yafu-1.32.1

[bsquared]

Now available, to fix a couple of bugs.

link.

[BudgieJane]

Thank you.

[Kosmaj]

Can you kindly provide the linux statically linked version.
I have troubles to locate the right gmp library.

Thanks!

[Kosmaj]

You can ignore my request above as I managed to build my own gmp library.

[bsquared]

Sorry, didn't see your request in time.

[WraithX]

I just tried out 1.32.1 and really like the ETA on the ecm curves. However, when the line gets "too long" it starts scrolling down the screen like so:

rho: x^2 + 1, starting 1000 iterations on C228
pm1: starting B1 = 150K, B2 = gmp-ecm default on C228
ecm: 30/30 curves on C218 input, at B1 = 2K, B2 = gmp-ecm default
ecm: 22/74 curves on C218 input, at B1 = 11K, B2 = gmp-ecm default
ecm: 51/51 curves on C201 input, at B1 = 11K, B2 = gmp-ecm default
ecm: 10/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 209 sec
ecm: 11/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 208 sec
ecm: 12/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 207 sec
ecm: 13/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 206 sec
ecm: 14/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 205 sec

Is there a way to shorten the line so that it doesn't scroll like this? Perhaps a change like so:

FROM:
ecm: 14/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 205 sec
TO:
ecm: 14/214 curves on C201 input, at B1=50K, B2=default ETA: 205 sec

And/or maybe a message before the ecm starts to let people know that "default" means gmp-ecm default, like so:

rho: x^2 + 1, starting 1000 iterations on C228
pm1: starting B1 = 150K, B2 = gmp-ecm default on C228
ecm: using gmp-ecm default B2 for each B1
ecm: 30/30 curves on C218 input, at B1=2K, B2=default
ecm: 22/74 curves on C218 input, at B1=11K, B2=default
ecm: 51/51 curves on C201 input, at B1=11K, B2=default
ecm: 10/214 curves on C201 input, at B1=50K, B2=default ETA: 209 sec

I know this is really just a cosmetic thing, but these changes should give us enough room that we don't get the scrolling except with "large" ecm curves. What do you and/or others think?

[kar_bon]

Originally Posted by WraithX

I know this is really just a cosmetic thing, but these changes should give us enough room that we don't get the scrolling except with "large" ecm curves. What do you and/or others think?

Using WIN change the screen width of the console window to 100 or 120.
Also setting the screenbuffer to a higher value (scrolling more output) is recommendable.

[richs]

How do I get YAFU to perform more sieiving rather than attempting to filter on a restart? Here is the last few lines of Msieve output from YAFU:

Mon Aug 27 12:51:40 2012 reduce to 3908966 relation sets and 3861521 unique ideals
Mon Aug 27 12:51:40 2012 ignored 9 oversize relation sets
Mon Aug 27 12:51:40 2012 commencing full merge
Mon Aug 27 12:55:08 2012 memory use: 122.0 MB
Mon Aug 27 12:55:08 2012 found 70708 cycles, need 805580
Mon Aug 27 12:55:08 2012 too few cycles, matrix probably cannot build

YAFU then abruptly shuts down.

Thanks!

[Dubslow]

What command are you using exactly? With nfs() -R (and no other NFS options) it should go right back to sieving after trying and failing at filtering. If you want it to do just filtering, you can do nfs() -R -ns and it will do at least one more round of sieving, but won't start filtering or LA.

[bsquared]

Originally Posted by Dubslow

What command are you using exactly? With nfs() -R (and no other NFS options) it should go right back to sieving after trying and failing at filtering. If you want it to do just filtering, you can do nfs() -R -ns and it will do at least one more round of sieving, but won't start filtering or LA.

I think you will have to do the latter. IIRC, there is an exit() buried in the msieve library for the "matrix probably cannot build" condition, in which case you have to force more sieving before trying to filter.

[bsquared]

Originally Posted by WraithX

I just tried out 1.32.1 and really like the ETA on the ecm curves. However, when the line gets "too long" it starts scrolling down the screen like so:

rho: x^2 + 1, starting 1000 iterations on C228
pm1: starting B1 = 150K, B2 = gmp-ecm default on C228
ecm: 30/30 curves on C218 input, at B1 = 2K, B2 = gmp-ecm default
ecm: 22/74 curves on C218 input, at B1 = 11K, B2 = gmp-ecm default
ecm: 51/51 curves on C201 input, at B1 = 11K, B2 = gmp-ecm default
ecm: 10/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 209 sec
ecm: 11/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 208 sec
ecm: 12/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 207 sec
ecm: 13/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 206 sec
ecm: 14/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 205 sec

Is there a way to shorten the line so that it doesn't scroll like this? Perhaps a change like so:

FROM:
ecm: 14/214 curves on C201 input, at B1 = 50K, B2 = gmp-ecm default ETA: 205 sec
TO:
ecm: 14/214 curves on C201 input, at B1=50K, B2=default ETA: 205 sec

And/or maybe a message before the ecm starts to let people know that "default" means gmp-ecm default, like so:

rho: x^2 + 1, starting 1000 iterations on C228
pm1: starting B1 = 150K, B2 = gmp-ecm default on C228
ecm: using gmp-ecm default B2 for each B1
ecm: 30/30 curves on C218 input, at B1=2K, B2=default
ecm: 22/74 curves on C218 input, at B1=11K, B2=default
ecm: 51/51 curves on C201 input, at B1=11K, B2=default
ecm: 10/214 curves on C201 input, at B1=50K, B2=default ETA: 209 sec

I know this is really just a cosmetic thing, but these changes should give us enough room that we don't get the scrolling except with "large" ecm curves. What do you and/or others think?

I'm ok with that, although I guess I'd slightly prefer getting rid of the words "input" and "at" and keep gmp-ecm where it is. As well as get rid of the spaces.

I think you still have commit access to the SVN repository; mind making these changes yourself?