Optimal depth?

[vjs]

A bit of a wag LOL...

But I took some offense to axn's post. ( sorry to point a finger directly that's not my intent).

As a reality check: sieving from 25P to 200P will eliminate roughly 1-ln(25P)/ln(200P) = 5.2% candidates only. So as a rule of thumb, no more than 5% of resource should be spent in sieving.

So as a reality check could someone run some numbers for me.

1-ln(50P)/ln(200P) =
1-ln(100P)/ln(200P) =
1-ln(150P)/ln(200P) =



and

1-ln(400T)/ln(1P) =
1-ln(500T)/ln(1P) =
1-ln(400T)/ln(600T) =




Also as to

: Prime Grid

[Bananaman42]

So as a reality check could someone run some numbers for me.

1-ln(50P)/ln(200P) =
1-ln(100P)/ln(200P) =
1-ln(150P)/ln(200P) =



and

1-ln(400T)/ln(1P) =
1-ln(500T)/ln(1P) =
1-ln(400T)/ln(600T) =

1-ln(50P)/ln(200P) = 0.0348
1-ln(100P)/ln(200P) = 0.0174
1-ln(150P)/ln(200P) = 0.0072

1-ln(400T)/ln(1P) = 0.0265
1-ln(500T)/ln(1P) = 0.0201
1-ln(400T)/ln(600T) = 0.0119

The above results are rounded to 4 decimal places.

[vjs]

1-ln(400T)/ln(600T) = 0.0119

So if we currently sieved to 400T and were thinking about sieving to 600T we should only devote <3% of our total resources? Just checking... on the logic.

[Freeze_XJ]

I can see the logic of all the logaritms, but i prefer a faster way to calculate efficiency :
Right now i'm popping out a factor a day or so when sieving, and a full PRP-test takes about a month or more. (running 4core, so should correct for that... let's say a week) That means that sieving is about 7 times as efficient as is, so i'll continue sieving :P

@vjs : just a bit more than 1% of total resources... and probably not worth the effort... But hey, we're not even close yet Once sieving reaches 50T we reevaluate, can always switch, but it'd be much better if the PRP-ers switch to us first

[vjs]

yup...

simple math is generally pretty good.

Way back when a few of us were screaming for sievers...
We were pretty much ignored except for; the faithful few, those we begged, those be bribed, and those who knew...

Now that we are starting to approach APPROACH optimal numbers

[enderak]

Can someone let me know if my "back of the napkin" figuring here make sense..?

According to primestats.net, I am averaging about 2.5 factors/day when I run BOINC PSPSeive all-out on this machine. 6 out of the 15 k being sieved are applicable to SoB (40%) so essentially I am finding about 1 factor per day that benefits SoB, which saves 1 future PRP test. (Assuming a prime isn't found for that k before then, in which case the work would be "wasted")

Running Prime95 all-out on this same machine, I can complete PRP 4 tests in about 5.5 days, so about 0.73 tests per day.

So, by sieving instead of running the client on this machine, I am reducing SoB's queue by 0.27 additional tests/day, and the other 1.5 factors/day go to benefit PSP. (Bonus for them!)

Does this make sense, or am I miscalculating or misunderstanding something somewhere? (I am new to sieving and trying to get a better understanding of what I am actually accomplishing here )