+- +-
Say hello if visiting :) by Gecko
11 Jan 2023, 07:43:05 pm

Seti is down again by Mike
09 Aug 2017, 10:02:44 am

Some considerations regarding OpenCL MultiBeam app tuning from algorithm view by Raistmer
11 Dec 2016, 06:30:56 am

Loading APU to the limit: performance considerations by Mike
05 Nov 2016, 06:49:26 am

Better sleep on Windows - new round by Raistmer
26 Aug 2016, 02:02:31 pm

Author Topic: Report on new optimized Astropulse apps for Windows  (Read 56025 times)

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #15 on: 22 Nov 2008, 05:36:16 am »
I think Joe needs the most bone pointing training,   Skilled as he is in the ways that matter, placing curses is certainly not his forte.  Yourself and Joe should refer to the 'HowstuffWorks' site , article on 'Pointing the Bone'

Offline Richard Haselgrove

  • Messenger Pigeon
  • Knight who says 'Ni!'
  • *****
  • Posts: 2819
Re: Report on new optimized Astropulse apps for Windows
« Reply #16 on: 22 Nov 2008, 05:39:39 am »
There's an even fuller description at www.everything2.org

Maybe Joe sees himself more as the Nangarri?
« Last Edit: 22 Nov 2008, 05:43:24 am by Haselgrove »

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #17 on: 22 Nov 2008, 05:42:18 am »
Ha!, now that's a good one!.  Just rememeber this stuff's the real deal and not for kids to play with  ;D

Offline Richard Haselgrove

  • Messenger Pigeon
  • Knight who says 'Ni!'
  • *****
  • Posts: 2819
Re: Report on new optimized Astropulse apps for Windows
« Reply #18 on: 22 Nov 2008, 05:58:09 am »
The high incidence of overflows ...
Getting back (almost) on topic ...

It's interesting that there's such a high incidence of overflows at Main: whereas at Beta, I've only had that one short result, against 50 (so far) full-length ones at Beta. It's another weakness of the Beta testing paradigm that they tend to load a single 'tape' and let it split to the end, rather than taking shorter segments from a variety of data sources. Murphy's Law dictates that the sample tape chosen often turns out not to be a very interesting one.

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #19 on: 22 Nov 2008, 06:06:37 am »
mmmm.  I'd have to put that into the 'not our problem' category... On the wacko extreme interpretation of things you see it is entirely possible the solar system is surrounded by rapidly rotating pulsars or otherwise an approaching invasion fleet uses spread spectrum technology for its communications... Of course the algorithm could just be wrong, but I'm no phd, so I'll leave that to the experts.

Offline Richard Haselgrove

  • Messenger Pigeon
  • Knight who says 'Ni!'
  • *****
  • Posts: 2819
Re: Report on new optimized Astropulse apps for Windows
« Reply #20 on: 22 Nov 2008, 08:52:17 am »

@All testers
Please keep eye on indices.txt in working directory (if it become non-zero length or not).
And if you keep result file, please, keep indices.txt too along with result and WU files.


ap_03no08af_B2_P1_00143_20081121_30048 (1065721003)
ap_03no08ag_B2_P1_00100_20081121_20392 (1066329522)

- both just started - have chunky indices.txt files (attached). If I'm reading the guesstimates right, these both have a high chance of pulsing out before their allotted span. Have suspended networking for a result capture.

[attachment deleted by admin]

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Report on new optimized Astropulse apps for Windows
« Reply #21 on: 22 Nov 2008, 10:11:06 am »
Fine, thanks!
It would be usefult to get these WU + results + final indices.txt in their final state. To have some reference point.

"At sample 1024, data_chunk_now 256, strength 13.369141 > 12.100000"
So, we can see what limit established for pulses. If pulse is bigger it should be quenched ...

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #22 on: 22 Nov 2008, 10:23:29 am »
NoNo, not Pulse I think... sample... Big difference... Correct me if I am wrong please.

Leaps-from-Shadows

  • Guest
Re: Report on new optimized Astropulse apps for Windows
« Reply #23 on: 22 Nov 2008, 10:44:09 am »
Leaps - I don't want to burst your bubble, but Beta is currently overpaying credit - the tracking average was skewed by a disasterously slow v4.37 build. Main is paying as near as dammit 760 cobbles, making Cruiser's AP rate ~0.00807454. Still respectable, but not such an outright winner.

I won't be commenting on this at Main: the difference isn't all that significant, and it's nice to see someone saying something nice about AP credit for a change.
Don't worry ... I figured that Beta would have a different pay scale, which is why I pointed out that it was a work unit from Beta.  I'll be adding new posts to all the threads with updated results for all the v5 work units completed for Main and Beta, so people can see the differences.  I've got 11 more in my queues (six for Main, five for Beta), plus the last v4.36-branded work unit being crunched with the v4.37 SSE3 EPF app.

Some questions from a new tester...

I have to work in a couple of hours, and the v4.36-branded work unit will complete while I'm gone.  If I have it set for no network activity, will the results files be preserved?  If they are preserved ... then where?  The slots directories?

Looking in the slots directories, I can already see a couple of non-zero indeces.txt files.  Will these be preserved as well, or overwritten by new ones when new work units start?

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #24 on: 22 Nov 2008, 10:50:39 am »
I have to work in a couple of hours, and the v4.36-branded work unit will complete while I'm gone.  If I have it set for no network activity, will the results files be preserved?  If they are preserved ... then where?  The slots directories?

Looking in the slots directories, I can already see a couple of non-zero indeces.txt files.  Will these be preserved as well, or overwritten by new ones when new work units start?

- Yes, they'll stay there with network activity disabled.
- Yes the files will be preserved
- WU file will be in the project folder
- corresponding results and indices.txt will be in the slot (I think  ;D)
- What we are mostly after is those that early exit.  If you can keep/copy the WU file, result and indices, then report the file, the result page online will have useful info too, I find making a text file with the result web address at the top, followed by entire task page copy/pasted, is very useful as it has stderr contents.

[Note: If you have a 'nice' early exit, say <50% progress, then I'll point you to a FTP place to deposit the files]

[EDIT: example stderr & details text file attached to clarify  ;D]
jason


[attachment deleted by admin]
« Last Edit: 22 Nov 2008, 10:55:31 am by Jason G »

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Report on new optimized Astropulse apps for Windows
« Reply #25 on: 22 Nov 2008, 11:41:40 am »
NoNo, not Pulse I think... sample... Big difference... Correct me if I am wrong please.
Don't know what "sample" is then.
Sample can be (0,0), (0,1), (1,0), (1,1). How it can be more than 12 ?
Probably it's power[] sample, so - pulse.
But sure need to check with actual code...

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #26 on: 22 Nov 2008, 12:09:12 pm »
Right, polarised signal already decoded,  complex signal  partially blanked.

Comparison:
- A single sample amplitude variation = AM radio (benefit = simple, Disadvantage = short range, noisy )
- peak moving in frequency per time interval = FM radio (benefit = longer range, disadvantage = complex decoding)
- broadband pulses = primitive marconi/morse code telegraph style( very high efficiency/ low power because part of signal can be enough to sufficiently detect, because other interferece will only mask a small part.  Long Range. Highly illegal now, because they saturate all neighbouring frequencies. (this would mess up things like telescopes and police radio  ;D) [Square wave shape at source]

So broadband pulse is not one high sample:
   - Fundamental frequency + harmonics of the fundamental in constructive interference produce the pulse. [ multiple sine waves]
   - shape is determined by bandwidth + power [ impure square wave]

Now shift the signal source out 100 light years [ but stationary] :
   - different frequency components are 'dispersed' at different rates, so phase alignment is 'stuffed'
   - square wave scattered now looks like noise
   - dedispersion reconstitutes square wave through constructive interference generating highest peak.

Now put signal on orbiting planet:
   - Doppler effect causes chirp and its reverse as planet approaches and accelerates away.
   - could be coming toward us too.
   - Now dispersed signal is also chirped, so we need dedisperson and dechirp....  :o... big cpu cycles

A pulse is not a AM signal peak, but a combination of sine wave frequencies added together.  What we detect will only ever be approximations, however, we can do this approximation on many more places/times in the sky than anybody before in history.  The worst radar removal could do is remove the 'guts' of a pulse, which we would probably detect anyway.  My 'feeling' is the abundance of overflows we get for now are either radar blanking artefacts, or genuine interstellar sources.

Jason

   

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Report on new optimized Astropulse apps for Windows
« Reply #27 on: 22 Nov 2008, 12:19:51 pm »
My 'feeling' is the abundance of overflows we get for now are either radar blanking artefacts, or genuine interstellar sources.

Jason
 

Or wrong threshould value ;)
sngle pulse finding algorthm compares power with threshould.
"
if (power[m] > state.thresh[l]) {
"
So, at least this simplest form of pulse is power[] elment. (next round - coadded power[] elements, next - coadded and summarized through different data chunks in FFA).

Offline Jason G

  • Construction Fraggle
  • Knight who says 'Ni!'
  • *****
  • Posts: 8980
Re: Report on new optimized Astropulse apps for Windows
« Reply #28 on: 22 Nov 2008, 12:23:46 pm »
Don't be too quick, Better too low a threshold than too high  ;).  I reckon some of these may be early human transmissions bouncing around out there too,  Ever heard of the Queen Mary phenomenon ?

Offline Raistmer

  • Working Code Wizard
  • Volunteer Developer
  • Knight who says 'Ni!'
  • *****
  • Posts: 14349
Re: Report on new optimized Astropulse apps for Windows
« Reply #29 on: 22 Nov 2008, 12:36:42 pm »
Don't be too quick, Better too low a threshold than too high  ;).  I reckon some of these may be early human transmissions bouncing around out there too,  Ever heard of the Queen Mary phenomenon ?

What is queen Mary?

 

Welcome, Guest.
Please login or register.
 
 
 
Forgot your password?
Members
Total Members: 97
Latest: ToeBee
New This Month: 0
New This Week: 0
New Today: 0
Stats
Total Posts: 59559
Total Topics: 1672
Most Online Today: 33
Most Online Ever: 983
(20 Jan 2020, 03:17:55 pm)
Users Online
Members: 0
Guests: 27
Total: 27
Powered by EzPortal