srsieve/sr2sieve enhancements

[KEP]

Originally Posted by rogue

I cannot build a 32-bit Windows exe. I can only build a 64-bit Windows exe.

Thanks for the clarification of the Mersenne factors. Could you then please in plain layman english clarify for me, at what p value it is safe to use 64-bit srsieve? I read something with "LIM..." in the bug file, but it really didn't make it clear to me when I was clear of danger of falls factors and wich factored candidates that should be reinserted into the srsieve.out file...

Take care

KEP

[rogue]

Originally Posted by KEP

Thanks for the clarification of the Mersenne factors. Could you then please in plain layman english clarify for me, at what p value it is safe to use 64-bit srsieve? I read something with "LIM..." in the bug file, but it really didn't make it clear to me when I was clear of danger of falls factors and wich factored candidates that should be reinserted into the srsieve.out file...

That was only for older versions of srsieve. My changes started with a version of srsieve that did not have that bug.

[cheesehead]

Originally Posted by KEP

I'm asking because currently mersenne@home is appearing to do things the wrong way in their mersenne-prime searching. (At best they are only tripple cheking work that has already been done, but most likely they are waisting several thousand CPU years on doing work that will lead to no new discoveries and eliminate nothing new from the quoue of work). Before I continue to debate with the Polish guy at mersenne@home,

Kenneth,

I suggest backing off for a while on debating with the Polish guy (Bobrecki, right?) at Mersenne@home.

As it is right now, Mersenne@home isn't damaging anything. Leave their duplicative effort alone for now, because it isn't hurting anything.

If we send too much criticism to Bobrecki without convincing him to cooperate, he could cause us lots of trouble by deciding to leapfrog over the ranges GIMPS is currently testing and unleash his BOINC minions onto higher ranges. So, don't provoke him!

I propose that we prepare a document carefully explaining the advantages of cooperation and the disadvantages of competition, then politely introduce it to Bobrecki. Let's brainstorm ways in which a BOINC-based project could cooperate with GIMPS. I'll start a thread for this in Lounge.

Okay?

[KEP]

Originally Posted by cheesehead

Kenneth,

I suggest backing off for a while on debating with the Polish guy (Bobrecki, right?) at Mersenne@home.

As it is right now, Mersenne@home isn't damaging anything. Leave their duplicative effort alone for now, because it isn't hurting anything.

If we send too much criticism to Bobrecki without convincing him to cooperate, he could cause us lots of trouble by deciding to leapfrog over the ranges GIMPS is currently testing and unleash his BOINC minions onto higher ranges. So, don't provoke him!

I propose that we prepare a document carefully explaining the advantages of cooperation and the disadvantages of competition, then politely introduce it to Bobrecki. Let's brainstorm ways in which a BOINC-based project could cooperate with GIMPS. I'll start a thread for this in Lounge.

Okay?

I'm fine with that. I just wanted to stop an apparent waist of resources, mostly because I didn't read carefull enough what he told me about the factoring, but as Rogue explained and I understood when reading more carefully what Bobrecki wrote to me, I have no intentions on debating him. I like your idea more, because weather it be BOINC or elsewhere, a waist of resources is always a waist and should be adverted (if possible). However I support your idea, because I am not able to explain to Bobrecki in well enough english, what I find troubeling about Mersenne@home and the way it apparently does trial division that has already been carried out by GIMPs.

Good luck on your effort.

Kenneth!

[cheesehead]

Originally Posted by cheesehead

I'll start a thread for this in Lounge.

It's "Brainstorming ways to convince Mersenne@home to cooperate, not compete" here: http://www.mersenneforum.org/showthread.php?t=16850

Originally Posted by KEP

I'm fine with that. I just wanted to stop an apparent waist of resources, mostly because I didn't read carefull enough what he told me about the factoring, but as Rogue explained and I understood when reading more carefully what Bobrecki wrote to me, I have no intentions on debating him.

Also, let's keep in mind that a waste of resources on the small present scale of Mersenne@home isn't really much of a tragedy.

After all, the participants there are learning some things, and I'm personally interested in seeing the ways that their BOINC-based project is organized differently from ours -- we may be able to get good ideas from their independently-developed approach.

However I support your idea, because I am not able to explain to Bobrecki in well enough english,

How about good enough Polish? (If not you, someone you know?) :-)

what I find troubeling about Mersenne@home and the way it apparently does trial division that has already been carried out by GIMPs.

Again, let's not be concerned with their current harmless sub-optimization, and keep in mind that there are other outcomes (i.e., learning) there than only the bare numbers.

Good luck on your effort.

Come over to the Lounge thread and contribute your ideas! :-)

[KEP]

Originally Posted by cheesehead

How about good enough Polish? (If not you, someone you know?) :-)

Well I'm Danish, so I think I would have more luck in english then in writing something and have google transkate it to Polish :)

I saw your thread. I'm not really conviced that they want to compete with GIMPs and according to their news, it appears that some users at GIMPs has had bad hardware or software wich has missed at least 2 factors, so afterall a tripple and quadrupple check might afterall be just the right thing to do, in order to ensure that the data is actually correct. Heck, even I've decided to support them for a while, now where I understood that the trial factoring went to:

2kp+1 (k<=1G, k<=10G and k<=~73G)

and not to:

p=1G (Phase 1)
p=10G (Phase 2) and
p=73G (Phase 3)

but I'm still looking forward to the time n>1G, since it appears that the new discoveries will come more frequently at that depth (even for Phase 1).

Take care

Kenneth

[MyDogBuster]

sr1sieve question.
sr1sieve writes to the output file every hour.
Does it then read that file as input for the next hour's sieve?
That would speed things up a tad with less terms to test (assuming it found any the last hour).

[rogue]

Originally Posted by MyDogBuster

sr1sieve question.
sr1sieve writes to the output file every hour.
Does it then read that file as input for the next hour's sieve?
That would speed things up a tad with less terms to test (assuming it found any the last hour).

No. It only writes as a checkpoint.

[pepi37]

Can this appp be used in Boinc sieve project on CPU?

[rogue]

Originally Posted by pepi37

Can this appp be used in Boinc sieve project on CPU?

For what project? If someone wants to try, they are welcome to do so.

[pepi37]

Any of this three?
Cullen/Woodall Prime Search (Sieve)
The Riesel Problem (Sieve)
Proth Prime Search (Sieve)