LLR Version 3.8.11 released - Page 2

Jean Penné · 2013-12-25, 06:47

Quote:

Originally Posted by pinhodecarlos

Complete test on 4 numbers, went back to previous version.

Please, would you show what type of candidates are you testing?
Thank you by advance,
Jean

pinhodecarlos · 2013-12-25, 09:01

In bold with LLR 3.8.11. The rest with LLR 3.8.9.

5*2^3788690-1 is not prime. LLR Res64: E8A5BCE8FD57EAC6 Time : 4465.550 sec.
5*2^3788730-1 is not prime. LLR Res64: 3CCE7561AC6103BD Time : 4431.953 sec.
5*2^3788764-1 is not prime. LLR Res64: B20C45C63C96F395 Time : 4447.727 sec.
5*2^3788810-1 is not prime. LLR Res64: D9A8F0723DB0845C Time : 4416.190 sec.
5*2^3788824-1 is not prime. LLR Res64: 7F373226C125A8C8 Time : 4402.961 sec.
5*2^3788920-1 is not prime. LLR Res64: DC4A38A6D1214BD0 Time : 4574.177 sec.
5*2^3788958-1 is not prime. LLR Res64: 50D1F4F64FA56F71 Time : 4551.266 sec.
5*2^3788988-1 is not prime. LLR Res64: BE1168AA07206781 Time : 4432.525 sec.
5*2^3789004-1 is not prime. LLR Res64: 1888B618CC568562 Time : 4446.892 sec.
5*2^3789028-1 is not prime. LLR Res64: A739CDC00DA238CD Time : 4395.631 sec.
5*2^3789060-1 is not prime. LLR Res64: 21F99E8373C8109D Time : 4406.748 sec.
5*2^3789084-1 is not prime. LLR Res64: 2D968EB15DE6EB0D Time : 4417.853 sec.

paulunderwood · 2013-12-25, 21:54

Quote:

Originally Posted by Jean Penné

It is not mine... But I think the best is to ask that to him!

George, please write the code for efficient b^n-b^m-1 mod reduction.

Prime95 · 2013-12-25, 22:32

Quote:

Originally Posted by paulunderwood

George, please write the code for efficient b^n-b^m-1 mod reduction.

Sounds painful.

Jean Penné · 2013-12-26, 07:19

Quote:

Originally Posted by pinhodecarlos

In bold with LLR 3.8.11. The rest with LLR 3.8.9.

5*2^3788690-1 is not prime. LLR Res64: E8A5BCE8FD57EAC6 Time : 4465.550 sec.
5*2^3788730-1 is not prime. LLR Res64: 3CCE7561AC6103BD Time : 4431.953 sec.
5*2^3788764-1 is not prime. LLR Res64: B20C45C63C96F395 Time : 4447.727 sec.
5*2^3788810-1 is not prime. LLR Res64: D9A8F0723DB0845C Time : 4416.190 sec.
5*2^3788824-1 is not prime. LLR Res64: 7F373226C125A8C8 Time : 4402.961 sec.
5*2^3788920-1 is not prime. LLR Res64: DC4A38A6D1214BD0 Time : 4574.177 sec.
5*2^3788958-1 is not prime. LLR Res64: 50D1F4F64FA56F71 Time : 4551.266 sec.
5*2^3788988-1 is not prime. LLR Res64: BE1168AA07206781 Time : 4432.525 sec.
5*2^3789004-1 is not prime. LLR Res64: 1888B618CC568562 Time : 4446.892 sec.
5*2^3789028-1 is not prime. LLR Res64: A739CDC00DA238CD Time : 4395.631 sec.
5*2^3789060-1 is not prime. LLR Res64: 21F99E8373C8109D Time : 4406.748 sec.
5*2^3789084-1 is not prime. LLR Res64: 2D968EB15DE6EB0D Time : 4417.853 sec.

Here are my results on an AMD 4 cores and Linux 64bit llr :
With llr64 V3.8.9 :

[Wed Dec 25 15:25:19 2013]
Using AMD K10 FFT length 224K, Pass1=896, Pass2=256
V1 = 4 ; Computing U0...done.
[Wed Dec 25 18:25:56 2013]
5*2^3788920-1 is not prime. LLR Res64: DC4A38A6D1214BD0 Time : 10837.484 sec.
Using AMD K10 FFT length 224K, Pass1=896, Pass2=256
V1 = 4 ; Computing U0...done.
[Wed Dec 25 21:25:41 2013]
5*2^3788958-1 is not prime. LLR Res64: 50D1F4F64FA56F71 Time : 10784.490 sec.

And with llr64 3.8.11 :

[Wed Dec 25 15:25:23 2013]
Using AMD K10 FFT length 224K, Pass1=896, Pass2=256
V1 = 4 ; Computing U0...done.
[Wed Dec 25 18:25:32 2013]
5*2^3788920-1 is not prime. LLR Res64: DC4A38A6D1214BD0 Time : 10809.519 sec.
Using AMD K10 FFT length 224K, Pass1=896, Pass2=256
V1 = 4 ; Computing U0...done.
[Wed Dec 25 21:24:55 2013]
5*2^3788958-1 is not prime. LLR Res64: 50D1F4F64FA56F71 Time : 10762.600 sec.
As you can see, V3.8.11 is (very slightly) faster...
Regards,
Jean

pinhodecarlos · 2013-12-26, 08:42

I left the new version running through the night and it was as you said very slightly fast although in the first time I used it it was slightly slow. I don't understand because there was nothing on the background running beside two instances of LLR.

Anyway, thank you for the new version.

Carlos

KEP · 2013-12-26, 11:22

Well I can tell you all this much, that both using version 3.8.9 and 3.8.11 the timings for the same FFT length and the same k and a slightly different n value can be +/- 5% in total testing time, compared to the previous test completed, with the same FFT length and same k and a slightly smaller n. I suspect it to have something to do with my RAM not being fast enough to cope with the AVX processing, so more data goes into the RAM than the RAM can process. Anyway the difference appears to be the same on both versions, so I'll stick with 3.8.11

Take care

KEP

Jean Penné · 2013-12-26, 16:58

Thanks to Iain Bethune who built them, the two Mac/Intel binaries are also now available.
Regards,
Jean

pepi37 · 2013-12-26, 17:26

Who can build LLR with GUI but 64 bit?

pinhodecarlos · 2013-12-27, 13:33

Quote:

Originally Posted by pepi37

Who can build LLR with GUI but 64 bit?

Once I saw the GUI version on LLR in here. Maybe you can contact the administrator.

http://www.bc-team.org/downloads.php?cat=7

pepi37 · 2013-12-27, 14:36

That was 32 bit LLR -GUI :(

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLR Version 3.8.20 released	Jean Penné	Software	30	2018-08-13 20:00
LLR Version 3.8.19 released	Jean Penné	Software	11	2017-02-23 08:52
LLR Version 3.8.15 released	Jean Penné	Software	28	2015-08-04 04:51
LLR Version 3.8.9 released	Jean Penné	Software	37	2013-10-31 08:45
llr 3.8.2 released as dev-version	opyrt	Prime Sierpinski Project	11	2010-11-18 18:24

2013-12-25, 09:01	#13
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 2·5·11·47 Posts	In bold with LLR 3.8.11. The rest with LLR 3.8.9. 52^3788690-1 is not prime. LLR Res64: E8A5BCE8FD57EAC6 Time : 4465.550 sec. 52^3788730-1 is not prime. LLR Res64: 3CCE7561AC6103BD Time : 4431.953 sec. 52^3788764-1 is not prime. LLR Res64: B20C45C63C96F395 Time : 4447.727 sec. 52^3788810-1 is not prime. LLR Res64: D9A8F0723DB0845C Time : 4416.190 sec. 52^3788824-1 is not prime. LLR Res64: 7F373226C125A8C8 Time : 4402.961 sec. 52^3788920-1 is not prime. LLR Res64: DC4A38A6D1214BD0 Time : 4574.177 sec. 52^3788958-1 is not prime. LLR Res64: 50D1F4F64FA56F71 Time : 4551.266 sec. 52^3788988-1 is not prime. LLR Res64: BE1168AA07206781 Time : 4432.525 sec. 52^3789004-1 is not prime. LLR Res64: 1888B618CC568562 Time : 4446.892 sec. 52^3789028-1 is not prime. LLR Res64: A739CDC00DA238CD Time : 4395.631 sec. 52^3789060-1 is not prime. LLR Res64: 21F99E8373C8109D Time : 4406.748 sec. 52^3789084-1 is not prime. LLR Res64: 2D968EB15DE6EB0D Time : 4417.853 sec.

2013-12-26, 08:42	#17
pinhodecarlos "Carlos Pinho" Oct 2011 Milton Keynes, UK 5170₁₀ Posts	I left the new version running through the night and it was as you said very slightly fast although in the first time I used it it was slightly slow. I don't understand because there was nothing on the background running beside two instances of LLR. Anyway, thank you for the new version. Carlos

2013-12-26, 11:22	#18
KEP Quasi Admin Thing May 2005 3EB₁₆ Posts	Well I can tell you all this much, that both using version 3.8.9 and 3.8.11 the timings for the same FFT length and the same k and a slightly different n value can be +/- 5% in total testing time, compared to the previous test completed, with the same FFT length and same k and a slightly smaller n. I suspect it to have something to do with my RAM not being fast enough to cope with the AVX processing, so more data goes into the RAM than the RAM can process. Anyway the difference appears to be the same on both versions, so I'll stick with 3.8.11 Take care KEP

2013-12-26, 16:58	#19
Jean Penné May 2004 FRANCE 1160₈ Posts	Thanks to Iain Bethune who built them, the two Mac/Intel binaries are also now available. Regards, Jean

2013-12-26, 17:26	#20
pepi37 Dec 2011 After 1.58M nines:) 11010100011₂ Posts	Who can build LLR with GUI but 64 bit?

2013-12-27, 14:36	#22
pepi37 Dec 2011 After 1.58M nines:) 11010100011₂ Posts	That was 32 bit LLR -GUI :(