Discussion related to cado-nfs

[eric.jeancolas@free.fr]

Hi Paul(s)

Just a question of logic.
Issuing a grep tasks.I params* on parameters/factor, I get the following 
params.c105:tasks.I = 12
params.c110:tasks.I = 12
params.c115:tasks.I = 12
params.c120:tasks.I = 12
params.c125:tasks.I = 13
params.c130:tasks.I = 14
params.c135:tasks.I = 13
params.c140:tasks.I = 13
params.c145:tasks.I = 13
params.c150:tasks.I = 13
params.c155:tasks.I = 14
params.c160:tasks.I = 14
params.c165:tasks.I = 14
params.c170:tasks.I = 14
params.c175:tasks.I = 14

Well, about params.c130, it seems have a continuity check. Shouldn't be set 
to 13 ?

By the way, I confirm, tasks.sqrt.threads must be adapted to the hardware. 
And as this is not the longest task, you may set it to a low value.
On my 8 MB CPUs, I set it to 2.

Best regards.

Eric. 
----- Mail original -----
De: "Paul Zimmermann" <Paul.Zimmermann@inria.fr>
À: "Paul Leyland" <paul.leyland@gmail.com>
Cc: cado-nfs@inria.fr
Envoyé: Mardi 4 Juillet 2023 15:25:01
Objet: Re: [cado-nfs] Parameter documentatin.

       Hi Paul,

> Date: Thu, 29 Jun 2023 15:08:21 +0100
> From: Paul Leyland <paul.leyland@gmail.com>
> 
> A quick question: are all settable parameters documented 
> parameters/factor/params.c90 or is there some additional documentation I 
> have yet to find?

all parameters should be there. If you find a missing one, please report.

> I'm running a factorization which is at, or somewhat beyond, the present 
> limits of my system and would like to understand what else I can 
> optimize.  As just one example, it is quite important to set 
> tasks.sqrt.threads to a small value to keep memory usage within 
> reasonable limits.

Good luck!
Paul