LLRnet servers for CRUS

[em99010pepe]

The candidates are running on 4 cores. For now it seems that the standard client is 21 % faster than llrnet based on llr 3.5.0. Tomorrow I'll post the timings.

[gd_barnes]

Sounds great Carlos. Thanks for help cleaning up our 'mess'. 4 cores should knock it out in < 1 week.

David, I just thought of a more likely reason why your web page shows one more k/n pair remaining than was in your list. I think it may be because it is counting the header line as a k/n pair, i.e. 100000000:P:1:2:257, which it should not be.

So when having your script or program compute the pairs remaining for display on the web page, it just needs to look at the # of lines remaining in the file and subtract 1.


Gary

[IronBits]

One is the loneliest number... Sounds good.
I will use find /c " " knpairs.txt to get an accurate line count, thus pairs remaining.
There is no space in the first line, so it gets skipped

[em99010pepe]

Originally Posted by em99010pepe

The candidates are running on 4 cores. For now it seems that the standard client is 21 % faster than llrnet based on llr 3.5.0. Tomorrow I'll post the timings.

1000 seconds faster per candidate.

llrnet:

64494*2^1937858+1 is not prime. RES64: 069AD8EFB4AA8A0A Time: 7396.367 sec.
64494*2^1938146+1 is not prime. RES64: 560645A6CB1D4FD7 Time: 7139.231 sec.
64494*2^1938506+1 is not prime. RES64: FF5B25D71A310702 Time: 7514.914 sec.

llr 3.7.1c

64494*2^1955306+1 is not prime. Proth RES64: D1E445EF4A40DF51 Time : 6389.697 sec.
64494*2^1955426+1 is not prime. Proth RES64: 57365A75AE2CB681 Time : 6380.260 sec.
64494*2^1955954+1 is not prime. Proth RES64: A22BFC12F6A94C7A Time : 6381.957 sec.

[gd_barnes]

Originally Posted by em99010pepe

1000 seconds faster per candidate.

llrnet:



llr 3.7.1c

Most impressive speedy machine! It looks like your total CPU time will be just under the lowest estimate that I gave you in the PM. 6390 * 257 = 1.661M secs. = 19.23 CPU days or < 5 days on your 4 cores! Sweet!

Is this your own personal quad?


Gary

[em99010pepe]

Originally Posted by gd_barnes

Is this your own personal quad?

Yes it is but the most impressive is the difference in timings between the two versions. Of course those times were a comparison between the same core. As you already know as the FFT increases (or the number size) the 4 cores can't handle the same output speed due to memory bottleneck. So the average timings are for core 0 of 6470 secs per candidate, core 1 of 6390 secs, core 2 of 6440 secs and core 3 of 6470 seconds.

It would be very funny to find a prime but I don't think that will happen.

[em99010pepe]

Carlos is done with the remaining range from server #2.