Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.programming.threads > #2780

A new algorithm of a Parallel Conjugate gradient solver library

Show all headers | View raw

Hello,


I have come up with a new algorithm of my Parallel Conjugate gradient 
solver library, now it has become cache-aware, but you have to notice 
that this new cache-aware algorithm is more efficient on multicores, 
since i have benchmarked it against my previous algorithm and it has 
given a scalability of 5X on a Quadcore over the single thread of my 
previous algorithm , that's a really big improvement ! and now i have 
made it NUMA-aware so that the memory transfers from the memory to the 
CPU have become parallelized so that it becomes really scalable on NUMA 
architecture, and this is also a big improvement ! but to make it 
NUMA-aware you have just to allocate your arrays of your matrix in 
different NUMA nodes using threads, and that's easy to do.

Hope you will be happy with my new algorithm and parallel library,
frankly my new algorithm have to be written as a PhD paper
or such.. but i have let it like that at this moment.

You can download my new Parallel Conjugate gradient solver library 
version 1.1 from:

https://sites.google.com/site/aminer68/parallel-implementation-of-conjugate-gradient-linear-system-solver



Thank you,
Amine Moulay Ramdane.

Back to comp.programming.threads | Previous | Next — Next in thread | Find similar | Unroll thread

Thread

A new algorithm of a Parallel Conjugate gradient solver library Ramine <ramine@1.1> - 2014-12-16 15:01 -0800
  Re: A new algorithm of a Parallel Conjugate gradient solver library Ramine <ramine@1.1> - 2014-12-18 15:35 -0800

csiph-web