Registered Memory (-- edit: --) is Not the Problem

[ewmayer]

Perhaps CADO-NFS, much like the GIMPS p-1 clients, is querying the system for available RAM and using some large fraction of that for efficiency reasons. Does the program have the equivalent of a -maxalloc option?

[charybdis]

Originally Posted by ewmayer

Perhaps CADO-NFS, much like the GIMPS p-1 clients, is querying the system for available RAM and using some large fraction of that for efficiency reasons. Does the program have the equivalent of a -maxalloc option?

I'm fairly sure it doesn't do this. There are options that can reduce memory usage while sacrificing quite a lot of speed, but the client doesn't invoke these automatically if there isn't enough memory, it just spits out an error. With the parameters that Ed is using, the client should require ~4GB and there isn't any way to speed it up by using more memory than that.

[EdH]

Originally Posted by charybdis

Absolutely. This sounds like a significant bug and the sooner the CADO developers find out about it, the better.

Unfortunately, I seem to have been dropped or somehow off the cado-nfs list, or some such event. I'm not getting any emails any longer. The first thing I would need to do is reestablish my membership.

OTOH, if no one else is experiencing this. . .

Perhaps tomorrow I'll do some more testing with various revisions and see if I can pin down a change and then chase down my cado-nfs email problem.

[charybdis]

Originally Posted by EdH

Unfortunately, I seem to have been dropped or somehow off the cado-nfs list, or some such event. I'm not getting any emails any longer. The first thing I would need to do is reestablish my membership.

The address changed a few months ago, you can sign up again from here.

[EdH]

Originally Posted by charybdis

The address changed a few months ago, you can sign up again from here.

I appear to still be a member, so I attempted to send a message to the list. I'll see if it makes it through and what may result.