For Intel CPUs
Little-endian
Signed integer (2's complement), unsigned integer, bitfield, and IEEE floating point
8-bit, 16-bit, 32-bit, and 64-bit
I don't think it matters in representation in our use cases, but using the standard types H5T_STD_[I|U]8[L|B]E for int and uint or the ANSI C9x-specificH5T_NATIVE_* might be an alternative.
We are currently using
H5T_INTEL_*
types forint
anduint
CollectionType
s.Following the documentation they are designed:
I don't think it matters in representation in our use cases, but using the standard types
H5T_STD_[I|U]8[L|B]E
forint
anduint
or theANSI C9x-specific
H5T_NATIVE_*
might be an alternative.