W1066 Lost Extended floating point precision. Reduced to Double (Object Pascal)

From Appmethod Topics
Jump to: navigation, search

Go Up to Error and Warning Messages (Object Pascal) Index

Extended precision floating-point operations are supported by the Object Pascal 32-bit Windows compiler but are not supported by the 64-bit compiler. In consequence, extended precision floating-point computations can be performed only on Win32 and OS X, not on Win64.

In Win64, the precision of floating-point computations was reduced from extended (10-byte values) to double (8-byte values). We do not recommend using extended precision floating-point values. The Intel x86 architecture supports 10-byte real data types, but the trend of new processor architectures is to perform floating-point computations in parallel and with at most double precision (see the Intel® Streaming SIMD Extensions).

For reading extended precision floating-point numbers from a file, see TExtended80Rec.

See Also