[QUOTE=KEP;389952]I'm throwing 8 cores on R3, so I'm making this reservation:

Starting new range:

minK= 13000000002 (13G)
maxK= 21000000000 (21G)

I'm going to use the starting up bases script to n=2500, using "pfgw_win_3.7.8_20141127". (ETA for this part is about March 1st 2015).

From n=2501 to n=25K, I'm going to use a sievefile and primality test with LLR 3.8.13 (using LLR's StopOnPrimedK=1).

Regards

KEP[/QUOTE]

I found those sieve files with many k values to be annoyingly slow work. The most pleasant way to do this (in my opinion) is to use the starting script to n=25K and invoke it with parameter -f30 which gives a considerable overall speedup.

[QUOTE=Puzzle-Peter;389972]I found those sieve files with many k values to be annoyingly slow work. The most pleasant way to do this (in my opinion) is to use the starting script to n=25K and invoke it with parameter -f30 which gives a considerable overall speedup.[/QUOTE]

It may be the most pleasent to use the starting up script, to n=25K, however, it is not the most efficient. The difference between running the starting up script to n=1000 and n=2500 is also very big. In stead of ~200000 k's remaining at n=1K, I'll only have to sieve ~80000 (or less) remaining k's, from n=2501 to n=25K. It is always much faster to set the -f to 0 and then run to a low n limit, since factoring is much slower than doing a single primality test (at least thats what I seem to feel like) for n<=2500 on both my Hasswell and my Sandy Bridge.

Thanks for your input.

Take care :smile:

Kenneth

Kenneth, I am going to decline this large of a reservation on our pages. Please reserve ranges in much smaller pieces such as P=1G or 2G at a time and come back as you complete your ranges to reserve more. It is easy to lose interest and other participants may wish to join in in the mean time and there will be a big hole in the search range in the mean time. You will find that the size of the sieve files to be difficult to deal with and there is no efficiency gain by sieveing millions of k's vs. hundreds of thousands of k's. Thank you.

[QUOTE=gd_barnes;390107]Kenneth, I am going to decline this large of a reservation on our pages. Please reserve ranges in much smaller pieces such as P=1G or 2G at a time and come back as you complete your ranges to reserve more. It is easy to lose interest and other participants may wish to join in in the mean time and there will be a big hole in the search range in the mean time. You will find that the size of the sieve files to be difficult to deal with and there is no efficiency gain by sieveing millions of k's vs. hundreds of thousands of k's. Thank you.[/QUOTE]

Okay :smile: ... no further work to be done on R3 ... I may stress, that factoring compared to sieving is much slower and the time it will take to bring a 1G range from n=1 to n=25K on a single core (using my initial combination, including sieving) is only a little more than 3 months on a Hasswell and a little more than 4 months on a Sandy Bridge, so the reservation were not all out of proportion. However, I'm not going to continue my reservation and I'm not gonna make a new reservation for a smaller range. :smile:

[B]Double Reservation[/B]

k=13G-14G as new to n=25K

k=9G-13G for sieving from n=25001 to n=100K, this range is split as follows:

k=9G-10G with 6014 k's
k=10G-11G with 5808 k's
k=11G-12G with 5962 k's
k=12G-13G with 5984 k's

I hope this reservation can be accepted :smile:

KEP

Reserving R3 from n=100-125k for BOINC (t17_b3_20M-1G_100k-500k)

R3 20M<k<1G 100k<n<125k done

244 primes found - 1217 remain

Results emailed - Base released

R3 k=7e9-8e9; n=25e3-100e3 COMPLETE

Result mailed

Lennart

Reserving R3 from n=125-150k for BOINC (t17_b3_20M-1G_100k-500k)

What about finishing R3 upto 25k?

[QUOTE=henryzz;391069]What about finishing R3 upto 25k?[/QUOTE]

This would require Rebirther to run the new bases script in PFGW which he has not implemented. Plus there are huge files to be uploaded. This should be taken on by a group of individuals.
If we find a few people interested in doing so I'll throw in a few cores starting mid-January.