[QUOTE=henryzz;488186]It is worth noting that unlike the ggnfs siever cado can sieve below the fb bound.[/QUOTE]
I think you mean, that cado can sieve with fb entries above the currently sieved special-q.
IMO that's slower overall, at least for small composites like c145. That's the reason I limited the fb.

Error?
 

[CODE]Info:Linear Algebra: Starting
Warning:Command: Process with PID 23973 finished with return code 255
Error:Linear Algebra: Program run on server failed with exit code 255
Error:Linear Algebra: Command line was: /home/wombatman/cado-nfs-2.3.0/build/BenMeekins-PC/linalg/bwc/bwc.pl :complete 'thr=8' 'nullspace=left' 'matrix=/tmp/cado.ui87i89u/c165.sparse.bin' 'wdir=/tmp/cado.ui87i89u/c165.bwc' 'cpubinding=/home/wombatman/cado-nfs-2.3.0/parameters/misc/cpubinding.conf' 'cantor_threshold=2048' 2> /tmp/cado.ui87i89u/c165.bwc.stderr.3
b'Missing parameters: m and/or n at linalg/bwc/bwc.pl line 360.\n[/CODE]

Any ideas what this is? Sieving and merging completed without issue. This error occurs immediately upon the start of linear algebra.

Since I can't edit the previous post, the solution was to add the appropriate parameter line to the new c165 params file (and in this case, to the snapshot as well). That's my bad. The lines I added were the following, copied from pre-existing c165 params file:

[CODE]###########################################################################
# Linear algebra
###########################################################################

tasks.linalg.bwc.interval = 1000
tasks.linalg.bwc.interleaving = 0
tasks.linalg.m = 64
tasks.linalg.n = 64

###########################################################################
# Characters
###########################################################################

tasks.linalg.characters.nchar = 50[/CODE]

CADO-NFS on RPis
 

I've been playing with using RPis as a mini-cluster with CADO-NFS. I'm running into what I think is a memory issue. The short version is this: the RPis can do polynomial selection workunits up to at least C17x (I think C177 is the highest I've done). However, they cannot handle sieving workunits above C117 (C119 and above led to the error detailed below), giving the following error repeatedly:

[CODE]Error:Lattice Sieving: Program run on localhost failed with exit code -9
Error:Lattice Sieving: Stderr output follows (stored in file /tmp/cado.znzw_dvb/c120.upload/c120_sieving_196000-197000#2.2w0rleb9.stderr0):
b'madvise failed: Invalid argument\n'[/CODE]


I know I can take this to the CADO-NFS programmers, but I figured I would check here first in case there's a simple solution. This is on an RPi Model 3B+(1GB RAM).

If you're using 2.3.0, the C120 params file uses I = 14 for no good reason. C115 file uses I = 12. If you're using the git 3.0.0, then params.c120 uses I = 12 and I dunno why you're getting crashes.

I=12 should fit comfortably in 1GB.

It's the 2.3.0 version. Thanks!

[QUOTE=wombatman;490983]I know I can take this to the CADO-NFS programmers, but I figured I would check here first in case there's a simple solution. This is on an RPi Model 3B+(1GB RAM).[/QUOTE]
Could you time RSA-100 to compare the RPi 3B+ with my (ancient) Odroid-U2 ARM board?

CADO-NFS 3.0-dev timings for RSA-100 on Odroid-U2:
[URL]http://mersenneforum.org/showpost.php?p=481269&postcount=129[/URL]


RSA-100 with Msieve-lasieve4-python script on Odroid-U2 in:
[URL]http://mersenneforum.org/showthread.php?t=23471[/URL]

Sure. Started it just now with default parameters, 4 threads.

[CODE]Info:Square Root: finished
Info:Square Root: Factors: 37975227936943673922808872755445627854565536638199 40094690950920881030683735292761468389214899724061
Info:Square Root: Total cpu/real time for sqrt: 441/124.699
Info:Polynomial Selection (size optimized): Aggregate statistics:
Info:Polynomial Selection (size optimized): potential collisions: 2890.72
Info:Polynomial Selection (size optimized): raw lognorm (nr/min/av/max/std): 2920/32.780/37.719/38.560/0.695
Info:Polynomial Selection (size optimized): optimized lognorm (nr/min/av/max/std): 2920/32.780/36.417/38.550/0.985
Info:Polynomial Selection (size optimized): Total time: 403.24
Info:Polynomial Selection (root optimized): Aggregate statistics:
Info:Polynomial Selection (root optimized): Total time: 510.88
Info:Polynomial Selection (root optimized): Rootsieve time: 507.48
Info:Generate Factor Base: Total cpu/real time for makefb: 8.42/2.29804
Info:Generate Free Relations: Total cpu/real time for freerel: 245.18/62.0037
Info:Lattice Sieving: Aggregate statistics:
Info:Lattice Sieving: Total number of relations: 2689258
Info:Lattice Sieving: Average J: 1023.74 for 88142 special-q, max bucket fill: 0.707584
Info:Lattice Sieving: Total CPU time: 40586.2s
Info:Filtering - Duplicate Removal, splitting pass: Total cpu/real time for dup1: 24.95/68.3106
Info:Filtering - Duplicate Removal, splitting pass: Aggregate statistics:
Info:Filtering - Duplicate Removal, splitting pass: CPU time for dup1: 67.6s
Info:Filtering - Duplicate Removal, removal pass: Total cpu/real time for dup2: 108.2/94.0727
Info:Filtering - Singleton removal: Total cpu/real time for purge: 117.94/84.1812
Info:Filtering - Merging: Total cpu/real time for merge: 386.29/355.652
Info:Filtering - Merging: Total cpu/real time for replay: 35.57/37.4631
Info:Linear Algebra: Total cpu/real time for bwc: 18146.5/0.00117993
Info:Linear Algebra: Aggregate statistics:
Info:Linear Algebra: Krylov: WCT time 2941.91
Info:Linear Algebra: Lingen CPU time 303.06, WCT time 84.66
Info:Linear Algebra: Mksol: WCT time 1477.02
Info:Quadratic Characters: Total cpu/real time for characters: 19.53/9.2079
Info:Square Root: Total cpu/real time for sqrt: 441/124.699
Info:HTTP server: Shutting down HTTP server
Info:Complete Factorization: Total cpu/elapsed time for entire factorization: 61033.9/16947.3
Info:root: Cleaning up computation data in /tmp/cado.1xpfmivz
37975227936943673922808872755445627854565536638199 40094690950920881030683735292761468389214899724061[/CODE]

***Note that everything except for the linear algebra was done on 2 threads.

Polyselect: 23 minutes, 42 seconds
Sieving: 40586.2 seconds (~5.5 hours real time)
Filtering: 639 secs (~10 minutes)
Linear Algebra: 2941.91+84.66+1477.02 = 4503.59 seconds (~1.25 hours) ***4 threads here
Sqrt: 125 seconds
Total (CPU/Real): 61033.9/16947.3 seconds

[QUOTE=wombatman;491045]
***Note that everything except for the linear algebra was done on 2 threads.

Polyselect: 23 minutes, 42 seconds
Sieving: 40586.2 seconds (~5.5 hours real time)
Filtering: 639 secs (~10 minutes)
Linear Algebra: 2941.91+84.66+1477.02 = 4503.59 seconds (~1.25 hours) ***4 threads here
Sqrt: 125 seconds
Total (CPU/Real): 61033.9/16947.3 seconds[/QUOTE]
Cool, thx!
I'm a bit surprised that it is not faster than the quad-A9 in the ODROID-U2?! Even after taking into account the clockspeed difference (1700 MHz for the U2 vs. 1400MHz of the RP3B+)
Were you using the CADO 3.0-dev latest git version?

No, this was with the CADO-2.3.0 release. I haven't compiled the latest version.