Use arm neon instructions to optimize sgemm_beta operation

[WIP] Use arm neon instructions to optimize tcopy operation

AVX2 ZGEMM3M kernel

prefetching for dgemm_beta

add in runtime cpu detection for zarch

modified:   KERNEL.ARMV8
	modified:   KERNEL.TSV110
	new file:   sgemm_ncopy_4.S

Changes to build on AIX in POWER8 mode

This bug is not there in 8.x nor in the 9.0 daily snapshot.

Fix precompiled kernels on POWER9 and make their use conditional on (old) gcc version

Fix SEGV in cdot_power9

used in KERNEL.POWER9

We were corrupting r2 because the local entry wasn't being
setup correctly.

The ZARCH implementations of ?sum contain a cut & paste-error: An inline
assembly argument is named "sum", but the assembly references "asum"
instead.  The mismatch causes a build error.  This is fixed.

revised fix windows compatible for #2313

Signed-off-by: Wang, Long <long1.wang@intel.com>

Further Power8 big-endian corrections

Signed-off-by: Wang, Long <long1.wang@intel.com>

For large matrix, e.g. M=N=K, and M>1290, int mnk=M*N*K will overflow.
This will lead to wrong branching to single-threading. The performance
is downgraded significantly.

Signed-off-by: Wang, Long <long1.wang@intel.com>