I have begun to sieve 1G-2G from 100k to 500k. It won't proceed quickly.
If rebirther is interested in boincing the file, I'll submit it; if not, I will test some of it myself.

[QUOTE=VBCurtis;398993]I have begun to sieve 1G-2G from 100k to 500k. It won't proceed quickly.
If rebirther is interested in boincing the file, I'll submit it; if not, I will test some of it myself.[/QUOTE]

Please STOP your sieving. Lennart has already sieved from k=20M to k=4G for n>100K to n=500K. You can as far as I know (it may be wrong) begin sieving the range k>4G to k=7G from n=100K to n=500K :smile:

Thanks for your support.

Great, thanks! I only got to 1e9 anyway, part-time office machine.

I'll begin 4-5G, from 100k to 500k. Since there is so much already sieved, I'll let it run at the office for a while.

[QUOTE=VBCurtis;399028]Great, thanks! I only got to 1e9 anyway, part-time office machine.

I'll begin 4-5G, from 100k to 500k. Since there is so much already sieved, I'll let it run at the office for a while.[/QUOTE]

Nice :smile:

Reserving k=16G-20G to n=25K.

[QUOTE=VBCurtis;399028]Great, thanks! I only got to 1e9 anyway, part-time office machine.

I'll begin 4-5G, from 100k to 500k. Since there is so much already sieved, I'll let it run at the office for a while.[/QUOTE]

I was disappointed to discover sr2sieve does not work for this range. The documentation for sr2sieve states k must be smaller than 2^32, which is a tick over 4G. So, Lennart's 1G-4G sieve is the highest that can be done with sr2sieve.

In the distant future of R3, I think that means we'll have one sieve for all the k's under 4G to leverage sr2's speed, with the other 92% of this base stuck with relatively slow srsieve.

I'm at 1e9 on the sieve, barely started. Just under 17M candidates in the sieve presently. A laptop core manages 10kp/sec, while an i5 desktop does 22kp/sec. Slow.

R3 k=3677878 is complete to n=2M, with a great deal of help by the users at BOINC. It took approximately 3 weeks to do in BOINC, what would have taken me at least 14 weeks to do :smile:

Way to go BOINC. Thx for the support Reb. Looking forward to seeing srbase reach 20 TeraFLOPS (and preferably higher)

For R3 k=9G-13G, 9912 k's remain and the testdepth is at n=66K. ETA is about 85 days from now. Dependant on how much of a slow down it will be going from FFT lenght=12K to FFT length=16K.

Ps. I have a sieve file for R3 k=3677878 wich includes the range from n=2000001 to n=5M, wich is sieved concurrently with the n=1000001 to n=2M range and is already sieved to p=192T. If anyone want's to take this k for a further swing, I'm more than willing to hand it over to anyone interested in either preparing the sievefile for BOINC or sieve it further for a personal effort. At p=192T, there remains 50594 candidates in the sievefile for n=2000001 to n=5M :smile:

I was disappointed to find that srsieve 1.0.4 win64 version produces "DOES NOT DIVIDE" errors starting at 4294e6 for my 4G-5G base 3 sieve. Each error causes srsieve to crash, and is repeatable. I tried skipping the bad p, but the program runs no more than a few seconds before producing another error.

I happen to have 0.6.17 here from the old days, which is running through the problem range without error. Speed is within 2% of 1.0.4.

Hrmmm... just checked the google-pages download spot, and 0.6.17 is latest-source. I wonder what this 1.0.4 is!! Guess I won't use it anymore.

[QUOTE=VBCurtis;400617]I was disappointed to find that srsieve 1.0.4 win64 version produces "DOES NOT DIVIDE" errors starting at 4294e6 for my 4G-5G base 3 sieve. Each error causes srsieve to crash, and is repeatable. I tried skipping the bad p, but the program runs no more than a few seconds before producing another error.

I happen to have 0.6.17 here from the old days, which is running through the problem range without error. Speed is within 2% of 1.0.4.

Hrmmm... just checked the google-pages download spot, and 0.6.17 is latest-source. I wonder what this 1.0.4 is!! Guess I won't use it anymore.[/QUOTE]

I have the source and include it in the distribution. It is a miscompile. 1.0.4 had problems. You should be using 1.0.8.

Reb sent me 1.0.6, which is working fine. Thanks for your reply!

[QUOTE=VBCurtis;400636]Reb sent me 1.0.6, which is working fine. Thanks for your reply![/QUOTE]

That version should work as well.