Improved params files for CADO

[firejuggler]

first try with cado, for unconnected's c203

13966734844620395430991487348745167746919939062454191641301594834859712167254112852425351951454323376159536913937113093639020912129186373021867801289559387189650881337161223227226030924322175928067799243

editing params.c200 tp

tasks.polyselect.degree = 5 
 
tasks.polyselect.P = 10000 
tasks.polyselect.admin= 5e6 
tasks.polyselect.admax = 7e6 
tasks.polyselect.adrange = 1e3 
tasks.polyselect.incr = 60 
tasks.polyselect.nq = 3125 
tasks.polyselect.nrkeep = 200

Does that seem reasonnable?

[swellman]

See my answer in the original thread.

[EdH]

In preliminary testing, strategy 2 at smaller composites appears to make no discernible difference in memory use, relations returned or time spent. This may allow for less work in rewriting my client scripts.

My concern is that I'm currently running two clients on machines with 8GB. But my solution looks like running one of those with strategy 2 and the other with strategy 0 on a permanent basis. I "think" that will fit, but further testing is needed, of course.

[EdH]

Well, that didn't work! The clients I tried to add strategy 2 to, crashed trying to run polyselect. This is going to take some more complicated scripts.

[charybdis]

Originally Posted by EdH

Well, that didn't work! The clients I tried to add strategy 2 to, crashed trying to run polyselect. This is going to take some more complicated scripts.

Oh, of course adjust-strategy isn't a valid option for polyselect so it will crash

How about adding tasks.sieve.run = false at the server level to get only polyselect to run, and then for sieving remove it and run clients with --override adjust-strategy 2?

[EdH]

Originally Posted by charybdis

Oh, of course adjust-strategy isn't a valid option for polyselect so it will crash

How about adding tasks.sieve.run = false at the server level to get only polyselect to run, and then for sieving remove it and run clients with --override adjust-strategy 2?

The trouble is that ...run=false doesn't send a 410 to tell the clients to stop, so all commands have to be supplied by the server and I have many clients that aren't going to handle strategy 2. I do already have separate sets of clients that begin for polyselect and las, but the polyselect are never told to stop, so they just keep running.

i will have to think about this a bit. . .

[charybdis]

Originally Posted by EdH

The trouble is that ...run=false doesn't send a 410 to tell the clients to stop

That feels like an omission/error on the part of the developers. I bet if you dropped them a line on the mailing list you would get a quick fix.

[EdH]

Originally Posted by charybdis

That feels like an omission/error on the part of the developers. I bet if you dropped them a line on the mailing list you would get a quick fix.

I've thought about doing that for quite some time, but I'm lazy and I figured if nobody else seems to need it, maybe it's just me. I think it would benefit those who use multiple machines for poly searching only, but maybe everyone is using single machines. Perhaps I'll send off an email. I'll think about it a bit.

Thanks!

[charybdis]

You don't know how many people have asked themselves the same question and thought the same thing

I've run polyselect on many machines before. We're certainly not the only ones. It's probably a 5 minute fix for the developers; I'm sure they won't mind.

[EdH]

Originally Posted by charybdis

You don't know how many people have asked themselves the same question and thought the same thing

I've run polyselect on many machines before. We're certainly not the only ones. It's probably a 5 minute fix for the developers; I'm sure they won't mind.

Well, I sent a message, but you'll see I went and messed it up. I grabbed a digest and replied to it, forgetting to remove the prior content.

[thornado123]

Originally Posted by VBCurtis

Poly select params posted for C190 and C195. Sieve params are CADO default, and are not optimal.
These params take about 1/3rd the CADO-default time; if you are not impressed with the E-score from your run, double P and run again.
Thanks to user hnoey for extensive testing on a series of C193s to develop these parameters.

Hi

Can I use the C195 parameter file when factoring a 193 cipher number (or should I search for a specific parameter file?)

I am new to this forum and I want to factor a 640-bit integer to prove a point about its security after reading about FAAS (https://github.com/eniac/faas).

I am making my own version of FAAS because theirs aren't functional right now :(

Best regards