You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Int128/Uint128: presumably similar to Complex128 (encode as pair of Uint64). The holdup: is the sign bit portable?
HDF5 allows defining arbitrary integer datatypes by specifying their number of bits. This would seem to be the most natural approach.
Otherwise, a similarly "natural" approach would be to manually decompose them into their lower and upper 64 bits, and storing these as Uint64 and Int64, respectively. (Uint128 is then stored as two UInt64.) This is portable; there is no computing architecture today that deviates from the signed two's complement representation.
The text was updated successfully, but these errors were encountered:
The documentation says:
HDF5 allows defining arbitrary integer datatypes by specifying their number of bits. This would seem to be the most natural approach.
Otherwise, a similarly "natural" approach would be to manually decompose them into their lower and upper 64 bits, and storing these as
Uint64
andInt64
, respectively. (Uint128
is then stored as twoUInt64
.) This is portable; there is no computing architecture today that deviates from the signed two's complement representation.The text was updated successfully, but these errors were encountered: