SETI@home 7.28 for ARM-Android released...

Message boards : News : SETI@home 7.28 for ARM-Android released...
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 18 · Next

AuthorMessage
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1547
Credit: 26,836,438
RAC: 1,577
United States
Message 50137 - Posted: 23 Mar 2014, 21:16:34 UTC

There are two versions. A version that supports NEON, and a "safe" version the just uses VFP instructions. Devices that support NEON could get either, although there should be a 3:1 preference for the neon version.

Please run whichever version you get, as cancelling results will skew the completion statistics.
ID: 50137 · Report as offensive
Profile David S
Volunteer tester
Avatar

Send message
Joined: 10 Sep 13
Posts: 1187
Credit: 2,511,129
RAC: 319
United States
Message 50141 - Posted: 23 Mar 2014, 22:25:36 UTC - in response to Message 50137.  
Last modified: 23 Mar 2014, 22:32:17 UTC

There are two versions. A version that supports NEON, and a "safe" version the just uses VFP instructions. Devices that support NEON could get either, although there should be a 3:1 preference for the neon version.

Please run whichever version you get, as cancelling results will skew the completion statistics.

Will let my phone get going immediately.

[edit]
Now running two, non-NEON.

Thanks, Eric.
David
signature sent back to alpha testing
ID: 50141 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,439,427
RAC: 0
United Kingdom
Message 50145 - Posted: 23 Mar 2014, 23:07:30 UTC - in response to Message 50137.  

My 2012 HTC One S with it's Qualcomm Snapdragon S4 is already stalling on the v7.28 (armv6-neon) apps, and requiring a suspend/resume to get them going again.

Claggy
ID: 50145 · Report as offensive
Profile David S
Volunteer tester
Avatar

Send message
Joined: 10 Sep 13
Posts: 1187
Credit: 2,511,129
RAC: 319
United States
Message 50146 - Posted: 24 Mar 2014, 0:15:45 UTC - in response to Message 50145.  

My 2012 HTC One S with it's Qualcomm Snapdragon S4 is already stalling on the v7.28 (armv6-neon) apps, and requiring a suspend/resume to get them going again.

Claggy

No problems here. 1h45m, 3.8%. Log is full of checkpoints.
David
signature sent back to alpha testing
ID: 50146 · Report as offensive
arkayn
Volunteer tester
Avatar

Send message
Joined: 16 Jan 07
Posts: 155
Credit: 194,400
RAC: 0
United States
Message 50147 - Posted: 24 Mar 2014, 2:39:38 UTC

It will be a while until my phone gets freed up from the Collatz WU, but I have taken it off NNT for a run on here again.
ID: 50147 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1628
Credit: 11,798,184
RAC: 181
Sweden
Message 50150 - Posted: 24 Mar 2014, 9:24:14 UTC

I will allow new work, when I get home after work. I haven't my charger with me today, because I did not think any nwew app would be released yet.
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 50150 · Report as offensive
Profile David S
Volunteer tester
Avatar

Send message
Joined: 10 Sep 13
Posts: 1187
Credit: 2,511,129
RAC: 319
United States
Message 50151 - Posted: 24 Mar 2014, 13:24:20 UTC

Seems like my two tasks are heading for about a 45 hour runtime. I'd do the math as it is right now, but I left the house in a hurry and forgot the d@mn thing.
David
signature sent back to alpha testing
ID: 50151 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1628
Credit: 11,798,184
RAC: 181
Sweden
Message 50156 - Posted: 24 Mar 2014, 16:19:56 UTC
Last modified: 24 Mar 2014, 17:09:39 UTC

OK home now, I'm on. Got myself a SETI@home v7 v7.28 (armv6-neon)

Added: Compared to v7.27, this version is a rocket on steroids :-)
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 50156 · Report as offensive
Dr Who Fan
Volunteer tester
Avatar

Send message
Joined: 6 Apr 11
Posts: 13
Credit: 186,742
RAC: 82
United States
Message 50163 - Posted: 25 Mar 2014, 8:04:48 UTC

Have 1 arm6-vpf task running under NativeBoinc on host 69846 (Nexus 7 tablet) now for 22h:10m - 47.325% done.

My calculations show it should have total run-time of ~46.83 Hrs.

ID: 50163 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,439,427
RAC: 0
United Kingdom
Message 50166 - Posted: 25 Mar 2014, 9:58:10 UTC

My 2012 Nexus 7 completed a v7.28 (armv6-neon) task last night, and validated.

The 2012 HTC One S continues to produce inconclusive and invalids with v7.28 (armv6-neon), and frequently stalls,
(lots of time on app, but progress has stopped, CPU-Z doesn't report both cores at max, and when you suspend Boinc,
tasks say suspending for just over 20 seconds, before changing to suspended, and Elapsed time drops back many hours)

Claggy
ID: 50166 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 50167 - Posted: 25 Mar 2014, 10:27:33 UTC
Last modified: 25 Mar 2014, 10:27:51 UTC

Eric, 7.27 task that ran offline finished finally on G-tab - no SIGSEGV exception at the end and full stderr preserved! so, my Gtabs get SIGSEGVs under NativeBOINC and under stock BOINC but not when app runs offline! Hope this info will be useful.
Now switching to 7.28 debugging.
ID: 50167 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 50168 - Posted: 25 Mar 2014, 10:53:04 UTC - in response to Message 50167.  

Also, worth to mention, there was no app kills through whole offline run (almost week). I played Majesty and browsed internet there, no low-memory conditions though.

I moved result back in slot and resumed that task from NativeBOINC. It resumed from last checkpoint, 99,98% and then finished OK, BUT (!) there was SIGSEGV now on exit! So, no SIGSEGV running offline, SIGSEGV running under BOINC. Clear proof:

Stderr output

<core_client_version>7.0.36</core_client_version>
<![CDATA[
<stderr_txt>
Unable to resolve function unwind_backtrace_signal_arch
Unable to resolve function acquire_my_map_info_list
Unable to resolve function release_my_map_info_list
Unable to resolve function get_backtrace_symbols
Unable to resolve function free_backtrace_symbols
Unable to resolve function format_backtrace_line
Unable to resolve function load_symbol_table
Unable to resolve function free_symbol_table
Unable to resolve function find_symbol
one or more symbols not found. stackdumps unavailable
setiathome_v7 7.27 $Revision: 1146 $ arm-linux-androideabi-g++ (GCC) 4.6 20120106 (prerelease)
libboinc: BOINC 7.3.1

Work Unit Info:
...............
WU true angle range is : 0.444884
Optimal function choices:
--------------------------------------------------------
name timing error
--------------------------------------------------------
v_BaseLineSmooth (no other)
v_GetPowerSpectrum 0.015335 0.00000
stackdumps unavailable
19:58:08 (22439): Can't set up shared mem: -1. Will run in standalone mode.
setiathome_v7 7.27 $Revision: 1146 $ arm-linux-androideabi-g++ (GCC) 4.6 20120106 (prerelease)
libboinc: BOINC 7.3.1

Work Unit Info:
...............
WU true angle range is : 0.444884
Optimal function choices:
--------------------------------------------------------
name timing error
--------------------------------------------------------
v_BaseLineSmooth (no other)
v_GetPowerSpectrum 0.020496 0.00000
v_ChirpData 0.855561 0.00000
v_Transpose4 0.046207 0.00000
FPU opt folding 0.123256 0.00000

Flopcounter: 36632555675524.375000

Spike count: 3
Autocorr count: 0
Pulse count: 5
Triplet count: 0
Gaussian count: 0
04:12:48 (22439): called boinc_finish

Unable to resolve function unwind_backtrace_signal_arch
Unable to resolve function acquire_my_map_info_list
Unable to resolve function release_my_map_info_list
Unable to resolve function get_backtrace_symbols
Unable to resolve function free_backtrace_symbols
Unable to resolve function format_backtrace_line
Unable to resolve function load_symbol_table
Unable to resolve function free_symbol_table
Unable to resolve function find_symbol
one or more symbols not found. stackdumps unavailable
Restarted at 99.98 percent.

Flopcounter: 36632816770998.125000

Spike count: 3
Autocorr count: 0
Pulse count: 5
Triplet count: 0
Gaussian count: 0
14:36:08 (25561): called boinc_finish
SIGSEGV: segmentation violation

Exiting...

</stderr_txt>
]]>

Here the link: http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=16230984
ID: 50168 · Report as offensive
Profile David S
Volunteer tester
Avatar

Send message
Joined: 10 Sep 13
Posts: 1187
Credit: 2,511,129
RAC: 319
United States
Message 50171 - Posted: 25 Mar 2014, 13:42:07 UTC - in response to Message 50137.  

There are two versions. A version that supports NEON, and a "safe" version the just uses VFP instructions. Devices that support NEON could get either, although there should be a 3:1 preference for the neon version.

So what are the four tasks my phone has gotten so far? Three VFPs and one NEON.

http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=6193973 finished in 24.1 hours Run time, 12.2 hours CPU time. All three of us got identical -9s.

My next task .... just finished while I was typing this at 35:41:47.
http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=6193823 All of us agree on 5 spikes and 1 Gaussian.

And I have just learned something: the time displayed while the task is still on the device is the Run time that ends up appearing on the web page, not the CPU time.

My 3rd 7.28 http://setiweb.ssl.berkeley.edu/beta/workunit.php?wuid=6197407 looks like it's heading for about 15:40 of Run time when it's done. One host has returned that WU so far and it's not a -9. I suppose I'll finish it during the outage, if there is one today.

.... Updating my initial comment, the phone has now downloaded a 5th task, and it's another VFP.

The early returns seem to be showing a winner here. Have to see what other devices think of it, of course, but from my perspective, get this puppy up on Main in place of 7.21.
David
signature sent back to alpha testing
ID: 50171 · Report as offensive
jason_gee
Volunteer tester

Send message
Joined: 11 Dec 08
Posts: 198
Credit: 658,573
RAC: 0
Australia
Message 50175 - Posted: 25 Mar 2014, 14:41:06 UTC

I'm seeing a similar weight to vfp at the moment, but even more 7.24s.
Chaos: When the present determines the future, but the approximate present does not approximately determine the future.
Edward Lorenz
ID: 50175 · Report as offensive
Dr Who Fan
Volunteer tester
Avatar

Send message
Joined: 6 Apr 11
Posts: 13
Credit: 186,742
RAC: 82
United States
Message 50181 - Posted: 25 Mar 2014, 22:34:07 UTC - in response to Message 50163.  

Have 1 arm6-vpf task running under NativeBoinc on host 69846 (Nexus 7 tablet) now for 22h:10m - 47.325% done.

My calculations show it should have total run-time of ~46.83 Hrs.

Result ID 16430458 finished sooner than I expected, total runtime of 34:08:50 and CPU run time of 33:32:45.

ID: 50181 · 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: 26,836,438
RAC: 1,577
United States
Message 50182 - Posted: 26 Mar 2014, 0:14:49 UTC - in response to Message 50181.  

So you can see how the apps are doing, I've built pages for them (and every other active app). Thus far only about 100 hosts (of the 900 or so out there) have gotten the new apps. Of those, have returned early errors. That's not surprising since most of the 900 machines out there are crunching a 7.27. The ones that were erroring out on 7.27 would be the first to get 7.28, and the most likely to return an error. It usually works out that way.

setiathome_v7_arm-android-linux-gnu__armv6-neon.html
setiathome_v7_arm-android-linux-gnu__armv6-vfp.html
setiathome_v7_arm-android-linux-gnu__armv7-vfpv4.html
setiathome_v7_x86-android-linux-gnu.html
setiathome_v7_windows_intelx86__opencl_intel_gpu_sah.html
astropulse_v6_windows_intelx86__opencl_intel_gpu_100.html
astropulse_v6_x86_64-apple-darwin__opencl_ati_mac.html
astropulse_v6_x86_64-pc-linux-gnu__cuda_opencl_100.html
astropulse_v6_x86_64-pc-linux-gnu__opencl_nvidia_100.html
astropulse_v6_x86_64-pc-linux-gnu__opencl_ati_100.html
astropulse_v6_windows_intelx86__cal_ati.html
astropulse_v6_windows_intelx86__opencl_ati_100.html
setiathome_v7_x86_64-pc-linux-gnu__opencl_ati5_cat132.html
setiathome_v7_x86_64-pc-linux-gnu__opencl_ati_cat132.html
setiathome_v7_x86_64-pc-linux-gnu__opencl_ati_sah.html
setiathome_v7_x86_64-pc-linux-gnu__opencl_ati5_sah.html
setiathome_v7_windows_intelx86__opencl_ati5_cat132.html
setiathome_v7_windows_intelx86__opencl_ati_cat132.html
setiathome_v7_windows_intelx86__opencl_ati5_sah.html
setiathome_v7_windows_intelx86__opencl_ati_sah.html
setiathome_v7_powerpc-apple-darwin.html
setiathome_v7_i686-apple-darwin.html
setiathome_v7_x86_64-pc-linux-gnu.html
setiathome_v7_i686-pc-linux-gnu.html
setiathome_v7_windows_intelx86.html
setiathome_v7_windows_intelx86__cuda50.html
setiathome_v7_windows_intelx86__cuda42.html
setiathome_v7_windows_intelx86__cuda32.html
setiathome_v7_windows_intelx86__cuda23.html
setiathome_v7_windows_intelx86__cuda22.html
astropulse_v6_windows_intelx86__cuda_opencl_100.html
astropulse_v6_windows_intelx86__opencl_nvidia_100.html
astropulse_v6_powerpc-apple-darwin.html
astropulse_v6_i686-apple-darwin.html
astropulse_v6_x86_64-pc-linux-gnu.html
astropulse_v6_windows_intelx86.html
astropulse_v6_i686-pc-linux-gnu.html
ID: 50182 · Report as offensive
jason_gee
Volunteer tester

Send message
Joined: 11 Dec 08
Posts: 198
Credit: 658,573
RAC: 0
Australia
Message 50183 - Posted: 26 Mar 2014, 1:06:57 UTC - in response to Message 50182.  
Last modified: 26 Mar 2014, 1:11:11 UTC

If I'm looking at the data right, I'm a little surprised to see the OpenCL and Cuda applications outstripping for success, stock PC host CPU applications in a lot of, perhaps most, cases.

Is there any easy way to drill into a given plan class (for the sakes of argument, windows x86 stock v7 MB CPU app) to characterise the failures at all ? For example what proportion of failures have truncated or missing stderr on outliers/overflows, which have been shown to trigger instant invalids ? (and is a known boincapi thread safety issue)
Chaos: When the present determines the future, but the approximate present does not approximately determine the future.
Edward Lorenz
ID: 50183 · Report as offensive
Profile David S
Volunteer tester
Avatar

Send message
Joined: 10 Sep 13
Posts: 1187
Credit: 2,511,129
RAC: 319
United States
Message 50184 - Posted: 26 Mar 2014, 1:48:15 UTC - in response to Message 50182.  

So you can see how the apps are doing, I've built pages for them (and every other active app). Thus far only about 100 hosts (of the 900 or so out there) have gotten the new apps. Of those, have returned early errors. That's not surprising since most of the 900 machines out there are crunching a 7.27. The ones that were erroring out on 7.27 would be the first to get 7.28, and the most likely to return an error. It usually works out that way.

setiathome_v7_arm-android-linux-gnu__armv6-neon.html
setiathome_v7_arm-android-linux-gnu__armv6-vfp.html

So if I'm reading it correctly, my phone is the only Samsung I605 that has gotten VFPs. 1/4 is in process and 3/4 have been returned as success.

I'm having more brain trouble parsing the table for Neons. It lists 2 hosts for that app, half of the hosts are in process, but all of the results are in process and no successes or errors have been returned. How can all of the results be in process if only half of the hosts are in process? (My single Neon is still in process.) I guess the question is, how did someone else's I605 get listed as a host for this app without being assigned a result for this app? Or am I totally misinterpreting it?
David
signature sent back to alpha testing
ID: 50184 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 2 Jul 13
Posts: 505
Credit: 3,688,332
RAC: 10
United States
Message 50186 - Posted: 26 Mar 2014, 2:06:23 UTC

I'm still waiting for more AstroPulses. I just added my Mac Mountain Lion host and decided not to add an app_info. Now if I could just get some tasks for astropulse_v6_x86_64-apple-darwin__opencl_ati_mac all would be well.
ID: 50186 · Report as offensive
arkayn
Volunteer tester
Avatar

Send message
Joined: 16 Jan 07
Posts: 155
Credit: 194,400
RAC: 0
United States
Message 50187 - Posted: 26 Mar 2014, 2:35:07 UTC

20 hours in, I am at 69%, this is with the NEON app. Most likely a shorty.
ID: 50187 · Report as offensive
1 · 2 · 3 · 4 . . . 18 · Next

Message boards : News : SETI@home 7.28 for ARM-Android released...


 
©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.