This release adds the following features:
- add support for performance monitoring using PAPI
- add options for specifying the number of rows/columns per thread
- add support for reading gzip-compressed Matrix Market files
- align allocations to page boundaries if HAVE_ALIGNED_ALLOC is defined.
- add support for symmetric matrices and matrices with integer and pattern fields
- add compile-time option for selecting 32- or 64-bit integers for row/column offsets
- add CSR SpMV kernels where rows or nonzeros are partitioned evenly among threads