LLR - new version

Cruelty · 2006-04-12, 13:15

There was no announcement, however version 3.7.0 is out and it seems that there are some speed improvements over 3.6.2, at least for P4...

Flatlander · 2006-04-12, 13:54

Quote:

Originally Posted by Cruelty

There was no announcement, however version 3.7.0 is out and it seems that there are some speed improvements over 3.6.2, at least for P4...

... but, as expected, not on my Athlon XP. LLR3.5 is still fastest.

Cruelty · 2006-04-12, 13:55

Quote:

Originally Posted by Flatlander

... but, as expected, not on my Athlon XP. LLR3.5 is still fastest.

And is still BETA

Flatlander · 2006-04-12, 14:02

Have there been any problems reported with 3.5?

paulunderwood · 2006-04-12, 14:44

From the Readme.txt that come with LLR 3.7.0:

Quote:

The new Gwnums allow non zero padded Irrational Base FFTs with k's up to
20 bits, and zero padded ones with k's up to 48 bits, so this version is
a major enhancement of the previous IBDWT version of LLR, which uses the
Colin Percival's method, and is really fast only for k's up to 9 bits
(However, I will continue to support it some time).

So for larger "k" there should be a huge speed up

Flatlander · 2006-04-12, 21:47

The same comment is also in the 3.5 readme.txt.

VBCurtis · 2006-04-13, 19:03

I was disappointed to not find any speed boost with k=399239 or k=3803443215 on a P4 CPU running LLR 3.7. 399239 is under 20 bits, so I had hope. In fact, the bigger k (just under 32 bits) is the same speed as the 19-bit k.

-Curtis

Cruelty · 2006-05-16, 13:53

There is new 3.7.1 version, however it is not mentioned in readme what the change from 3.7.0 actually is...
I haven't tested it yet either - perhaps someone else already did it?

edorajh · 2006-05-16, 15:00

Jean Penné did describe changes made in 3.7.1. version in the PrimeNumbers yahoo group....

Quote:

This update only concern the Gaussian Mersenne prime tests.
I implemented the random shift of the units bit position for the
Gaussian Mersenne norm primality testing, as George Woltman did for
the LL test of Mersenne numbers. It works perfectly well, and it is
important for the double check works. On the other hand, none of the
features have been changed, so I did not changed the version number of
the program...

Link...

http://groups.yahoo.com/group/primen.../message/17999

2006-04-12, 13:15	#1
Cruelty May 2005 2³×7×29 Posts	LLR - new version There was no announcement, however version 3.7.0 is out and it seems that there are some speed improvements over 3.6.2, at least for P4...

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LLR Version 3.8.5 is available!	Jean Penné	Software	11	2011-02-20 18:22
LLR Version 3.8.0 is now available!	Jean Penné	Software	22	2010-04-28 07:45
Which version for P-III's?	richs	Software	41	2009-01-07 14:40
Version 24.14	Prime95	Software	13	2006-02-15 16:32
Which LLR version to use...	Cruelty	Riesel Prime Search	1	2005-11-10 15:17

2006-04-12, 14:02	#4
Flatlander I quite division it "Chris" Feb 2005 England 31·67 Posts	Have there been any problems reported with 3.5?

2006-04-12, 21:47	#6
Flatlander I quite division it "Chris" Feb 2005 England 31·67 Posts	The same comment is also in the 3.5 readme.txt.

2006-04-13, 19:03	#7
VBCurtis "Curtis" Feb 2005 Riverside, CA 2·2,339 Posts	I was disappointed to not find any speed boost with k=399239 or k=3803443215 on a P4 CPU running LLR 3.7. 399239 is under 20 bits, so I had hope. In fact, the bigger k (just under 32 bits) is the same speed as the 19-bit k. -Curtis

2006-05-16, 13:53	#8
Cruelty May 2005 2³·7·29 Posts	There is new 3.7.1 version, however it is not mentioned in readme what the change from 3.7.0 actually is... I haven't tested it yet either - perhaps someone else already did it?