Stats milestone
 

I was just looking at he stats page and I think there was a time when the stats went up to 2 million, and there was arround 125,000 candidates to test and 2milllion seemed so far off in the distance considering we were prp'ing around the 250,000 mark.

I just checked and there are just 2106 test to go until all candidates are tested to a minimum of n=2,000,000. and that will be a great milestone for the project.

I'd like to give a BIG thanks to all who have contributed in anyway at all, esp. for Lars :bow: for his brilliant job on this project.

Thanks

Foots

I agree:bow: :bounce:

Let's hear it for Lars!!!!! :banana:

Greenbank posted something about his ranges in this thread, which I do not remember, I guess it is lost now. May be he can post it again.

Citrix

It was about getting drunk and manipulating his machines so they don't fall idle.

But what he absolutely should repost is this link to his personal stats page.

H.

Milestones to achieve in the near future.

PRP all candidates below 2M
Total untested <1.5M
Sieve to 100 trillion
Double check to 400k
Find one more prime if possible:smile: :bounce:


Citrix

[url]http://www.greenbank.org/cgi-bin/proth.cgi[/url]

I'm awake again (and sober). I'll be updating and improving it later on this afternoon.

[QUOTE=Citrix]Milestones to achieve in the near future.

PRP all candidates below 2M
Total untested <1.5M
Sieve to 100 trillion
Double check to 400k
Find one more prime if possible:smile: :bounce:


Citrix[/QUOTE]
One down four to go.

Double check is larger then 400K.

Thanks to some work by Rogue (Mark R) and myself, I've got a nice little speed increase:-

This is combined SoB+PSP sieving on a 2.5GHz G5 PPC running MacOS X 10.4:-

Old:-

pmin=125686741988123 @ 321 kp/s
pmin=125686751988137 @ 357 kp/s

New:-

pmin=125686761988153 @ 589 kp/s
pmin=125686771988177 @ 585 kp/s

:-)

ETA for my 8T range down from ~43 days to ~26 days.

WOW:w00t:

Is that only for the G5 or everybody? New releases planned? Give us more information, pleeease.

But we are not yet done with the acclamations: :bow: :bounce: :shock: :banana: :banana: :banana:

[QUOTE=hhh]WOW:w00t:

Is that only for the G5 or everybody? New releases planned? Give us more information, pleeease.

But we are not yet done with the acclamations: :bow: :bounce: :shock: :banana: :banana: :banana:[/QUOTE]

It's G5 specific I'm afraid.

I just tried the same changes on the x86 code and it made almost no difference (the rate was sometimes higher, sometimes lower than before, within just 1%).

Still got quite a way to go before a fully tested release is available. I'm still relying on GMP for a few functions (multinv and gcd). I then have to test it against all previously found factors to make sure it doesn't miss anything!

Green bank, is the MAC binary available for everyone?Where do I get it? I have a G5 sitting idle, I could put it to some good use.

Citrix

[QUOTE=Greenbank]pmin=125686761988153 [b]@ 589 kp/s[/b][/QUOTE]

I remember when SoB finally broke the 100 kp/s barrier. :nostalgia:
Does your current version top the 1 mp/s frontier when only working on the 8 SoB k's? :banana:

Just found it: [url]http://www.free-dc.org/forum/showpost.php?p=100260&postcount=20[/url] :banana:

[QUOTE=Citrix]Green bank, is the MAC binary available for everyone?Where do I get it? I have a G5 sitting idle, I could put it to some good use.

Citrix[/QUOTE]

Not quite yet, it still relies on GMP for some functions (gcd and multinv), and you have to patch GMP with a bunch of other files to make it faster for G5 PPC. Not the easiest thing to do.

If I get any time today I'll be working on removing the need for GMP. Then I have to test it all thoroughly.

Citrix,

Right, I think I'm almost there for the G5 binary. It no longer relies on any GMP and so it is much easier to use.

A few questions though:-

Which version of MacOS do you have on your G5?[LIST][*]it must support 64-bit binaries, so that means a minimum of 10.4.x[/LIST]Is it a single, dual, single-cpu dual-core of dual-cpu dual-core?
What speed processor(s)?
How much memory?

Will you be sieving PSP, SoB or PSP+SoB combined?

I will be sieving combined ranges. I do not know much about the MAC, I am not a mac user. I know that I am allowed run DC applications on it. So if there is anything specific you want to know, I will have to go and look it up.(Which may take upto a week)

OK, I've been fiddling and found a few more % here and there.

Speeds as follows for a 2.5GHz G5 PPC:-

SoB: (8k) p=1800893475068332 (yes, that's 50 bits!) at 1100kp/s
SoB: (8k) p=1023348220675453 (49 bits) at 1120kp/s
PSP: (12k) p=80T (46 bits) at 867kp/s
Combined SoB+PSP (20k) p=125T (46 bits) at 630kp/s

Tomorrow I'll PM you with details of where you can download it if all goes well. It just has to run through my set of test ranges which takes several hours (I run them overnight).

[QUOTE=Citrix]Milestones to achieve in the near future.

PRP all candidates below 2M
Total untested <1.5M
Sieve to 100 trillion
Double check to 400k
Find one more prime if possible:smile: :bounce:


Citrix[/QUOTE]

Ok, so now less than 1.5M candidates are left. 2 down!!! :bounce: :bow: :showoff:

[QUOTE=Greenbank]OK, I've been fiddling and found a few more % here and there.

Speeds as follows for a 2.5GHz G5 PPC:-

SoB: (8k) p=1800893475068332 (yes, that's 50 bits!) at 1100kp/s
SoB: (8k) p=1023348220675453 (49 bits) at 1120kp/s
PSP: (12k) p=80T (46 bits) at 867kp/s
Combined SoB+PSP (20k) p=125T (46 bits) at 630kp/s

Tomorrow I'll PM you with details of where you can download it if all goes well. It just has to run through my set of test ranges which takes several hours (I run them overnight).[/QUOTE]

Thanks!

Only 127 tests left to go where n < 2M.
And only 2.8T below 100T to be returned as completed sieved.

Citrix, any luck with the G5 version of proth_sieve?

[QUOTE=Greenbank]

Citrix, any luck with the G5 version of proth_sieve?[/QUOTE]

Haven't had time to test it. Just running busy. It might take me a while.

Green bank, I ran proth_sieve on the MAC computer. It did not work. The computer said fastsieve needs fastsieve application to work. (Don't know what that means):popcorn:



I and mac's do not mix well:furious::rant: I will stick to pc's in the future.:alien:

All tests under 2.0 M have been done once.


Last goal of finding a prime is left.

Can we set a new PRP target (or 2)

PRP to 2.5M
PRP to 3M
find a prime.


Can we have a counter of PRPs to these values on the stats page, just like
the <2.2M that is already there.

Cheers
Martin

No problem to include the suggested milestones in the stats.
I hope to include them this weekend but at first i have to catch up with all the mails.

Lars

cheers for that. looking good and something to aim for.

only 4K tests to 2.5M! :banana:
only 20k tests to 3M! :yucky:

we will eat those tests up! :popcorn:

Could we have the numbers of tests under 1 Million digits (3321929) instead of 3M.

Lets try to reach 1 Million digits by the end of the year. I think it is a possibility.:rolleyes: :banana:

There is no way we will hit 1 million digits by the end of the year with out an influx of new users. Each test takes longer as the value of n increases. As you start to get to 3 million the rate of tests completed in 30 days will only be about 60% of what we do now.

Of course that doesn't stop us from trying :)

I was hoping we would find a prime soon and that would even out the increase in size of n

we are at <20k tests for 3M

and 30800 for 1 million digit. Does the sieving at the current depth eliminate candidates at this amount? Because it seems to drop faster than tests are being done.

I am with Citrix, 23k tests done in the last 180 days. thats 1/2 year. With an elimination due to sieving and finding a prime, plus some new members, 1 million by the end of the year is not unreachable. That should be the goal.

WOOO!

[B]Does the sieving at the current depth eliminate candidates at this amount? Because it seems to drop faster than tests are being done.[/B]

Sieving at any depth removes candidates at any n level.

can we have some stat about the number of residues for which the double check and the first check did not match and a third check was needed.

Thanks.

Citrix,

I'm pretty sure that LTD commented about this before the number of non-matching residues is zero or very close to zero thus far.

This makes sence for PSP since errors seem to be more time dependant as opposed to n-size. The PRP tests for PSP are still very very short relative to mersenne or SoB. I personally wouldn't worry about missed primes yet. Think of it this way the first pass n value is less than half of sob's second pass n and computers are alot faster now for PSP compared to SoB or mersenne when their n's were at the same level.

As for sieve, the question about is sieve eliminating candiates faster than prp, the answer is probably yes and no. We are probably eliminating more k/n pairs per day over the entire 991-50M range. However, I'm pretty sure PSP is testing more n in the 2-5M range than sieve is eliminating.

I'd like to see a "all things sieve page" for PSP not nessarily a daily update but something monthly or weekly?

In either case the sieve should certainly continue at least, I'd even suggest more effort in that area.

Thank you for the information. I was basically trying to get the stats page to refelect if any triple check were pending so we could finish them, as they arise.

For the moment i see no chance that i will find the time to create something like "all things sieving". There are to many other things that need to be done.

That normal stats pages need to be redesigned. They are very ugly in my opinion.
Then there are the addional stats requests like showing factors removed
from reservation files by sieving or some performance graphs.
There are some more internal optimisations needed.(At the moment the stats run does more then it needs to do)
Very important also. I like to have a web based factor/residues submission form that puts the manual results into the DB automatically.

My problem is that there is the real world alsowhere i have a very demanding job.


One more thing if there is anybody who has suggestions for better structured stats please tell me. I am a programmer and not a designer. I never manage it to produce something with a nice look and feel. ( One of the reasons why i only programm backend things in my job).

Lars

Edit: Forgot to answer the question about the error rates. Still the rates are extremely low. Problem with giving numbers is that sometimes it is not possible to distinuish if there was an wrong residue or if one of the result comes from an older llr client which produced PRP compatible results and the other one came from the new client. There is no way for the llrnet server to prduce such an information.
To be sure i would need to make a fourth check. (which i make sporadic when there are alot of errors from one user tomake sure that it is only incompatible results and not a PC problem)

[QUOTE=ltd]
Then there are the addional stats requests like showing factors removed
from reservation files by sieving or some performance graphs.


Lars[/QUOTE]

How many tests have we done so far, for which a factor was found later?
(No need to put this on the stats page, I am just curious about the number of avoidable tests)

Thanks

That is easy to answer.

When looking at all k in the DB we have 10190 factors for prp tested candidates while there were 146957 prp tests done inb total.

When we look only at the k values that are active at the moment we have the following numbers: 92741 prp tests done and 7183 factors for candidates which are already prp tested.

Thats alot of CPU time we could have saved.

One more number that shows that we should not push double cheching:

When looking at the active k values and check how many factors are found for candidates that have already been double checked you find that there are 1764.

The factors returned today by vaughan, hhh, and AMDave alone saved us from doing 14 double checks and even better it removed a test that would have been handed out by the llrnet server within the next 24 hours.

So sieving is very important and i encourage everybody with an AMD to help sieving.

I think we will find this to be true even with the current levels of prp until sieve reaches about 1000T. More than 5X higher than the current level, if sieve were around this level then I'd think about pushing prp a little harder.

I'd seriously not even think about secondpass or error rates until prp = 5M. At that point at least 1 k and probably more like 2 or 3k will be elimated.

Could p-1/ECm/P+1 have avoided all those tests and saved time? Should we start a factoring thread. We are almost at 2.5M.
How long does each p-1 test take and what bounds will be best?

My internal tests show that with our sieving depth at the moment "p-1" would start to make sense somewhere around at least n=3.5M.

For "p+1" or "ECM" the numbers are even worse as for p+1 you need three runs with around the same runtime as "p-1".

And very simplified "ECM" is runing lots of "p-1". (Yes i know that you do not run "p-1" but the runtime per curve is in the same range)

So there is no time saving by trying these factoring methods.

Lars

Could you post your calculations?

I do not have the exact numbers anymore but i will try to remember as much as possible.

I did the test when we were sieving around 120T (we are now at 160T).
I took 100 pairs in the 2.5M range and ran "p-1" tests on all of them.
Sorry but i am not sure anymore about B1 and B2.
Then i counted the number of factors and the time needed to find that on average finding a factor needed more then 3 times doing a PRP test.
Then i took a small snapshot at 3.5M (only 40) and found that the time moved to a little bit over a factor needs around two times a prp test.

So with beeing above 160T in sieving and with a factor beeing worth something between 1.6 and 2 times a PRP test i see no use in doing p-1 at the moment.

Lars

Thanks for the information. I will stick to PRP and sieving.

I trust the B1B2-choice of prime95 until further examination. So, if it says this number doesn't need P-1, I would let it be. It says so at the moment.
Let's sieve so that this remains like this.

We have a new best performance ever.:banana: :banana:

In the last seven days we were over 100 results every day on the llrnet server alone. :showoff: :showoff:

Look here for a display of the increased performace within the last 30 days.

[url]http://www.psp-project.de/llrnetstats.php[/url]

:w00t: :w00t:

Edit: at least some of the typos.

It is because of PCZ. Nice graph!

yeah, the top end of the LLRNET range is at 2.65M!!! while the reservations are almost taken up to 2.5M

Wont be long until 3M at this new rate.

[QUOTE=Footmaster;81782]There is no way we will hit 1 million digits by the end of the year with out an influx of new users. Each test takes longer as the value of n increases. As you start to get to 3 million the rate of tests completed in 30 days will only be about 60% of what we do now.

Of course that doesn't stop us from trying :)[/QUOTE]

Well, thanks to outInder006 and PCZ, chances are good that we reach 1M digits in time. :smile:

[QUOTE=Citrix;73518]Milestones to achieve in the near future.

1. PRP all candidates below 2M
2. Total untested <1.5M
3. Sieve to 100 trillion
4. Double check to 400k
5. Find one more prime if possible:smile: :bounce:

Citrix[/QUOTE]

1. Check!
2. Check!
3. Check!
4. Check!
5. Check!

:george: (I like this one!)



Let's propose new milestones:

1. PRP all candidates below 4M
2. Total untested <1M
3. Sieve to 400 trillion
4. Double check to 1M
5. Find one more prime if possible:smile: :bounce:

Your milestones depend on each other. For instance, 5. (prime) is necessary for 2 (number of tests in queue)
I like 3(sieve) , and think 4 (double check) is too high (low error rate), but OK.

A year or two?

[QUOTE=hhh;86827]A year or two?[/QUOTE]

Let's see. :wink:

I am with Mystwalker. We have had some heavy duty crunchers recently who have done alot to increase the LLRnet daily figures quite a bit. As we have just gotten a new prime, there is hope for these milestones.

My new crunching soopa-computa arrives tomorrow to help in the fight. I cant wait to see what these crazy new Core 2 processors do. :geek:

I'm quite interested in how your core 2 system handles this stuff, both the llr and the sieving... quite interested. Myself, I've been buying amd x2's as the amd handles the sieving pretty decent. Of course not compared to the powerpc chip.

And in regards to the "goals" it's really getting harder and harder to place guesstimates as the technology is rocking right along. Just really depends on how much new tech that people interested in this project purchase.

I have my pc up and running.

I have yet to try sieving, as I had a wee look and P4 is supposedly the best at PRPing. But I will try when my run ends.


I have a Core 2 Duo 6400, 2 2.4Ghz cores. 2Ghz DDR2 800Mhz memory, Sata II harddrive @ 8.9ms response time........and a beverage holder on the front. You push the wee button and the beverage holder slids out and is perfect for a cup of tea.
I am running 2 LLR V3.70 at 7.4ms per iteration. Which I think is pretty good. My 3Ghz P4 at work runs 1 at 9.4ms per iteration. so it is at least 100% increase over that.

will repost my sieving performance when I try it out in a few days. But for other things the Core 2 is very fast.

[QUOTE=Chino112;87872]
will repost my sieving performance when I try it out in a few days. But for other things the Core 2 is very fast.[/QUOTE]

That would be truly decent!!!!! and also post the range that you ran it on so I can do a comparisson here on my stuff. I'm really curious as I'm holding off on buying anything until I see what the sieving performance is on the duo. I've been sieving here so far, and don't see changing that. :)

Bruce

I sent you a PM.

But I am getting around 900kp/s over two instances. so 450-465kpp/s on each individually.

Hope this is good, cause I love it!

yep, that's doing pretty good on combined stuff. :)

[QUOTE=statswebsite]Countdown to finish all n<2^21: 4 tests to go[/QUOTE]

why not move the lowest tests from time to time in the llrnet server? those 4 tests are pending forever :wink:

they have been reserved by someone have they not.

they are taking a while tho....

I'm generally a big double check and factoring P-1 advocate in SoB.

But for PSP lets not even talk about double checking or P-1 until we reach 5M. I think this is will be a reasonable point to consider P-1 again, it will all depended on sieve depth.

In the mean time lets try to sieve as deep as possible. 400T by the end of the year is IMHO do-able with some work, we will be close regardless. If we can push hard in sieve we won't have to P-1.

400T? You meant 300. Soon it's X-mess, and then the end of the year is already there.
Well, anyways, you are right, the sooner we sieve the better. Personally I would stop the SoB-only sieving at Mike's reservation page and sieve even that together with PSP. But that shall not be my consideration. Now, in the wintertime, some top-producers are back, so I am optimistic that sieve remains popular.
With my P4, I find a factor every few hours, and could run 3 tests a day; but the factors save much longer tests, for higher n, so it's a good buissiness to do sieving.
H.

Nope I really mean 400T,

We only have about 40T to go until 300T. We should be able to do that. There are also a few ranges between 300T to 400T that are already complete. I'm hoping we make a big push for 400T and get the majority of it done. I try not to set goals that are easy although this one might be way to hard. We will see where we are Jan 1 07.

I also don't think finding another prime soon is unreasonable.

214519*2^1929114+1
and
222361*2^2854840+1

are pretty far appart even when comparing the previous ones. Hopefully the prime gods will treat us to another in short order rather than making us wait too long.