New sieve program....

[em99010pepe]

Please go to Free-DC forum and check this thread:

http://free-dc.org/forum/showthread....threadid=10069

Cheers,

Carlos

[Citrix]

I see only a 5% increase in new sieve.

[em99010pepe]

Originally Posted by Citrix

I see only a 5% increase in new sieve.

It's not the final version!

Carlos

[Joe O]

Originally Posted by Citrix

I see only a 5% increase in new sieve.

Silly me, I thought any increase would be welcome. I guess I will have to withdraw the program until there is an acceptable increase. Would 10% do?

[Citrix]

Joe,

I like 5% also. I would even prefer a 0.000...1% increase. But since your program is currently not bug proof. I will stick to proth_sieve, unless you make your program so fast that missing a few factors doesn't matter anymore.

Waiting for such a fast program.

Citrix

[Joe O]

Originally Posted by Citrix

Joe,

I like 5% also. I would even prefer a 0.000...1% increase. But since your program is currently not bug proof. I will stick to proth_sieve, unless you make your program so fast that missing a few factors doesn't matter anymore.

Waiting for such a fast program.

Citrix

Have you seen 0.30C miss any factors? If so please give me the particulars.

[Citrix]

Nothing so far to report, I will let you know if I see any. I know proth_sieve misses some.

[Joe O]

Originally Posted by Citrix

Nothing so far to report, I will let you know if I see any. I know proth_sieve misses some.

I'm also interested in specifics where proth_sieve misses a factor.

[Citrix]

Email ltd (Lars) , he can give you a list of factors from the DB.

Citrix

[Citrix]

I have a suggestion to make.

consider the Pohlig-Hellman algorithm that proth_sieve uses. Now when the probability of finding a factor is less than 0.001% for a prime then we skip that p.That way we can sieve faster and deeper and find more factors even though we will miss a few.

I think if this will be implemented, we can find more factors faster. What do you think?
You can give it a try and inform us of the results.

Note:- 0.001% can be any arbitary number, where it is most optimal and can be deepdent on p, where p is the prime being tested.

Thanks,
Citrix

[Citrix]

2090037773 | 1*2^1896586+1
2090076869 | 1*2^2567662+1
2090099741 | 1*2^2154742+1
2090190901 | 1*2^1990658+1
2090281541 | 1*2^2645926+1
2090494501 | 1*2^2787326+1
2090593409 | 1*2^1719238+1
2090699453 | 1*2^31342+1
2090903141 | 1*2^2944934+1
2091284873 | 1*2^1039406+1
2091594709 | 1*2^873682+1
2091643537 | 1*2^2125654+1
2091853069 | 1*2^2745214+1
2092308833 | 1*2^1469318+1
2092944397 | 1*2^3260038+1
2093003257 | 1*2^499762+1


Missed by proth_sieve for k=1.