[QUOTE=gribozavr;93193]Not yet...
Celeron is sieving 1000T and on,
Athlon is doing 959T-1000T.[/QUOTE]

At what pace is the sieving removing candidates? If it nears the 2 minutes a k-candidate, then the entire sieving range should be released and sieving on another range begin :) Also how high a range are we sieving? (25G or 50G range?)

On athlon: 1 k every 28 sec
On celeron: 1k every 71 sec

And, as far as I can see, the rate isn't going to drop soon: it is near-constant since around 600T.

I started to sieve 100e6-25e9, but now I'm sieving 1400e6-25e9

[QUOTE=gribozavr;93206]On athlon: 1 k every 28 sec
On celeron: 1k every 71 sec

And, as far as I can see, the rate isn't going to drop soon: it is near-constant since around 600T.

I started to sieve 100e6-25e9, but now I'm sieving 1400e6-25e9[/QUOTE]

Ok, there is still enough to gain on sieving. I think then that it is all in place to release 100M maybe 200M to PrimeGrid, in stead of 1G+ ranges, sinces, you are removing ~8 candidates every time my computer is removing 1 candidate :) Any approximity of the amount of k's left? I guess it must be around 8,260,000 based on a rate of 350 remaining candidates for every 1 M range - is this a close estimate or am I far from the real amount of n's left?

Now there are no more than 8308076 k left. Even I don't know the real value because I would have to merge the both computers' files.

Just to let you all know, I'm from PrimeGrid. I have all 6 of my computers running the TPS beta 100% 24/7. They consist of:

K6-2+ @ 400 MHz
K6-III+ @ 600 MHz
2 X PIII @ 1.0 GHz
AthlonXP 2100 (452) @ 2.04 GHz
Athlon64 3000 (939) @ 2.19 GHz
Athlon64 3700 (939) @ 2.66 GHz

I have an Athlon64 3000 (939) and 3500 (939) showing up within the next week or two, and they will be on 24/7 also. After I find the ultra stable overclock of each of course. :P

[QUOTE=KEP;93207]Ok, there is still enough to gain on sieving. I think then that it is all in place to release 100M maybe 200M to PrimeGrid, in stead of 1G+ ranges, sinces, you are removing ~8 candidates every time my computer is removing 1 candidate :) Any approximity of the amount of k's left? I guess it must be around 8,260,000 based on a rate of 350 remaining candidates for every 1 M range - is this a close estimate or am I far from the real amount of n's left?[/QUOTE]


I think that is more practical (for Gribozavr, me and Rytis) to have 1G release instead of 100M. Moreover the 1G release is for all TIPS users, not only Primegrid.

Rytis have you solved the bugs of beta LLR version ? When do you think it will become official version (days, weeks,... ) ? Do you think to have both LLR and Primegen running or will you stop Primegen ?

I am sorry guys, I will not be able to maintain the system for a few days as I have some big problems in my family (my grandmother died. RIP :( ). This also probably means that we'll run out of LLR work on PG.

If you could get me a big range (300M?) all in one file, I would probably be able to insert it, but don't expect any more fixes for now.

[QUOTE=Rytis;93392]I am sorry guys, I will not be able to maintain the system for a few days as I have some big problems in my family (my grandmother died. RIP :( ). This also probably means that we'll run out of LLR work on PG.

If you could get me a big range (300M?) all in one file, I would probably be able to insert it, but don't expect any more fixes for now.[/QUOTE]

My sincere condolences, Rytis.

I asked Gribozavr to send me 1400-1700M; if he sends me the file I'll assign this range to Primegrid.

condolences
 

My condolences Rytis…very sorry to hear that. Take your time...

------------------------------------------------------

TPS…correct me if I’m wrong but it seems impractical to be LLRing right now when sieving can remove 8 k’s to LLR’s 1 k.

Riesel Sieve already has a BOINC sieving effort going on in which Rytis helped them out quite a bit…maybe they can return the favor in setting up TPS BOINC sieving.

By running LLR and Sieve together, you give people the option to separate their intel/amd machines to run what’s most optimal for them.

As PrimeGrid TPS has already run out of WUs, I’m heading to RS for a while until Rytis returns.

[QUOTE=jmblazek;93412]My condolences Rytis…very sorry to hear that. Take your time...

------------------------------------------------------

TPS…correct me if I’m wrong but it seems impractical to be LLRing right now when sieving can remove 8 k’s to LLR’s 1 k.

Riesel Sieve already has a BOINC sieving effort going on in which Rytis helped them out quite a bit…maybe they can return the favor in setting up TPS BOINC sieving.

By running LLR and Sieve together, you give people the option to separate their intel/amd machines to run what’s most optimal for them.

As PrimeGrid TPS has already run out of WUs, I’m heading to RS for a while until Rytis returns.[/QUOTE]

Unfortunently distributed sieving is not as easy as it looks, you'll will have a lot of work done, which isn't nescessary, since you'll be working on big datapackages containing the same k's for a lot of different ranges of p. So a lot of candidates will show up as factored several times, while that want be the case when doing sieving on one machine in a very undistributed way.

Actually riesel sieve returned the favor, by helping rytis set up the LLR client, by letting him recieve the scripts needed to run the LLR work :) But of course one favor can always be returned with more than one favor back...

Finally, my condolences to you and your family Rytis, just take your time my friend...

Regards!