unusually low P-1 bounds

[ixfd64]

Several of my computers are currently doing P-1 factoring in the 53.8M range. I usually allocate 384 MB of memory per core, and the bounds I get are usually B1=630000, B2=16222500 (4.2385 GHz-days).

However, I recently requested a few P-1 assignments in the 50.1M range manually, and the bounds for those are B1=280000, B2=5460000 (only 1.4934 GHz-days).

Is this normal? For the record, the bounds I get for a 49M exponent are B1=570000, B2=14107500 (3.2916 GHz-days).

[cheesehead]

Originally Posted by ixfd64

Several of my computers are currently doing P-1 factoring in the 53.8M range. I usually allocate 384 MB of memory per core, and the bounds I get are usually B1=630000, B2=16222500 (4.2385 GHz-days).

Presumably, those exponents had not yet been LLed, so the last parameter on their worktodo Pfactor lines was 2 (finding a factor saves 2 LLs) -- right?

However, I recently requested a few P-1 assignments in the 50.1M range manually, and the bounds for those are B1=280000, B2=5460000 (only 1.4934 GHz-days).

Had those exponents already been first-time LLed? What's the last parameter on their worktodo lines?

[ixfd64]

The worktodo.txt format is:

Pfactor=<assignment key>,1,2,<exponent>,-1,68,2

[markr]

Originally Posted by cheesehead

Had those exponents already been first-time LLed? What's the last parameter on their worktodo lines?

These ones had a prior LL test - there may have been others with slightly different bounds than ixfd64's example.

Exponent	Status	Data

50175679	No factors below	2^69
	P-1	B1=280000, B2=5460000
	Unverified LL	9CA5E2E3BBEB38__ by "mister_phil" on 2010-12-25
	History	9CA5E2E3BBEB38__ by "mister_phil" on 2010-12-25
	History	B1=280000, B2=5460000 by "ixfd64" on 2011-03-31

50185963	No factors below	2^69
	P-1	B1=280000, B2=5460000
	Unverified LL	AC782724D87460__ by "ANONYMOUS" on 2010-08-31
	History	AC782724D87460__ by "ANONYMOUS" on 2010-08-31
	History	B1=280000, B2=5460000 by "ixfd64" on 2011-03-30

50175679	No factors below	2^69
	P-1	B1=280000, B2=5460000
	Unverified LL	9CA5E2E3BBEB38__ by "mister_phil" on 2010-12-25
	History	9CA5E2E3BBEB38__ by "mister_phil" on 2010-12-25
	History	B1=280000, B2=5460000 by "ixfd64" on 2011-03-31

50185963	No factors below	2^69
	P-1	B1=280000, B2=5460000
	Unverified LL	AC782724D87460__ by "ANONYMOUS" on 2010-08-31
	History	AC782724D87460__ by "ANONYMOUS" on 2010-08-31
	History	B1=280000, B2=5460000 by "ixfd64" on 2011-03-30

[cheesehead]

Originally Posted by ixfd64

The worktodo.txt format is:

Pfactor=<assignment key>,1,2,<exponent>,-1,68,2

for the 50.1M exponents?

[ixfd64]

Originally Posted by cheesehead

for the 50.1M exponents?

Yes.

[cheesehead]

Do you happen to have a saved copy of the Pfactor assignment lines you received for the 50.1M exponents that you can post?

[ixfd64]

Here's two of them:

Pfactor=A5E36281D18ADB9E5C6C5FAD90B9DE87,1,2,50171491,-1,69,1
Pfactor=38FD532B5233142BCF520A0FA832B338,1,2,50175119,-1,69,1

[KingKurly]

Originally Posted by ixfd64

Here's two of them:

It's the ,1 at the end that causes the difference. That is the number of LL tests that will be saved by running the P-1 test. On an exponent that has never been LL tested, that number would normally be 2 (the LL and the LL-D) instead of 1. Prime95/mprime set their bounds based on how many LL tests are assumed to be saved in the event of success. You can pick numbers between 0 and 10, though I guess picking 0 wouldn't make much sense, would it...?

[cheesehead]

Originally Posted by ixfd64

Here's two of them:

So, the answers to some of my previous questions should (:-) have been:

Originally Posted by cheesehead

Had those exponents already been first-time LLed? What's the last parameter on their worktodo lines?

Answers: yes, 1 :-)

Originally Posted by cheesehead

for the 50.1M exponents?

Answer: No, for those exponents the last parameter was 1 rather than 2. :-)

... which would have more quickly led to the reason for their P-1 limits being lower than for your 53.8M exponents.