Fixes for PGI compiler

power9 cgemm/ctrmm. new sgemm 8x16

Fix build on PPC970 for FreeBSD

FreeBSD needs those macros too.

FreeBSD needs DCBT_ARG=0 as well.

Fix build on FreeBSD/powerpc64.

Signed-off-by: Piotr Kubaj <pkubaj@anongoth.pl>

Fix DTRMMKERNEL register save for power8 64-bit mode (Fix for #2166)

PGI assembler does not like the initialization in the constraints.

fixes compile failure with pgi 18.10 as reported on OpenBLAS-users

Avoid unintentional activation of TLS codepath via USE_TLS=0

fixes #2149

Remove any inadvertent use of -march=native from DYNAMIC_ARCH builds

from #2143, -march=native precludes use of more specific options like -march=skylake-avx512 in individual kernels, and defeats the purpose of dynamic arch anyway.

Add gfortran workaround for potential ABI violation

for #2154

Thread safety tester using C++11 threading (cleaned history)

Add gfortran workaround for C->FORTRAN ABI violation

for #2154 (see gcc bug 90329)

for #2154 (see gcc bug 90329)

improved power9 zgemm,sgemm

Separate implementations of AMAX and IAMAX on arm

Fix detection of Skylake processors when using GCC

21eda8b5 introduced a check in getarch.c to test if the compiler is capable of
AVX512. This check currently fails, since the used __AVX2__ macro is only
defined if getarch itself was compiled with AVX2/AVX512 support. Make sure this
is the case by building getarch with -march=native on x86_64. It is only
supposed to run on the build host anyway.

for #2151

As noted in #1912 and comment on #1942, the combined implementation happens to "do the right thing" on hardfp, but cannot return both value and index on softfp where they would have to share the return register

Revert "Add softfp support in min/max kernels"

Add softfp support in min/max kernels

Build and run utests independently of fortran

fix for #1912

Do not try ancient PGI hacks with recent versions of that compiler