View Single Post
Old 2007-10-05, 15:26   #6
gd_barnes
 
gd_barnes's Avatar
 
May 2007
Kansas; USA

2×3×19×89 Posts
Default

Quote:
Originally Posted by VBCurtis View Post
Since multiple people are interested in supplying LLR work, I set a second core on sieving; I should be able to provide a 25k chunk every 5-6 days this way. I suppose I can also suspend my fanatic over-sieving of ranges; the 500-530 chunk will be sieved to nearly 30T as it is.
Oh, NOW the truth comes out! I just KNEW you were over-sieving! (Although for the huge range of n=470K-4M that you're sieving, I can't really tell.) Great work! With your sieving, LLRing on k=470K-500K is just smoking...at n=483K this morning...running on 2 cores after about 30-35 hours. With the fftlen jump at n=476K, it was about half at the faster rate and half at the slower rate at this point.

I'll be out of town from tonight until Sunday and my k=5 machine may just end up sitting idle for a good portion of Sunday without a concerted effort to switch LLRing between machines, which I don't want to mess with. Darn the idle CPU cycles!


Quote:
Gary, I'll send you 530-550 roughly next friday, around 33T on the sieve. Since the FFT length just jumped at 480, this range should be pretty close to optimally sieved at that level. I think there will be another jump around 550-575, so I'll run the sieve quite a bit deeper before releasing 550-600.

-Curtis
Great! Works for me on 530-550.

I'll run some quickie tests to see where the fftlen jumps again. No use to over-sieve more than you have to!

Final note to all...I still plan on splitting any first prime that I find on this with BlisteringSheep, who has contributed a lot of sieving to many efforts at RPS without being able to LLR them.


Gary
gd_barnes is offline   Reply With Quote