What's next

[engracio]

Since we got another prime in the bag, what is next in the project. Projections, predictions, etc.?

When can we expect the next prime based on the cpu power? Will people come back crunching since they left after the 9th prime?

e

[umccullough]

New/updated client software would be nice... with some easier configuration options - easier to setup for multiple cores/processors, easier to configure remotely, etc.

And I still want to get a newer BeOS/Haiku port :/

[Zoink]

First, great congrats to Angotonfy. seems like a worthy cruncher to get the credit for this one, he's done a lot of work for the project. Can't help to envy him a bit though.

Some predictions for the future would be really neat however. I guess this means efforts will go up about 15% for the remaining Primes now that we've left another one behind us.

When this is done, will there be free beer for everyone?

[Frodo42]

Nice to see another prime, congrats to everyone.

I am afraid I am one of those that will not be back anytime soon crunching, but I hope that others will. I didn't stop crunching due to the lack of primes, but due to my power bill and no longer having electrical heating.
This project is still dear to me, I might be back sometime, but new primes is not what is going to bring me back.

[vjs]

Since we got another prime in the bag, what is next in the project. Projections, predictions, etc.?

When can we expect the next prime based on the cpu power? Will people come back crunching since they left after the 9th prime?

e

First off I was hoping that one of these small primes would go next either, 19249, 22699, or 67607. Primarily b/c of the number of tests per 1M n-range.

These tests represent 6.2%, 6.0%, and 5.2% of the tests respectively.

One might think that eliminating the heaviest k 10223, which accounts for 17.2% of the tests would be a better k to eliminate. However each k/n pair has a decreasing probablity of being prime as n increases. So doing more tests at lower n is benifital (hope this makes sence). 19249 was one of those k's that could have been that last to fall.

Regardless what will we see???... Humm, we will see a slight increase in the speed at which n increases but not much (6.2%). It probably won't be a noticeable difference. Hopefully we can get a little influx of new people for at least a short time.

.......................

If there were anything I would suggest doing currently, it would be taking one of the remaining small k's either, 22699 or 67607 up to about the 9M level. Just to see what happens. I was sort of hopping that we could switch the next.txt file back to the firstpass que so that mikes pages work again. Also I am very impressed by Louies and Angotonfy ability to hide what was going on.

I was sort of thinking next time we had a prime (this time) we would go to secondpass tests for a few weeks... get any new people started with some short tests to see if they like it, etc.

Any other suggestions and comments.

On the sieve and P-1 aspect. Engracio basically did all of the tests for n upto 14M so we are good on that front for a while. Talking about sieve alot of the heavy hitters are now doing the combined dat, as well as myself.

From the sieve aspect I don't think there will be too much of a change...

[hhh]

Talking about sieve alot of the heavy hitters are now doing the combined dat, as well as myself.

From the sieve aspect I don't think there will be too much of a change...

First of a ll let me say that I am very happy with this new prime; it was about time. Note that 19249 is prime, thus helping to solve the Prime Sierpinski problem as well.

Now, all the buzzwords are there. Sieve, PSP, combined dat, heavy accomplishment, change.

About two years ago, the sieving effort of SoB has started on working for candidates with n>20M, in order to prepare for the time when firstpass reaches this n in a few years. Soon afterwards, a discussion started about joining the sieve efforts of SoB and PSP, given the synergetic effects. The joint sieving was eventually started and has currently sieved up to almost 600T.

The idea was to get many more factors, with little more work, and to increase interest by an increased efficiency. Both worked very fine, IMO.
The sievers whom's majority came from the PSP side removed many many candidates beyond 20M for SoB and found even factors that slipped through the first sieving effort.

Though PSP is searching 11 k's currently, against 7 at SoB, the processing of a range with the PSP .dat-file takes only about 30% more time than with the SoB-one, while more than twice as much factors are found.

That's why I think it would be a good opportunity to switch to the combined dat for the main sieve effort of SoB as well. Organisational effort would be little to none, the combined .dat being already available, and as an estimation, interest in sieving could even rise by the new possibility of working for two projects while searching "hot" firstpass factors for SoB.

That's my opinion. Perhaps I'm not alone?

Yours H.

[Death]

*singin* You are not alone.....


=))))