Two questions from a newbie

1)
When configuring mprime, what's the difference between the two questions?

a) Request 10,000,000 digit numbers to test (N):
b) Request Mersenne numbers to run primality tests (N):

If I understand right a) is the same of b), but only for numbers > 10,000,000 digits? But now, since all number with < 10,000,000 digits seems to be already checked, a) and b) are the same?

2)
I subscribed some of my machine to GIMPS. However, having troble during initial configuration, I had to reinstall mprime some times.
Now I see in my account web page more machines that I have subsribed, some being the first attempts in configuring (so the page show 3 Pentium 4, with different ComputerID, when I really have only one).
How can I remove these old attempts? Is it done automatically after a some time?

Thanks

[rx7350]

As to your question #1, yes, you are correct in your interpretation about 'a' and 'b'. There are however, still many exponents below 10M that are still available for LL testing. Groups of these are released to the PRIMENET server for assignment fairly frequently, weekly at a minimum.

BTW, the leading edge of LL testing just crossed over the threshold between 2048K FFTs and 2560K FFTs, which will result in an increase in iteration times on systems currently in use, but will also increase the amount of credit when an exponent is completed.

[petrw1]

Originally Posted by rx7350

BTW, the leading edge of LL testing just crossed over the threshold between 2048K FFTs and 2560K FFTs, which will result in an increase in iteration times on systems currently in use, but will also increase the amount of credit when an exponent is completed.

Is not the threshold at 40,250,000 ... whereas the current LL line according to Primenet is 39,5xx,xxx???

If I understand well:
a) Request 10,000,000 digit numbers to test (N):
b) Request Mersenne numbers to run primality tests (N):
c) Request Mersenne numbers to double-check (N):
d) Request Mersenne numbers to factor (Y):

a) -> LL test for numbers > 10,000,000 digits
b) -> LL test for any number
c) -> LL test for "composite number verified only one time"
d) -> brute force factoring with 2^6x

Is this right?

[TimSorbet]

Originally Posted by petrw1

Is not the threshold at 40,250,000 ... whereas the current LL line according to Primenet is 39,5xx,xxx???

SSE2 2048-2560 threshold is 39,500,000 exactly. SSE 2048-2560 threshold is 40,300,000 exactly. (According to http://www.mersenne.org/bench.htm) So, you're both right, just for different CPUs.

[petrw1]

Originally Posted by Mini-Geek

SSE2 2048-2560 threshold is 39,500,000 exactly. SSE 2048-2560 threshold is 40,300,000 exactly. (According to http://www.mersenne.org/bench.htm) So, you're both right, just for different CPUs.

Yet, http://www.mersenne.org/status.htm suggests the 2560 cutoff at 40,250,000 ... which is neither 39,500,000 nor 40,300,000??

And so does the TPR P90 years calculator: http://www.teamprimerib.com/rr1/bin/calc_p90.php

[TimSorbet]

Originally Posted by petrw1

Yet, http://www.mersenne.org/status.htm suggests the 2560 cutoff at 40,250,000 ... which is neither 39,500,000 nor 40,300,000??

And so does the TPR P90 years calculator: http://www.teamprimerib.com/rr1/bin/calc_p90.php

Perhaps the benchmark page rounded it to 40.3M, when it's really 40,250,000, but that wasn't my point. My point was that SSE and SSE2 had different cutoffs making both of you correct and incorrect.

[rx7350]

Works for me. I stand corrected.

[petrw1]

I'm still confused

I have a few computers running Prime95 that are SSE2 type machines (mostly P4) according to the Benchmark page and according to the "Options ... CPU ..." screen and all these SSE2 machines are running LL/DC tests.

I have been running V24 since it became publicly available about 2 years ago.

I created an EXCEL spreadsheet using the ranges from the http://www.mersenne.org/status.htm page and applied it to the 74 LL/DC tests I have done ranging from 11M to 36M and my calculated points match exactly to what I have been awarded.

But now I realize these are NOT the correct ranges for SSE2 machines so my points should be different if I used these SSE2 ranges ... should they not?

For example according to the benchmark chart the change to 2048K FFT starts at 34.56M whereas the status.htm chart does not list the change until 35.10M. Therefore exponents that I tested between 34.56M and 35.10M should be awarded different points depending on the chart used. I've done 11 in this range (all with V24) and got the points awarded based on the 1792K FFT size.

Do I make sense? Can anyone help me make sense of this?

[markr]

Originally Posted by petrw1

But now I realize these are NOT the correct ranges for SSE2 machines so my points should be different if I used these SSE2 ranges ... should they not?

No - points are calculated using the ranges on the status page irrespective of the machine that returns the result. I suspect those ranges were for an older version of prime95, kept for simplicity as well as being close to both SSE2 & non-SSE2 ranges for a later version. Primenet can't know for sure if SSE2 was used or not.

It does make some minor anomalies. For some computers, a few assignments yield a little bit less* points than they "ought to". An even smaller number of assignments carry a small bonus for other computers.

(* fewer? less sounds wrong but I think it's right...)