Add support for RISC-V

Include sys/types.h for proper typedefs related to wait()

Catch invalid cpu count returned by CPU_COUNT_S

Should fix #1519

mips32 was seen to return zero here, driving nthreads to zero with subsequent fpe in blas_quickdivide

Correct precision of mips dsdot

Add macOS to the Travis testing matrix: Take 2!

Revert "Add macOS to the Travis testing matrix"

Add macOS to the Travis testing matrix

Add DragonFly to exports/Makefile

Its exclusion was an oversight on my part.

Fix thread races and infinite looping on systems with many cpus

Underline importance of NUM_THREADS setting for BUFFER allocation

Minor changes to wording and formatting in the README

The wording in some places is not grammatically correct. This change
also provides minor adjustments to the Markdown formatting which provide
modest improvements to readability.

Compile with cc rather than gcc whenever possible

following augray's suggestion from #1451, and incorporating ashwinyes' comments from #1141 on the importance of NUM_THREADS even for single-threaded builds.

On systems with more than 64 cpus, blas_quickdivide will sometimes return zero which creates bogus workloads when used for the stride calculation. This then leads to threads spinning incessantly waiting for a status change that never happens, as seen in #1497.
This patch also fixes several data races that were found by helgrind and/or tsan while debugging the issue.

Allow building on OpenBSD

Add workaround for old gcc and clang versions

With this change, OpenBLAS builds and all tests pass on OpenBSD 6.2
using Clang. Tested on x86-64 only, with and without DYNAMIC_ARCH=1.

Old gcc and clang do not handle constructor arguments, finally fix #875 as discussed there, using the fedora patch

Correct index variables used in MFlops calculation

Fixes #1474

Implemented missing vsx simd  kernels for power8 blas1/2 double. z13 modifications

Add multithreading support for Haswell DDOT

z13: improved zgemv_(t|n)_4,zscal,zaxpy

Disable CPU affinity by default again

This setting must have been changed unintentionally by my PR #1214 (probably leftover from unrelated tests)

Use generic/dot.c instead of the inferior arm/dot.c for x86 DSDOT

to resolve dsdot utest failure seen in #1492

copied from ashwinyes' implementation in dot_thunderx2t99.c

Re-enable DAXPY AVX microkernels  for x86_64

as the inaccuracies seen in the original testcase for #1332 appear to be due to an artefact that amplifies the very small rounding differences between FMA and discrete multiply+add