/yg/f/f-cpu/vhdl/stimulib/README.txt
created Sun Jul 21 07:02:36 CEST 2002 by whygee@f-cpu.org
version Tue Jul 23 04:29:50 CEST 2002 : going VHDL

This directory contains the necessary files for reading and
writing test stimuli and results in VHDL. Their format is really
straight-forward : each line contains hexadecimal data for each
cycle, grouped per vectors (MSB first) and separated by a space.
Comments are not (yet) supported. Both input and expected
output values are put in a single file. Unknowns are not supported.

The order of the vector bits does not differ from line
to line. This order must be coherent with all the langages.
I also try to reuse existing string-handling capabilities
included in the library of each langage. Hexadecimal data
is both compact and better supported than binary format in C,
it's also used in VHDL. Since the string library is not rewritten,
it should be rather fast (and less prone to bugs).

I am worried about many things, particularly that
the stimuli files are portable and easy to use, but not
constraining in such a way that additions will be complex.
It is kept highly simple for many reasons, way beyond the issues
of the coding and debugging efforts. Please don't mess too much
with that... The problem is not to add features but to make
the tests work anywhere.

WARNING : pretty few validity checks are possible.
VHDL should report errors if the HEX file is flawed
but few checks are done by the lib itself. We'll also
have to trust the software that creates the vectors
(it's not a network app and we don't need to prevent
hackers from breaking in). Format error are indirectly
caught (through wrong results) but it's not garanteed.
Don't edit the hex files too much...
