n > 9000000

[Keroberts1]

well we've pass N of 9000000 and noone comented? strange we didn't have any conjectures about when this would happen.

[Nuri]

I guess everbody was waiting for 10m. It's not still late anyway.. :p

A more interesting discussion would be (at least that's what I thought), what are the chances that we will find another prime for n<10m. What I'm interested in is the probability of that happening.

[ShoeLace]

i was actually thinking.. that the main queue current goes to 9999967 (ala 10M)

what would be the next upper range to add to teh queues.

would it be another 2m (ie 12M) , 5m (to 15M) or all the way to 20M.

personally i am thinking it will either be 12M or 20M, more likely 12.

i would probably prefer 20M but there are space/datasize considerations with that choice. (some 300000 extra entries )

anthing over 14M would mean moving all the current largest-prime queue which may also be difficult.

[vjs]

Wow 9M already??? I'm actually looking forward to that first 10M test however.

I'm with Nuri two things interest me... will we get another prime before 10M and can we predict what the n-level will be in november, also will the 10M tst come before or after a 300K sieve score?

[vjs]

Just wanted to also point out that garbage error-fix is finally above 5M again.
And secondpass has cleared the 1.9M hurdle.

All at roughly the same time we reached 9M in first pass.

[Joe O]

Originally posted by Nuri
I guess everbody was waiting for 10m. It's not still late anyway.. :p

A more interesting discussion would be (at least that's what I thought), what are the chances that we will find another prime for n<10m. What I'm interested in is the probability of that happening.

Do you mean[list=1]
Find a prime with n < 10M
Find a prime before we hand out a n> 10M test
both of the above
none of the above
[/list=1]

Since I believe there is a good possibility that a passed prime exist around 2.5 M < n < 3M, I can see us handing out tests with n > 10M, before we find that passed prime.
We really need to get behind the second pass effort.

[Nuri]

Exactly.

The probability that we will find another prime for n<10m.

So, this includes any prime with first time tests where n<10m AND any prime with second time tests where n<10m.

Thus, one should not only calculate the chances of finding one (or more) prime for n= 9m-10m range, but also the chances of us having missed a jewel(s) on the first run and it (they?) will pop up for second run for n = 2m-10m range.

The trick is not limited to that. On top of those, one has to also assume and calculate the probability of first time PRP finding a larger prime for some particular k even if we actually missed the first prime for that k for n<10m. Let me rephase that.. Let's say we've missed a prime for k=whatever during an unlucky test at n=8,xxx,xxx. Let's also assume that there is actually a second prime for that k, say at n=17,xxx,xxx. If, for example, first time PRP reaches 17m before second time reaches 8m, then we might not be able to find that n<10m prime. ;)


To be honest, it is beyond my capacity to calculate such a probability.

[vjs]

I have a feeling we will get a two primes very close together one in secondpass one of the denses ones then shortly there after a firstpass prime.

Let me check my crystal ball... :sofa:

secondpass tests in the range of n=2.5M to 3.0M and probably one of the heavier ones such as, 4847 or 55459.

So next prime
k=4847 or 55459
n=2.5M-3.0M

Looking into my crystal ball again .... :sofa:

Second prime ... oh know!!! :trash:

k=67607 or 19249

n<11M but not by much n=10.7M-10.95M

[Keroberts1]

really because my crystal ball was saying 22699 around 34xxxxxx

[Nuri]

MY crystal ball always suggests that the k/n pair I am testing will yield a prime. :D

May be I should change it.. :confused:

[kfx]

Is there anything special about test larger than 9 million, have done 1 test at never got over 2 million cems/sec used to get around 2.5 million. My new test is 8.8 milion and it´s up to 2.4 million ?

[Frodo42]

there is a FFT boundary just around here.

P-1 is switching between 1024K zero padded FFT and 768K depending on K, in my current P-1 range (9.15M-9.18M)