100 sievers?

[Moo_the_cow]

On this page:

http://www.aooq73.dsl.pipex.com/scores.htm

we can see that there are currently 89 sievers. When do you think we'll get to a hundred?

My guess - April 8

[Death]

March 8

I'll do some PR. =))

and it can be tomorrow
11 more registering at site and 11 more new factors =))

[Death]

90 CrackDaddy 58.51 ( 0.00) 22 ( 0.01) 1 3 19 0 0 0 0/750
91 ~timmay256 18.00 ( 0.00) 1 ( 0.00) 1 0 1 0 22 0 0
92 ~Annihilus 12.70 ( 0.00) 1 ( 0.00) 0 0 1 0 0 0 0
93 [unknown] 10.92 ( 0.00) 4 ( 0.00) 0 0 4 0 0 0/9050 0
94 ~Th0mmy 2.25 ( 0.00) 4 ( 0.00) 0 0 4 0 0 0 0
95 ~Larabil

[Keroberts1]

i believe unknown is actually some of alexR's factors from when he wasn't logged in so 94.

[Paperboy]

I guess me stopping once I finish my current range will hurt the push towards 100.

[Keroberts1]

well i don't think iunderstand why yo uwould stop but its only a count of total users to ever submit work s oit wwont change anything. If you don't mind me asking what are your reasons for stopping sieving? If its for scoring reasons then you should note that although in the short range factoring produces a higher score sieving produces a vast database of factors that will score i nthe future and i nthe long run give sieving a much higher score rating. As can be seen by the top of the scoreing charts where sievers dominate.

[Nuri]

Originally posted by Paperboy
I guess me stopping once I finish my current range .......

Paperboy, thanks for your contribution so far. Hope to see you around.

[Paperboy]

It has more to do with arstechnica about to take the #1 team spot away from Team Anandtech. I'll need to devote my resources back to prp. Although it is nice how there is only a 25% hit on sieving and a 50% hit on factoring when running them together on my p4 HT compared to running either one separately. (395 kp/s vs 300 kp/s and 40 min vs 60 min for factoring)

[Paperboy]

I guess I have to take my last statement back. Reserved a 500 block for my p3 1 ghz.

[Death]

95 ~timmay256
96 ~Annihilus
97 ~Th0mmy
98 ~Larabil

[Keroberts1]

bump

[ShoeLace]

well ive started some P-1 factoring on my P4.. does that mean i'll appear in this list at some point? :confused:

:elephant: :elephant: :elephant: :elephant: :elephant: :elephant: :elephant:

EDIT: yes i am so there are now 100 on the list.. woot!!

vvvv me vvvv

48* shoelace 90070.13+( 0.08) 1+( 0.00) 0 0 1+ ....


wasnt a bad score either at 90K

[Death]

and good start =)) you are on upper half

[Joe O]

100 ~Larabil

Our most recent new siever was #100

Possibly Shoelace or Silverfish.

Or it could have been AlanSimpson or levitate.

In any case, welcome aboard.




What is our next goal?

[vjs]

Joe...

I'm going to set a lofty lofy goal, I posted on my teams main site the following




We have seen some increased effort in the sieve department which I'm glad to see.

A couple of us have been trying to get this done for some time, and I only wish I had personally started sieving earlier.

Had we all started earlier, ie while around 3m, more of the founds factors would have eliminated more tests. Thus far with the sieve effort is almost at 500T (100% done) and we have eliminated 31% of the factors between 1m and 20m . Even if we find 4 more primes before 20m the majority of these factors were/are useful, and the more factors we find the better.

In addition we will more than likely double check more than half of the k's below 20m. Point being,... even if you find a k/n factor with n=<7m you have eliminated a double check. If it's more than the currrent threashold of ~7.5M you will have eliminated 2 checks.

Sad news is there is a point of diminishing returns, good news is we are not there yet. Our current client can sieve upto 1000T so basically we are half way there. Weather we sieve past 1000T is questionable, but there are k/n pairs beyond 20m that we will have to test as well, but not for a few years. Hopefully the project will have eliminated alot of k's by n=20m.

My main point is we should try to drive up the sieving T up as fast as possible to some T value and stop sieving.

Is our limit 1000T???

So lets set a goal off 1000T finishing up to 1000T.
So how long will it take us to get to 1000T???

Well that of course depends on how many people we have helping...

Here are some numbers....

Currently we are finishing around ~1000G per day which is rougly 11,500 kp/s.

At this rate we will get to 1000T in about 16 months.

So how many computers will it take to sieve to 1000T or 1P, in 6 months?

Well if we increased our total sieve speed to the <b>primorial rate</b> of 30029 kp/s we could do it.

This could be done with a combination <100 of reasonable xp1800 and P3-1000mhz computers.

If I had a hundred of "my fastest sieving computer" (a 2500mhz Barton) it would only take about 3 months.

Thank-you all for helping out.






I think the goal of 1000T (the current limit of proth cmov) in a reasonable time frame is a good one.

Then, give sieving a rest for a while and focus power elsewhere, look at what the factor density is etc.

If we get this far this quickly we could do a couple things.

Note: X=the current double check level.

1. Do a 20m<n<100m dat file to 1000T, then start from 1000T up with a X<n<100m dat file for p>1000T.

2. Start doing everything above 1000T with a x<n<100m dat file and new client???

3. Concentrate slow machines on double checks.


Best bet is probalby to #3 for a few months then, my preference is to do #2 until we get to a n value of say 18m. Then release a new dat for x<n<100m for everything p<1000T. We would be able to all p<1000T quite quickly before prp reaches 20m.

[Keroberts1]

we have experimented before with n ranges from x to 100M and the dat file is enormous. We also experiences tremendous slowdowns that were very possibly resulting from the design of prothsieve. I believe that if Mklasson would help out we could probably create a much better client to be working on that range. However my other input comes in the form of deciding when to stop the current sieve. The point of diminishing returns as far as value of eliminating a test vs the time it takes to find a factor is far beyond 1000T. Also, as we get to higher values the factor density dropps slower and slower. The only point when sieving becomes alot les valuable is when most of the factors we find are already passed by the sieve effort. So its more of a question of PRP depth than sieve depth. Or right now its a question of sieve depth but prp depth will be more important soon.

[vjs]

Yes I know about the other dat files, I downloaded some of them a couple months back.

I tried to test the 1m<n<100m dat file (29.2 Mb) on my machine the other day and it wouldn't work. Or at least I let it sit there for a very long time and it didn't report.

However the 20m<n<50m file (10.9Mb) worked just fine.

My speeds 1m<n<20m = 650 kp/s @500T
Same Machine 20m<n<50m = 620 kp/s @500T

The above is comparing a 20m range vs a 30m range. I just wonder if it makes sence to start collecting factors past 20m. I wouldn't restart the sieve from 0 just continue on with a new dat. Something reasonable

(double check prp)<n<(Maximum possible with <50% slow down)???

If square root holds true this would mean a 1m<n<80m dat file.

The only dat I didn't download to test :bang:



As for the factor density issue, yes the graph of factors vs p looks alot like a y=1/x plot.

In other words as we go to higher T we get less factors per G.
We have been on the "tail" portion since about 100T, I believe, and the number of factors per G isn't decreasing very fast anymore... this effect has been pretty much a non-issue since 400T.

But my question is "Are the number of factors actually the number of factors that actually eliminate a new k/n pair?" Not duplicates etc, I'm pretty sure it's eliminates a new k/n.

Anyways I think our points are the same, start using a larger range, or stop with the current range at some sieve value or prp value.

The question was regarding goals, all projects and sub projects should have some goals it makes them more interesting, I think sieving to 1000T/1P (The limit of proth cmov 0.42) is a reasonable goal. At 1000T we may reconsider where we are, new dat, new client, etc.