Discussion related to cado-nfs

[paul zimmermann <Paul.Zimmermann@inria.fr>]

       Hello,

> Hi, this device have 1GB RAM, and I don't have a c105.sqrt.log yet, but I 
> have a file named
> c105.sqrt.stdout.1 this is the output:
> 
> relations in 0.1s -- 1.1 MB/s -- 13270.6 rels/s # Read 2048 relations in 
> 0.1s -- 2.2 MB/s --
> 25968.1 rels/s # Read 4096 relations in 0.1s -- 4.3 MB/s -- 50214.6 rels/s 
> # Read 8192 relations in
> 0.2s -- 4.0 MB/s -- 46283.5 rels/s # Read 16384 relations in 0.3s -- 5.2 
> MB/s -- 60356.4 rels/s#
> Read 32768 relations in 0.4s -- 6.8 MB/s -- 79373.1 rels/s# Read 65536 
> relations in 0.9s -- 6.1
> MB/s -- 70366.4 rels/s# Read 131072 relations in 2.6s -- 4.3 MB/s -- 
> 50073.5 rels$# Read 262144
> relations in 6.6s -- 3.4 MB/s -- 39423.8 rels$# Read 524288 relations in 
> 15.1s -- 3.0 MB/s --
> 34641.6 rel$# Done: Read 876946 relations in 27.1s -- 2.8 MB/s -- 32342$
> ------
> Thanks

this seems fine. In the development version my c105.sqrt.stdout.1 file ends 
similarly:

# Done: Read 936288 relations in 2.5s -- 33.0 MB/s -- 374986.3 rels/s

and the c105.sqrt.stderr.1 file ends with:

Total usage: time 4s (cpu), 3s (wct) ; memory 97M, peak 177M

Then you should have two other files, c105.sqrt.stderr.2 and 
c105.sqrt.stdout.2,
the latter containing the factors found. On my side c105.sqrt.stderr.2 ends 
with:

Total usage: time 196s (cpu), 28s (wct) ; memory 561M, peak 1186M

thus it might exceed the memory of your computer. To reduce the memory usage, 
add the following
on the cado-nfs.py command line (or in your parameter file):

tasks.sqrt.threads=1

This will process only one dependency at a time. Then I get:

Total usage: time 24s (cpu), 24s (wct) ; memory 26M, peak 96M

Best regards,
Paul Zimmermann