (armv6). Now working. first % done increase passed OK. Will see if it can run till end now.
On G-tab v7.07 downloaded. Should this device recive 7.08 too or those plan classes mutually exclusive?
ID: 46653 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46654 - Posted: 22 Aug 2013, 15:35:47 UTC - in response to Message 46653.  

ARMv7 processors should get both versions for now. It seems likely that there will be little difference in speed, so we will probably just distribute the ARMv6 version on the main project.
ID: 46654 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46655 - Posted: 22 Aug 2013, 15:45:23 UTC - in response to Message 46652.  


First run of the 7.08 app on my probablamatic HTC One S ended after 0 seconds and a SIGILL: illegal instruction, normally it runs for a little while before erroring.


I'm not sure what's going on with 64283. It doesn't appear to be significantly different than other hosts that are running successfully. Is it able to run E@H OK? Is there anything unusual about the setup? Did you install to an SD card rather than to internal storage?
ID: 46655 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46656 - Posted: 22 Aug 2013, 20:02:26 UTC - in response to Message 46655.  

Did you install to an SD card rather than to internal storage?

for the record, on my Galaxy mini NativeBOINC configured to run from SD card. And app works OK.
ID: 46656 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46657 - Posted: 22 Aug 2013, 20:07:11 UTC

Hm... only 0.893% after 4h14min...
For comparison, third party app suspended on 0.299% after 10 mins.
It would be hard to collect enough statistics with such speed. More chances to get comparison on G-tab.
ID: 46657 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46658 - Posted: 22 Aug 2013, 22:00:24 UTC - in response to Message 46656.  

Did you install to an SD card rather than to internal storage?

for the record, on my Galaxy mini NativeBOINC configured to run from SD card. And app works OK.


Some BOINC devices mount their SD card with the "noexec" mount option, which could be a problem. Its not entirely clear to me whether that moves the run directory (/data/data/edu.berkeley.boinc/) to the SD card as well.
ID: 46658 · Report as offensive
JagDoc
Volunteer tester

Send message
Joined: 25 Aug 13
Posts: 17
Credit: 1,261,510
RAC: 275
Germany
Message 46662 - Posted: 25 Aug 2013, 14:50:25 UTC

On my Toshiba Excite Pro (Tergra4) Host ID: 64580 all WU with v7.08 end with:
process exited with code 193 (0xc1, -63)
</message>
<stderr_txt>
SIGILL: illegal instruction

On my Odroid U2 Host ID: 64579 the WUs run since 4h without problem.
3 x WU with v7.07 (11-14% after 4h)
1 x WU with v7.08 ( 4,8% aufter 4h)

cu JagDoc
ID: 46662 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46663 - Posted: 26 Aug 2013, 2:40:12 UTC - in response to Message 46662.  

That's a full ARMv7 processor, so it shouldn't be finding any illegal instructions, unless they've done something weird with their ARM implementation. I'll look at the RPCs tomorrow to see if anything looks strange.
ID: 46663 · Report as offensive
Nvgnte
Volunteer tester
Avatar

Send message
Joined: 19 Oct 08
Posts: 6
Credit: 160,888
RAC: 0
Spain
Message 46672 - Posted: 27 Aug 2013, 18:42:09 UTC - in response to Message 46645.  

My first task 7.08 (ARMv6) crunched OK :)

Task 14598571
Descarga gratuitamente y comparte mi primer eBook: Amaneceres
ID: 46672 · Report as offensive
JagDoc
Volunteer tester

Send message
Joined: 25 Aug 13
Posts: 17
Credit: 1,261,510
RAC: 275
Germany
Message 46673 - Posted: 27 Aug 2013, 18:55:31 UTC

Now my Toshiba Excite Pro (Tergra4) Host ID: 64580 got some WUs with v7.07 run without problem.

Same with the Nvidia Shield Host ID: 64582 , v7.07 no problem, v7.08 error.

cu JagDoc
ID: 46673 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,440,339
RAC: 5
United Kingdom
Message 46680 - Posted: 30 Aug 2013, 17:51:09 UTC

Finally found ARM illegal instruction encoding.

My HTC One S is still throwing 'process exited with code 193 (0xc1, -63) SIGILL: illegal instruction' with the v7.09 (armv6) app (released yesterday),

With the 7.07 (armv7a) app it still manages to complete a variable amount of time on the Wu before getting a 'SIGSEGV: segmentation violation', normally this is within a few % of start, but it does sometimes manage to get to ~70% before failing,

The Nexus 7 has no problem with any of the apps so far (It's not run the v7.09 (armv6) app yet)

Claggy
ID: 46680 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46681 - Posted: 30 Aug 2013, 18:35:19 UTC - in response to Message 46680.  

I'm going to see if I can figure out how to get a stack dump for SIGILL.
ID: 46681 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46683 - Posted: 31 Aug 2013, 11:03:27 UTC

@Eric
Look prev messages from Mateusz. He provided some logs from crashing app. Maybe needed info can be extracted from there?
No crashes on my phone (I'm running under NativeBOINC so "bug-catcher" should be available).

But I'm afraid 7.08 too slow vs custom app. >100 hours and only 20,1% done. And it's even not VLAR...
ID: 46683 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46684 - Posted: 31 Aug 2013, 11:08:00 UTC

And another (cosmetic) issue with stock android app:

<stderr_txt>
setiathome_v7 7.00 Revision: 1941 g++ (GCC) 4.4.6 20110731 (Red Hat 4.4.6-3)
libboinc: BOINC 7.2.0

Work Unit Info:
...............
WU true angle range is : 0.447883
Optimal function choices:
--------------------------------------------------------
name timing error
--------------------------------------------------------
v_BaseLineSmooth (no other)
vfp_GetPowerSpectrum 0.002403 0.00000
vfp_ChirpData 0.126063 0.00000
v_Transpose4 0.050126 0.00000
opt VFP folding 0.021716 0.00000

Flopcounter: 36584464432882.351562

Spike count: 0
Autocorr count: 0
Pulse count: 0
Triplet count: 0
Gaussian count: 1
12:14:17 (26557): called boinc_finish

</stderr_txt>


I'm not sure that bolded line is true. Especially "Red Hat".
ID: 46684 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46685 - Posted: 31 Aug 2013, 16:44:31 UTC - in response to Message 46684.  

Yes, that is undoubtedly wrong.
ID: 46685 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,981,656
RAC: 1,197
United States
Message 46686 - Posted: 31 Aug 2013, 17:36:25 UTC - in response to Message 46683.  

@Eric
Look prev messages from Mateusz. He provided some logs from crashing app. Maybe needed info can be extracted from there?
No crashes on my phone (I'm running under NativeBOINC so "bug-catcher" should be available).


Previous messages where?


But I'm afraid 7.08 too slow vs custom app. >100 hours and only 20,1% done. And it's even not VLAR...


I've downloaded Mateusz's source code, and the only significant difference with the stock app is which FFT he is using (FFTS vs FFTW). Unless he's not using the android NDK compiler.

Which host have you been using his app on?
ID: 46686 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46687 - Posted: 31 Aug 2013, 20:41:32 UTC

Also, looks like 7.08 MUCH slower than 7.07 on armv7 device capable to run both versions. My tablet running >70h already and only near 20%. Usually it takes much less time...
ID: 46687 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 46688 - Posted: 31 Aug 2013, 20:48:44 UTC - in response to Message 46686.  
Last modified: 31 Aug 2013, 20:53:18 UTC

@Eric
Look prev messages from Mateusz. He provided some logs from crashing app. Maybe needed info can be extracted from there?
No crashes on my phone (I'm running under NativeBOINC so "bug-catcher" should be available).


Previous messages where?



Hm... Can't find them. I thought I saw them in very this thread...

Which host have you been using his app on?

On SETI main:

http://setiathome.berkeley.edu/show_host_detail.php?hostid=6958030
http://setiathome.berkeley.edu/show_host_detail.php?hostid=6952254
http://setiathome.berkeley.edu/show_host_detail.php?hostid=6952249

Same hosts on beta (still running stock only):
http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=64312
http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=64311
http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=64393

G-tabs have 7.07 results already, Samsung mini can't complete its first 7.08 task still. But 7.08 takes much longer on G-tabs too and still no single complete result for 7.08.

EDIT: for first link on beta 7.06 took ~52h per task. Now that host running >70h and near 20 % only...

Both tablets just lie plugged and crunch, even WiFi disabled most of time...
ID: 46688 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 . . . 11 · Next

Message boards : News : SETI@home v7 7.06 released for ARMv7 Android


 
©2019 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.