Changes between Version 27 and Version 28 of Contribute

Timestamp:

06/27/08 11:53:03 (5 years ago)

Author:

nguyen

Comment:

--

Contribute

@@ -18 +18 @@
 The following are ideas you can work on:
  * '''Employ graphical processor units (GPU) to perform core matrix operations''' in MRMC. This  theoretically improves performance by many-fold. There is a related contest for creating efficient sparse-matrix multiplication algorithms for GPU's. Additionally, there is also a library for parallel GPU operations. See respectively: http://scriptroute.cs.umd.edu/gpucompete/ and http://www.gpgpu.org/developer/cudpp/
- * '''Improve the sparse matrix data-structure ''' in MRMC, in order to simplify the access to predecessor probabilities. Additionally, the performance and memory consumption can be improved. For ideas on this topic see [attachment:proposals-2008-06.pdf David N. Jansen's proposal].
- * '''Decrease memory consumption of the lumping algorithm''' by introducing a new data structure. More details can be found in [attachment:proposals-2008-06.pdf David N. Jansen's proposal]. 
- * '''Achieve more robust code''' in MRMC. This point may consist of restricting to a standard dialect like ANSI C/ISO C90, a naming convention for functions etc., improved error checking and a fixed comment style. Additionally tools like [http://www.splint.org/ Splint] can be used to assure correct functionality. For details see [attachment:proposals-2008-06.pdf David N. Jansen's proposal].
- * ...
+ * Please mail us your ideas!