- compatible with R version 3.0.x
- supports Open MPI in addition to MPICH
SPRINT is currently not available via CRAN, we have no access to a hardware configuration running the Solaris operating system and cannot adapt SPRINT for that platform as is required by CRAN.
- Adds pstringdistmatrix(), which computes (in this version) the Hamming distance between any strings (e.g. nucleotide bases, Next Gen Sequencing short reads).
- Makes SPRINT compliant with MPI3 (and the latest version of the mpich package)
- Provides simplified installation instructions
- Other bug fixes and updates (see User Guide)
- NOTE: This version is currently not compatible with R version 3 and higher, as we have yet to test this fully.
SPRINT version 1.0.5 is not available from CRAN or R-Forge at this time, as package submission guidelines have changed with the newly released R 3.0.x.
This version mainly adds support for Mac OSX computers. While this does not provide the high performance potential of multi-node multi-processor clusters, it allows users of Mac desktops and laptops to use their multi-core processors to best effect.
The main change in SPRINT 1.0.0 is the addition of four new functions to the SPRINT library of parallelized statistical R functions: a parallel apply function papply(), a parallel bootstrapping function pboot(), a parallel implementation of the random forest classification algorithm prandomForest() and a parallel version of the rank product analysis technique pRP().
The main change in SPRINT beta 0.3.0 is the addition of a clustering function, ppam(), to the SPRINT library of parallelized statistical R functions. It performs a Parallel Partitioning Around Medoids (PPAM) and is based on the pam() function from the cluster R package.
In addition, the implementation of the SPRINT pcor() function which performs the Pearson correlation in parallel has been extended to allow the correlation between two matrices.
The main change in SPRINT beta 0.2.0 is the addition of a permutation test function, pmaxT(), to the SPRINT library of parallelised statistical R functions. It parallelizes the mt.maxT() function found in the multtest R package.
Also changes to the implementation of the SPRINT pcor() function allows for the distance matrix to be returned as the output of pcor() instead of the correlation coefficient matrix.
Finaly, SPRINT Beta 0.2.0 has simplified installation process and a reduced set of requirements.
The changes in SPRINT beta 0.1 have been targeted at improving the scalability of SPRINT. SPRINT can now process larger data sets. The restriction on the size of the output data has been removed thanks to the use of R ff objects and binary files. SPRINT successfully returns output data which is larger than the memory of the computer.
Further improvements to the HPC harness using MPI/IO means that SPRINT now scales almost perfectly to 512 cores, providing a major improvement in performance and execution time.
No additional function has been added to the library of parallelised statistical R functions.
The SPRINT prototype is a demonstrator application to show that R functions can be parallelised to run on High Performance Computing (HPC) therefore using several cores and achieving a significant speed-up over its serial counter part. The SPRINT framework is made of two components:
A parallel version of the Pearson correlation has been implemented and benchmark of this early prototype show improved scalability over the serial version.