Cunningham ECM efforts

[yoyo]

I sent it also to Paul Zimmerman for his ecm record page https://members.loria.fr/PZimmermann...ds/ecmnet.html
yoyo

[R.D. Silverman]

Originally Posted by swellman

Yoyo found a p67 in 2,2222L, thereby fully factoring it!

Saving a big NFS effort! It is also nice to see that YoYo's efforts have not been futile.

Someone should send it to Paul Zimmermann at https://members.loria.fr/PZimmermann...ds/ecmnet.html

[frmky]

Originally Posted by swellman

Yoyo found a p67 in 2,2222L, thereby fully factoring it!

Yay! I've crossed that one off my list.

[pinhodecarlos]

And 2,2194L was done a few days ago.

[swellman]

Originally Posted by R.D. Silverman

Saving a big NFS effort! It is also nice to see that YoYo's efforts have not been futile.

Someone should send it to Paul Zimmermann at https://members.loria.fr/PZimmermann...ds/ecmnet.html

Yoyo took care of this. I sent a note to Sam Wagstaff as well.

In other news, I’ve enqueued three more from the Gang: 2,2246L, 2,1124+ and 2,1123+. All are SNFS difficulty 338. Hoping all numbers enqueued so far complete ECM by the end of April. Time will tell.

The rest of the Gang are SNFS 340-361. I’m holding off placing them in queue for now, in case Greg wants to change direction. Of course the entire Gang will be processed by Yoyo eventually.

[R.D. Silverman]

Originally Posted by swellman

Yoyo took care of this. I sent a note to Sam Wagstaff as well.

In other news, I’ve enqueued three more from the Gang: 2,2246L, 2,1124+ and 2,1123+. All are SNFS difficulty 338. Hoping all numbers enqueued so far complete ECM by the end of April. Time will tell.

The rest of the Gang are SNFS 340-361. I’m holding off placing them in queue for now, in case Greg wants to change direction. Of course the entire Gang will be processed by Yoyo eventually.

2,2206L??

[swellman]

Originally Posted by R.D. Silverman

2,2206L??

I believe 2,2206L is next on the big siever.

[Andrew Usher]

In other words, it has had enough ECM?

[swellman]

Originally Posted by Andrew Usher

In other words, it has had enough ECM?

An open question for numbers of this high difficulty, but it has already seen a LOT of ECM. If Greg chooses to skip it then I’ll throw it into Yoyo’s queue.

I’m more curious about what portion of NFS will eventually fail as we climb the difficulty ladder. I believe Greg already encountered issues with filtering but that was worked around. Building a proper matrix? LA?

At current pace, we should finish ECM on the 1987 list of base-2s by late 2024 or early 2025. How many survivors by then? 15-20 is my non-expert guess, though 0 would be wonderful.

[charybdis]

Originally Posted by swellman

I’m more curious about what portion of NFS will eventually fail as we climb the difficulty ladder. I believe Greg already encountered issues with filtering but that was worked around. Building a proper matrix? LA?

Might just be that collecting enough relations to form a matrix will eventually take impractically long. SNFS-330+ probably ought to use lims over 500M, but they are capped at 250M due to NFS@Home users' memory restrictions. Increasing lpbs to 35 has helped mitigate this, but currently they can't go any higher due to the issues with msieve as the number of relations approaches 2^32. Even if a 17e siever existed, it wouldn't help NFS@Home because of the much higher memory it would require.

[frmky]

Originally Posted by charybdis

Might just be that collecting enough relations to form a matrix will eventually take impractically long. SNFS-330+ probably ought to use lims over 500M, but they are capped at 250M due to NFS@Home users' memory restrictions. Increasing lpbs to 35 has helped mitigate this, but currently they can't go any higher due to the issues with msieve as the number of relations approaches 2^32. Even if a 17e siever existed, it wouldn't help NFS@Home because of the much higher memory it would require.

2,2194L and M are using lims of 300M (pushing the memory limit to the edge of 1GB/core but also crossing 2^28 for a small runtime hit) and lpbs of 35/36 (with prior duplicate and singleton removal to get around limitations of msieve).

Being able to use CADO for filtering then msieve for GPU LA would be great, but that will require work that I currently don't have time to do.