Re: [julia-users] Canned out of core linalg chunking algorithms?

2015-01-01 Thread Viral Shah
For now, we want to complete our Darray implementation and integration with MPI for parallel usage. Given how easy it is to get large numbers of CPUs, we want to address that use case first. The general out of core framework is the OS and swap - but it is not always the best. There are no good

Re: [julia-users] Canned out of core linalg chunking algorithms?

2014-12-31 Thread Lampkld
I would be more than happy to help, but that doesn't sound like a project within the domain of Julia and CS knowledge at the moment. If someone would set up a project framework, I could probably be better placed to contribute. On Wednesday, December 31, 2014 2:47:21 PM UTC-5, Jiahao Chen wro

Re: [julia-users] Canned out of core linalg chunking algorithms?

2014-12-31 Thread Jiahao Chen
We would very much like to have out of core functionality, but we don't have any active project to work on it. If you could spare the time, this would be a very valuable feature to work toward having in Julia. On Wed Dec 31 2014 at 10:38:44 AM Lampkld wrote: > Hello Julians, > > Are fallback ou

[julia-users] Canned out of core linalg chunking algorithms?

2014-12-31 Thread Lampkld
Hello Julians, Are fallback out of core chunking algorithms for common Darrays/shared mmap array linalg functions planned? This would allow trivial reuse of generic code. Thanks