BOINC effort for CRUS

[Ken_g6]

Hey, guys,

It looks like an invitation code is required to join the BOINC project. How would I go about getting one of those?

Thanks!

[KEP]

Originally Posted by Ken_g6

Hey, guys,

It looks like an invitation code is required to join the BOINC project. How would I go about getting one of those?

Thanks!

Just sent to you in a PM, as it is found on the frontpage of the BOINC site

Take care

KEP

[Lennart]

Originally Posted by Ken_g6

Hey, guys,

It looks like an invitation code is required to join the BOINC project. How would I go about getting one of those?

Thanks!

Look in the main page In the first section About SRbase


Lennart

[Neo]

Yeah, the instructions are not clear ...
You have to goto the website first, attempt to login, and then it will let you create an account.
That's when you use the invite code...
After that's done, then you follow the directions to add the project to your BOINC manager.

Neo

[vaughan]

Originally Posted by gd_barnes

Yes, Curtis's response is the point. It took me a little while to add your website to the exclusion (white) list on my anti-virus so that it would not block it. I barely know what I am doing and it took 2 attempts and about 15 minutes of my time. For people who don't know how anti-virus's work, you will likely lose them as potential searchers because they are unable to reach your site.

Ah that's good for us "early adopters" as it allows us to race ahead in the BOINC stats :)

Only problem I've had with this new BOINC project is that they run out of work too often. Primegrid never runs dry.

With respect to Sieving, PG also allows us to sieve using our NVidia GPUs. Dunno why Rebirther hasn't set that up for SRBase yet.

[KEP]

Originally Posted by vaughan

With respect to Sieving, PG also allows us to sieve using our NVidia GPUs. Dunno why Rebirther hasn't set that up for SRBase yet.

Well GPU is propably never gonna happen at SRBase, since the work that CRUS does is using either srsieve, sr1sieve or sr2sieve for sieving. None of these 3 programs has been ported to any GPU system. Last I heard of it, the problem is not with the applications themself, but with the required memory getting bigger than the available memory in the GPU.

KEP

[vaughan]

OK thanks for the info KEP.
It appears that SRBase is attracting some crunchers who wouldn't normally be seen on any of the non BOINC math projects. I think its good to see some activity.

[KEP]

Originally Posted by vaughan

OK thanks for the info KEP.
It appears that SRBase is attracting some crunchers who wouldn't normally be seen on any of the non BOINC math projects. I think its good to see some activity.

It is very good, because even though we are a few very dedicated crunchers at the mersenneforum, then weather we like it or not, our resources is getting stretched more and more thin, so it was about time to go BOINC if we would like to see all the started base move towards significant levels, within reasonable time. Eventually not even BOINC will be able to run out of work, because the n-range and the amount of remaining k's in a high-n level is going to be many enough to keep a sustainable flow of work at BOINC, weather or not sieving is implemented at srbase.

[vaughan]

KEP do you mean like 17 or Bust and its incredibly long running tasks?

Personally, I cannot stand such long runners and I'll switch to another project. For example, in November and December I was running a project called BOINC@Fiit and I had some tasks that ran for over 1000 hours. One completed OK but I lost the other two I was running on a different PC as I inadvertently re-booted it. The stupid project doesn't have checkpointing enabled so I lost all that CPU time (and electricity). Consequently, I'll never run that project again.

[KEP]

Originally Posted by vaughan

KEP do you mean like 17 or Bust and its incredibly long running tasks?

Personally, I cannot stand such long runners and I'll switch to another project. For example, in November and December I was running a project called BOINC@Fiit and I had some tasks that ran for over 1000 hours. One completed OK but I lost the other two I was running on a different PC as I inadvertently re-booted it. The stupid project doesn't have checkpointing enabled so I lost all that CPU time (and electricity). Consequently, I'll never run that project again.

Well eventually yes, but because CRUS is spread over 1031 bases on the Riesel and 1032 bases on the Sierpinski side (2063 bases in total), then there will for the next hundreds of years, be the possibility of offering work, where the test completes in just a few minutes.

For the bases with a low conjecture (conjectured k<=10K), it is most likely, that computation time required per test, will eventually reach the times that you currently see on SeventeenOrBust since the increase in computation speed will not be able to keep up compensating for the increase in the size of the numbers tested.

So as you can see, Rebirther has the possibility of offering plenty of small workunits, it may very well be so for the next hundreds of IRL years of using BOINC, before all 2063 bases reaches an n level where testing time will frequently be above 1 hour.

Another fact is that it may very well be several decades before all bases<=100 is tested to n=25K.

Personally I hope that Rebirther will eventually offer very long tests, since that will attract some entirely new kind of users. But who knows, what the future will hold for us

[Puzzle-Peter]

Originally Posted by KEP

Personally I hope that Rebirther will eventually offer very long tests, since that will attract some entirely new kind of users. But who knows, what the future will hold for us

I think it would be best to offer two or three subprojects so users can decide which runtime they prefer.