The following is a list of functions handled directly by evalhf. These functions are consequently executed without using Maple code, merely by calling the corresponding mathematical function from the C library or executing C code.

The functions listed above should execute at speeds significantly higher than the Maple-evalf equivalents.

In some cases where complex hardware floating-point values are involved, evalhf may call the evalf versions of these functions, which will incur the time cost of conversion to and from software floating point, and calculation using software floating point.

Maple kernel builtin functions not listed above are able to be executed within evalhf provided that they return a hardware-float data structure that is compatible with the evalhf subsystem.  These commands usually invoke compiled code, but may also be partially implemented using Maple code.  It is important to note that the builtin command and its arguments are processed outside of evalhf.  For example, consider the following example which uses the built-in function, rhs.  Evaluation of rhs and its arguments will be performed outside of evalhf so the non-hardware range structure returned by rtable_dims can be computed, and subsequently given to rhs, which will return a hardware compatible structure back to the evalhf subsystem.

A := Array(1..2,datatype=float[8]):

evalhf( rhs(rtable_dims(A,1) ) );

$2.$