I know, I know, who wants to start another language war :-)

But I thought I'd share a recent experience where MicroChip's C30 really=20
impressed me.

I needed to implement the XTEA encryption scheme, and needed performance. I=
=20
immediately thought of doing it in assembler for that reason. So, I spent=20
quite a bit of time writing a highly optimized version in assembler. I have=
=20
quite a bit of experience in assembler and I feel the version I wrote is at=
=20
least pretty close to as fast as it can get. I was pleased with the result:=
=20
1603 instruction times to encode each 8-byte block (at 60 MIPs on a PIC24E=
=20
that is nearly 300,000 bytes/sec!). Then, I decided to see how well C30=20
would do with it and wrote a straightforward implementation in C. I looked=
=20
at the generated code and was very impressed. Timing the C implementation=20
gave me 1889 cycles, which is only about 18% slower than the asm=20
implementation. As a final exercise, I coded up an assembler version=20
optimized for space instead of speed. It took 3016 cycles, but only used 89=
=20
instructions versus 132 instructions for the fast assembler version and 146=
=20
for the C version. So, in summary:

FAST ASM,             C VERSION,            SMALL ASM
1603  cycles,             1889 cycles,              3016 cycles
132 instructions,        146 instructions,         89 instructions

By the way: if anyone is interested in the source, I'd be glad to send it t=
o=20
you [ as long as you aren't in a foreign country on the no export list :-( =
]

-- Bob Ammerman
RAm Systems=20

--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.