Re: [cc65] Java VM built with cc65

From: Oliver Schmidt <ol.sc1web.de>
Date: 2007-08-09 00:38:37
Hi Dan,

First of all thanks for your reaction - and obviously taking at least
a peek into the NanoVM source code.

> It looks like a big gain would result from converting
> the small functions called by vm_run to macros. Also,
> sp should be assigned to zero page (although you need
> my patch to get the benefit).

Maybe I don't interpret the situation correctly but as far as I can
see Uz is - lets say somewhat reluctant - to accept your patches.
Nevertheless I presume that you'd like to see at least parts of them
in the main code line.

Probably Uz would be impressed by noticable sppedups of the NanoVM
with your changes. And if I could reproduce the performance
improvements I'd urge Uz as well to include the changes. Maybe it's
just a real world show case like the NanoVM missing ?!?

Regarding the function vs. macros issue: I seem to remember that your
improvements had a lot to do with function calls (esp.. leaf
functions) so would it make sense to check out what the changes can do
with the NanoVM code AS IS ?

My general issue is that I never compiled cc65 myself so far and don't
think it's that easy to do so for Windows. And additionally I have no
clue how to handlie those "magic" patch/diff things.

So after all it seems more reasonable to me that you take the NanoVM
source and do some tests - especially as you know exactly what type of
patterns / changes in th generated code you're looking for.

In case you have issues with building/using the NanoVM I'd be happy to
help if I can...

Does this approach make any sense to you?

Best, Oliver
----------------------------------------------------------------------
To unsubscribe from the list send mail to majordomo@musoftware.de with
the string "unsubscribe cc65" in the body(!) of the mail.
Received on Thu Aug 9 00:38:52 2007

This archive was generated by hypermail 2.1.8 : 2007-08-09 00:38:55 CEST