theSkyNet POGS

Gilthanis

Gilthanis

[H]ard|DCer of the Year - 2014
Joined
Jan 29, 2006
Messages
8,735
I have re-opened the topic of optimized apps and GPU apps at the project. It is now September and we are no further informed than we were when we were running our own.

http://pogs.theskynet.org/pogs/forum_thread.php?id=476&postid=3400#3400

Kevin asked folks to stop running the optimized apps in April, so by now there should be some kind of progress.

Edit: I created an account called HardOCPtest so that anyone whom wanted to test or contribute anonymously could do so.

•to attach a computer to your account without using the BOINC Manager. To do so, install BOINC, create a file named account_pogs.theskynet.org_pogs.xml in the BOINC data directory, and set its contents to:
<account>
<master_url>http://pogs.theskynet.org/pogs/</master_url>
<authenticator>26329_d70d2a7f7241a167e9e73e6603db3bd7</authenticator>
</account>

teamdaily.php
 
Last edited:
I hit my 1 million point goal so I am now pulling my systems from this project until they release the GPU app.
 
Hi All,

We’re planning an upgrade for POGS in the next few months that will see us add more calculations to the suite already conducted on each galaxy image, add more data types (such as higher resolution images from the WISE telescope, http://www.nasa.gov/mission_pages/WISE/main/, which collects infrared light) and much, much more.

There are some big changes, and some smaller ones, included in the upgrade. We’ve given a short summary of everything below, but if you’d like any more specific detail or have questions then please let us know.

As part of the upgrades we intend to run some simulated galaxies through the upgraded POGS code as a check on the accuracy of our the data processing. These galaxy images are created by a computer to mimic certain characteristics. Once the fake galaxies are processed by the upgraded POGS ‘spectral energy distribution’ fit code, we can compare the results to the values we used when creating the galaxies to confirm the upgraded code works as expected and is producing accurate science results for the astronomers.

We’re also upgrading the data retrieval pipeline, continuing to get enhanced infrared images from the WISE telescope for all galaxies that have already been processed, as well as some more galaxy data from other new sources. This new data will broaden our results on the galaxies that have been already processed - helping the astronomers to see more of the bigger picture, so to speak.

Kevin and the team will also upgrade the input data processing pipeline as well. This will include improving the smoothing of input images. They will also develop ‘parameter prior cubes’, sets of values that should be used in calculations on each galaxy image. This will stop us calculating every possible combination of parameters on every galaxy — saving time and resources. This also means we will be changing the credit system from the current points-per-pixel based model, as with ‘priors’ all pixels are no longer equal. We will let you know when the credit change will occur and will do our best to make sure it’s equitable and results in approximately the same amount of credit you currently earn.

We will also be upgrading the output processing to produce better results images. This will include things like stellar masking (from catalogs, or directly from the images) to remove foreground stars from the calculations, and we also start using a code called ‘galfit’ (http://users.obs.carnegiescience.edu/peng/work/galfit/galfit.html) to give a mass surface density of the galaxies (a new calculation) and also try another code called ‘CIGALE’ (http://cigale.lam.fr/) and others on the data.

All of this comes together to mean better science output from POGS - something we’re really keen on and we hope you are too!

--TheSkyNet Team
Click to expand...

http://pogs.theskynet.org/pogs/forum_thread.php?id=506#3591
 
Hopefully they will use my last fortan commit as a base, so some of my code optimizations will make it into the "real" world. I see on github that they have some Python code improving the source to work units already.
 
Here are a few snippits from an email they sent this evening to their contributors.

Helpdesk Holiday Shutdown
From Christmas Eve (24th of December) through to January 5th, theSkyNet helpdesk and forums will be unattended. We'll still see any Facebook messages you share with us and may be able to help that way - but if you get in touch for our help over the holidays via email or the forums we'll get back to you ASAP after the 5th of January.

TheSkyNet POGS Upgrades

After a very successful visit to Perth, Western Australia last month, our collaborator (and all round nice guy) Dave Thilker from Johns Hopkins University has a lot on his plate working with ICRAR's Kevin Vinsen to upgrade theSkyNet POGS. They're planning to upgrade almost everything POGS does, from the code your computers are using to process galaxy images through to new-and-improved galaxy images from more telescopes than have been involved in POGS before.

There'll even be a new credit system to go with the upgrades as they move to a more efficient method of processing the thousands of galaxy images we have ready to go. More details on the upgrades are available on the POGS forums, where you can also ask your questions and chat to Kevin/Dave about the changes.

We also have a student intern, Sam, working with Kevin on upgrading the speed of work unit generation so we can keep up with challenges held by BOINC teams on POGS. Sam is also improving the system that logs all the events that happen as theSkyNet POGS processes so Kevin and the team can check the system quickly and easily.

We're looking forward to the updated POGS system, but it may necessitate some down time in coming months. We'll keep you fully updated (with warning) when we plan to pause processing for the upgrades.
Click to expand...

There was more, but this is what I felt was probably the most important part.
 
GPU work

We know we have some very keen users who would like to use theSkyNet POGS on their GPUs. We&#8217;ve even had some volunteer work from community members along these lines - thank you! However, we&#8217;ve put the GPU work on hold for two reasons:
1) Currently we don&#8217;t think we generate data fast enough to feed GPUs. We&#8217;re actually on the edge keeping the pipeline filled at the moment, and GPUs would be an order of magnitude increase we don&#8217;t have resources for.
2) We have no funding to investigate the differences in the results from the prototype GPU code and the original Fortran 77 code. The original author of the MAGPHYS code is visiting ICRAR next month and we&#8217;ll discuss the issues of porting it to &#8216;C&#8217;, but we don&#8217;t want to waste your time processing results that may not be correct.

Virtual Box

To get a good fit for the galaxies, we really would like to try many other codes (some of which only run on *nix). I&#8217;ll then use Bayesian Ensemble methods to improve the fit we get from the data by running multiple codes. These codes are written in a variety of languages including python, and they&#8217;re under constant development. I want to be able to just deploy the code and let it run, rather than having to spend precious time and resources converting updates constantly.

The Virtual Box version will allow me to do that as I can have all the libraries precompiled in the VB file and I download the latest executables (as sticky files so you only download the once) and run all of them against all the pixels you have downloaded. The VB will be an automatic process within BOINC, you won&#8217;t have to do anything extra, and it will mean more science bang for our bucks.

The stand alone executable will remain as we have a solid Android user base and they can&#8217;t use VB, so there will be an option for standalone processing for those that don&#8217;t want to use the VB option.
____________
Regards
Kevin
-----
International Centre for Radio Astronomy Research
Click to expand...

http://pogs.theskynet.org/pogs/forum_thread.php?id=506#3698
 
Last edited:
Good news everybody!
We've got some new data from our friends at Johns Hopkins with a whole raft new improvements and additions.
What does this mean?
We still have to test the new data with the new clients to make sure everything goes according to plan which we would love your help with. First we&#8217;ll do some in house testing over the weekend but (hopefully) on Monday I&#8217;ll be putting out a call to volunteers to help us test some of this new data, including the brand new android 5 client.
Note: due to the setup of the test server we won&#8217;t be maintaining credit which may or may not disappear without warning.
____________
Cheers
Alex
theSkyNet.org webMaster
ICRAR
Click to expand...
http://pogs.theskynet.org/pogs/forum_thread.php?id=563#3972
 
I've thrown a quad socket box at this Beta to see how things go. The WUs are... long. They started out at 13h:51m; however have a WU at ~45% complete with 1h;25m elapsed and is suggesting it still has 7h:33m to go. Will keep an eye on this and see just how long it really takes. (E5-4640 @ 2.4Ghz, no hyper-threading)
 
Thankfully the initial numbers lied big time. ;)

Here are the results for the first 16 WUs; in seconds of course.

Code: 
	Run Time   CPU Time
Min	11,189	   11,071
Max	12,126	   11,995
Avg	11,783	   11,682

So a rough average of 195M/WU thus far.
 
POGS has work again! I started getting new WU's yesterday. The WU's are each taking about 4 hours to complete on my i7 machines.
 
Hi theSkyNet POGS volunteers,
The International Astronomical Union is allowing astronomical organisations (like theSkyNet) to submit potential names for 20 exoplanet systems outside our Solar System.

We get to submit a name, and we'd like your help to choose it.

If you get suggesting with names (that follow the IAU rules, see below) on June 10th we'll compile all the suggestions and take a vote for what name to submit to the IAU on June 15th!

Put your suggestions in replies on this thread on the message boards. Don't forget to tell us which system your name suggestion is for, see the list linked below (and if you've got a reason why you think the name is good, then add that in too!)

The Rules

The name must be 16 characters or less in length, preferably one word, pronounceable, non-offensive and not too similar to an existing name of an astronomical object (you can check your suggested name at Sesame Service (http://cds.u-strasbg.fr/cgi-bin/Sesame) or the Minor Planet Centre (http://www.minorplanetcenter.net/db_search)).

The names must not include names of pet animals, names of a purely or principally commercial nature, names of individuals, places or events principally known for political, military or religious activities, names of living individuals or names protected by trademarks or other forms of intellectual property.

More details on the rules are on the IAU website: http://www.nameexoworlds.org/the_process.

So is a list of the 20 exoplanet systems that are up for naming: http://www.nameexoworlds.org/the_exoworlds.

Happy suggesting! I'm looking forward to seeing what you come up with.
____________
Member of theSkyNet team.
Click to expand...

http://pogs.theskynet.org/pogs/forum_thread.php?id=588&postid=4261#4261
 
A GPU version was developed, but when I tested it against the CPU only version the numbers where out by as much as 5%.

Unfortunately we run this project on a shoe string budget and don't have the resources to develop it ourselves.

I have a student starting next year who will work with original developer of MAGPHYS (who now works in Australia) to produce a GPU version.
____________
Regards
Kevin
-----
International Centre for Radio Astronomy Research
Click to expand...

An update on optimised/GPU apps
 
From an email -
theSkyNet SourceFinder 2.0 (beta)
Coming soon to a BOINC manager near you!

Ryan (our intern) and Alex have been working really hard to bring SourceFinder up and running on BOINC. We expect we will be announcing its beta availability in the new year! We have all the data ready to go, and most of the BOINC infrastructure is coded and online, we're just finishing up the final touches with an assimilator (that puts the results you guys will send back to us into a database) and a few other checks and quality controls through the system.

Thank you so much for your patience with this one, Ryan was only with us for one day a week, and has now officially graduated and taken a job in industry so he won't be around much at all anymore! Congratulations from all of us at theSkyNet Ryan, and good luck in your new job.
Click to expand...
 
Yep, I've been trying to catch up on POGs recently before they shutdown which appear to be sometime later this year if I read it correctly.

In FB, we are now at 7th position and quite a few others seem to join the bandwagon recently?. BTW thanks to ChristianVirtual for putting up the stats.

We should keep this in mind as one of our FB strategies. Ooops I shouldn't mention it here. :censored:

upload_2018-3-10_9-37-14.png
 
pututu said:
Yep, I've been trying to catch up on POGs recently before they shutdown which appear to be sometime later this year if I read it correctly.

In FB, we are now at 7th position and quite a few others seem to join the bandwagon recently?. BTW thanks to ChristianVirtual for putting up the stats.

We should keep this in mind as one of our FB strategies. Ooops I shouldn't mention it here. :censored:

View attachment 57271
Click to expand...

Is this still CPU then?
 
CPU only because Kevin (the last remaining original admin) decided if he allowed an optimised client or a GPU client to be finalized that he would not be able to guarantee a supply of work. This is something they struggled with anyways. So, he decided to let people waste their CPU resources in order to "keep them busy" longer. A completely shit thing to do. Quite frankly the only one outside the current intern that really showed respect to the donors IMO was Alex. But he disappeared without a peep a few years back. Pretty much how they all disappeared. It's a shame because they started out pretty awesome.
 
Haven't done anything with Skynet for a while. Can never seem to remember the website that shows all the badges, ranks and so forth. Pretty sure my project rank was #2 last I checked... no way in hell I was going to overtake #1.
 
www.theskynet.org but last I read it was either down or having trouble. Sam was going to update it but with both projects nearing their end, it probably won't see light of day.
 
Gilthanis said:
www.theskynet.org but last I read it was either down or having trouble. Sam was going to update it but with both projects nearing their end, it probably won't see light of day.
Click to expand...

OK, was trying that website, but kept timing out, etc. Good to know I wasn't going crazy! Per the site I can access, looks like I've fallen down to 11th from however high I was. Haven't crunched a WU for them in ages though; so isn't too surprising. (Murphy.cz used to be way behind me, now he's out in front!)
 
fastgeek said:
OK, was trying that website, but kept timing out, etc. Good to know I wasn't going crazy! Per the site I can access, looks like I've fallen down to 11th from however high I was. Haven't crunched a WU for them in ages though; so isn't too surprising. (Murphy.cz used to be way behind me, now he's out in front!)
Click to expand...

Yeah well what are ya gonna do about it, Punk? :p
 
Welllll.... I could possibly do something.... but am more interested in doing Bio / Medical projects than anything else.
 
fastgeek said:
Welllll.... I could possibly do something.... but am more interested in doing Bio / Medical projects than anything else.
Click to expand...

pogs is a common medical condition often brought on after a night of heavy drinking and debauchery. If only I had a dollar for every time someone has said to me 'you look completely pogsed up after last night you prat'.

Take a shot at helping cure this terrible malady fastgeek, you know you want to :D
 
Now I've heard of sozzled; but being pogsed up is a new one on me. :p ;)
 
Their trophies website is up. Wow haven't checked this for a long time and there are plenty of trophies amassed already. Last count I have 58 trophies. I think they overkill on this one.
 
They did go a bit overkill on the trophies, which made it fun (I have 96 of them); but it peters out before too long. Kind of sad as it's such a great trophy whore project... which I honestly confess to being. For instance, I snatched up all the RAC trophies in something like a week. Now I'm 389,176,758 points away from whatever the next trophy is. Yeesh!
 
Yeah.. they made it fun for trophy hunting because there were several different ways to get them. Some were just for crunching on a specific day. However, Alex was the one really pushing that site and making improvements and he left in 2016. Sam was going to do some work on it but they are making the transition soon for potentially shutting both projects down. When this project started, it had a lot of potential. Then it just kinda stagnated about the time Kevin put a damper on making advancements with the application. From there it became boring and less responsive. I personally think they are keeping it going to milk the grant money as long as they can.
 
In less than a week from 7th place to 1st place. Wow. Just wow.
 
You must log in or register to reply here.
Back
Top