I thought I'd jump into the fray.

135 is tested to 1e15. Nothing.

I'm doing the same for 165-168. Nothing yet.

[QUOTE=rogue]I thought I'd jump into the fray.

135 is tested to 1e15. Nothing.

I'm doing the same for 165-168. Nothing yet.[/QUOTE]

I posted earlier:
135 60000000000000 52324121395455

I think it's time for a website

My domain hosting gives me huge storage (20GB), unmetered bandwidth, mysql and cgi-bin stuff.

I could probably have a progress and reservation system done just after the weekend.

I could even extend the client to automatically grab a work unit (say 30 minutes of work on a specific n and kmin,kmax range) from the website and upload the results/progress.

Of course, anyone else is welcome to create a website themselves, we can then vote as to whose is best and keep using that.

Placeholder: [url]http://octoproth.greenbank.org/[/url]

[QUOTE=smh]I posted earlier:
135 60000000000000 52324121395455

I think it's time for a website[/QUOTE]

Try [URL="http://frase.250free.com/Octoproth/octoproth.htm]this[/URL].

Edit: Sorry, haven't noticed Greenbank's reply...

No worries fetofs.

I want to make it clear that I'm not trying to take over this project. I'm happy to knock up a website for tracking that will be based on a network based client (although there will be a way to participate if you don't have a permanent network connection!).

If someone comes up with a better system/website/whatever then I'm happy to move over.

I've already made some good progress on a network client. It is already connecting to a server and obtaining work units as we speak. This is mainly for testing. The server is on the internal network at my work so there's no point in me sharing the code/client yet as you wouldn't be able to connect to it. I hope to get it all hosted externally on my own domain this weekend.

The main features will be:-
o Registration to the site (creating a username) so that your contributions can be tracked
o Progress/updates/statistics/etc
o Three work modes:-
o 1) A work unit from one of the next 5 n's for which an octoproth is not known. (default)
o 2) A work unit from the lowest uncompleted (i.e. search space not exhausted) n.
o 3) A work unit from a local file (so that users can work offline and on whatever ranges/n-values they want)
o Functions to manage your assigned work units (abandon them, etc).
o Separate submission for non-networked clients

Any more comments/suggestions I'd be happy to discuss...

I also want to make it clear, that I only want to link to the project on my web site. So please someone create a place for this.

The .NET version of octo.exe will not have network features.
It's around 70% complete, but may take awhile to iron things out.
P05.exe is my main focus, now that RMA.NET is ready.

Shane F.

I'd like to suggest to R.Gerbicz a few speed-ups in the PRP checking part.

1) Compute and PRP check 8 numbers one by one (not in groups of 4 like now), break on first composite.

2) Begin the above starting with 2^n-k (the smallest one) followed by the others in increasing order (now, k*2^n+-1 are checked first).

3) Compute 2^n and 2^(n+1) outside of the main loop as they don't change.

Blimey O'Reilly
 

I go away for a few days, and then check in again, and suddenly this has taken off again. It is strange how projects find new energy, this one started with no interest at all until I reposted it, and then I thought we had all given up on it.

More power to your collective elbows and happy hunting. When there is a true executable in Windows I may join the hunt again, as, at the moment, my new projects are looking a bit dull.

Regards

Robert Smith

new octo_2.0
 

I've done a new version of octo program. Possibly this is faster than previous version octo_new.

Now there is only one modular multiplication instead of eight per sieving prime in the main sieve. I've never thought that this is possible, it was a small trick. Now magic_constant is really large, I've decreased this value from 2*10^6 to 4*10^5, this means that you get an extra speed up if you reach another breakpoint in this case. I've also rearranged some tables to use less memory, now the program is using about only 550 KB, the previous version used about 810 KB. This is much better for L2 cache. I've rewritten almost the whole main sieve routine, probably this is also a speed up. Now the program will print out the version, the current version is 2.0 ( consider octo_new's version as 1.0 ).

Can you test my new code for different range values?
See my attachment for the program octo_2.0

[QUOTE=rogue]I thought I'd jump into the fray.

135 is tested to 1e15. Nothing.

I'm doing the same for 165-168. Nothing yet.[/QUOTE]

165-168 done to 1e15. Nothing.

Taking 160, 161, 169, 170 to 1e15

I'm certainly all for creating a webside for this project Greenbank, I don't think anyone will say otherwise. It was reasonably handled by the simple text file when there were about three people trying to find octos, now things have changed.