Only use the AVX512 version of memswap when compiling with the Intel compiler as GCC seems to produce invalid code.
Also allocate the offset arrays using an aligned allocation.
Fixes #428 (closed) and #430 (closed).
gitlab has been updated to version 16
Only use the AVX512 version of memswap when compiling with the Intel compiler as GCC seems to produce invalid code.
Also allocate the offset arrays using an aligned allocation.
Fixes #428 (closed) and #430 (closed).