You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
22 lines
1.1 KiB
22 lines
1.1 KiB
6 years ago
|
Project includes 6 versions of a knn algorithm implementation:
|
||
|
|
Serial - space optimized
|
||
|
|
Serial - time optimized
|
||
|
|
MPI parallel - blocking communications - space optimized
|
||
|
|
MPI parallel - blocking communications - time optimized
|
||
|
|
MPI parallel - non blocking communications - space optimized
|
||
|
|
MPI parallel - non blocking communications - time optimized
|
||
|
|
|
||
|
|
Project folder also includes some test files and execution results (stats folder).
|
||
|
|
|
||
|
|
In folder testFiles there is a dataset of 60000 points of 30 dimensions each, as
|
||
|
|
well as three IDX files extracted from Matlab storing correctly sorted IDs.
|
||
|
|
These files are named according to the convention: numberOfPoints_k
|
||
|
|
after the number of points and selected k that were used to run the Matlab
|
||
|
|
script.
|
||
|
|
|
||
|
|
To run any version first run make. Then copy testFiles/data.bin and one of the
|
||
|
|
IDX test files into the folder. Finally run with:
|
||
|
|
mpiexec -np numTasks ./prog.out numPoints numDimensions k data.bin idxFileName
|
||
|
|
|
||
|
|
To extract a new IDX file from Matlab run knn with the dataset and then extract
|
||
|
|
IDX variable to an ods/excel. Open the generated file and save as csv.
|