clMathLibraries / clSPARSE

a software library containing Sparse functions written in OpenCL
Apache License 2.0
173 stars 61 forks source link

Read explicit zeroes from Matrix Market files #168

Closed jlgreathouse closed 8 years ago

jlgreathouse commented 8 years ago

Fixes issue #146

As requested in the previous patch (#163), I've made changes to the matrix market reader functions (e.g. clsparseSCsrMatrixfromFile) to take an added boolean input. This input controls whether the function reads explicit zeroes that are stored in the matrix market file, or whether it ignores them and does not put them in the data structure.

Changed all the programs that eventually call the matrix market reader to (by default) read in explicit zeroes, as requested in this post by @kknox. Added command line parameters to all of the applications (except the samples) to control this.

@kvaragan, I was unable to reproduce the error you brought up in this post from the previous PR. I fixed some other errors in the test programs along the way, and I believe the test-conversion application only fails in the same places it failed before (e.g. Issue #153).