Re: FastCode RTL Replacement v0.20 Released!

Hi All

The Fastcode library design is very simple.

There is one of the following library units for each challenge

Direct calling
Conditional compilation
CPUID selection

and a patching unit if the challenge is based on a RTL function.

A direct calling unit has 9 winner functions = one for each target. All 9
functions are interfaced as plain function calls.
A conditional compilation unit has the same 9 functions. They are interfaced
via 1 function. The implementation is selected via a compiler directive.
CPUID interfaces one function and the implementation is selected via CPUID.

A Pacthing library unit has 9 winner functions = one for each target. The
function implementation is based on CPUID selection. The function is
"interfaced" via patching.

Which functions is selected via CPUID is defined in the CPUID unit. Each
processor maps to a target. This mapping is defined.

We made this design because different users prefers different designs. We
make available 4 different ways of using our winner functions. This way we
can satisfy Fastcode users better.

We do not have a library type called "FastCode RTL Replacement". This was a
new name invented by you. It clashes with the name of one of our targets.

You want to make patching units? Then follow the design of Fastcode paching
libraries.

Best regards
Dennis


.

Relevant Pages