Discussion Thread for the Continuous GW Search known as O2MD1 (now O2MDF - GPUs only)
Language
Copyright © 2024 Einstein@Home. All rights reserved.
Copyright © 2024 Einstein@Home. All rights reserved.
... this "behaviour" has been
)
... this "behaviour" has been thoroughly explained ...
San-Fernando-Valley
)
Where can I find the explanation for the three minute computation that used to take 50 seconds.
".... Is it necessary at
)
".... Is it necessary at all?"
Intellectual rigour I'd expect. :-)
O3 is going to be our best chance yet of finding a signal buried in all the other stuff - more 'celestial noise' than instrumental I hope. We hope that persistence of a signal will rise into the statistics produced from data analysis. This is an ambitious enterprise so a bit of extra time to be sure is just fine. For that matter I am pleasantly surprised that this project is still strong, productive and generating interest ( such as yours ) after 16 years. But this is the true subtlety of the task ... listening for whispers within whispers ....
Cheers, Mike.
( edit ) There are two sins to be escaped from in our work here. One is missing a signal of cosmic origin, the other is claiming one that ain't so.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
petri33
)
I don't think there has been an explanation about the difference in time between old and new. only the reason for this step. I don't know specifically, but it might be as you implied, making sure the GPU did it right by calculating it on the CPU.
I agree that the mutex you implemented in the SETI app was super nice. maybe you can find a way to convey the specifics of this code and behavior to the E@h devs so they could consider making that change. Do you think you could implement that in a custom application yourself like you did on SETI?
_________________________________________________________________________
Mike Hewson wrote:or that
)
O3 is getting to be make or break time for me. It looks to be much better than before, but if nothing is found, I would start to conclude that the model for the neutron stars themselves is wrong. They may emit in bursts as PAVEL_SUSLO suggests in the Science section.
Maybe the FGRPB1G app could be modified to look for burps?
Ian&Steve C. wrote: petri33
)
I can not take the credit for the implementation of the Seti-mutex code. It was, If I recall right, a Danish Setizen who did that. Major kudos ('kiitos/kunnioitus' in finnish) to hem. (him/her=hem, you never know these days and you have to be as sensitive as the receivers near The absolute zero. My guess is that the nickname was 'danishguy')
I have the code and I can ask if I can give it for this project too - should some instance be willing to implement it with options in project settings page to use or not when running two or more at a time.
The custom app(s) built by me are like a nice sunrise or sunset. Too far away at this time. No ref results, no testing environment, no nothing that the (great unofficial project) Lunatics implemented those Seti days to support individual piece by piece testing and optimization of the code.
The official public source code is the key for public interest: Some do the science, some do the initial write-up, some have tested an continue to do so, some write bug reports, some do scientific reports or something else. ... And then some (a lot of people) make the project more productive and I N T E R E S T I N G by contributing their knowledge in: optimization, public questions and answers, peer development, support, beta testing, ...
"a new rigid is flexible"
-
petri33
Indeed it was Oddbjornik’s
)
Indeed it was Oddbjornik’s code/idea. But I’m sure it required a bit of modification to be incorporated into your app. I don’t mean to take away that credit, just referencing that you put it in your application, and that the implementation/idea was very nice.
_________________________________________________________________________
Ian&Steve C. wrote: Indeed
)
Oddbjornik it was! Yes!
B.t.w Have you tried running O3SA on your RTX2080(vanilla or Ti) or on your other similar to mine to take a look of run times for reference? "One task only, just one. -- Sir? -- One ping only, just one. It is time."
Thanks Ian&Steve C.
Sure thing petri. I’ll see
)
Sure thing petri. I’ll see about running through a few tonight.
_________________________________________________________________________
Quote: "One task only, just
)
Ha ha hah. Big thanks, Petri for the meme.