Hi Gavin,
Am 03.04.2008 um 22:16 schrieb Gavin Romig-Koch:
the ultimate purpose would be the latter; changing the machine code
would then also be possible, but that's not what I want to do. It's
rather about approaching something like inline caches (other than
those already present in the id model).
Well, on modern hardware and modern OSs, I really don't think
nestling the inline caches amongst code makes sense - it plays havoc
with the cpu's memory cache lines, and disallows any use of the code
in a multi-threaded or multi-processed way. Much better, I think,
to allocate the caches out of thread local or process local memory.
that sounds sensible; but let me poke at the "havoc" thing a bit... I
have heard this stated informally several times. Is there some source
of related measurement information? Given that inline caching was
introduced to improve performance (and is still in use), it would be
interesting to see some actual benchmark results that nail this down.
Related question: does threaded interpretation still make sense these
days, what with all those sophisticated branch prediction units
around? Again: are there reliable sources?
Anyway, I'm trailing off. Sorry, but this is an interesting topic. :-)
This is a sketch of how I would go about this, as a first attempt:
[...]
Your suggestions sound worthwhile, thanks a lot; I will have a look at
the places in the source code you mentioned. It seems you have
forgotten that "actives" example you announced, though. ;-)
Best,
Michael
--
Dr.-Ing. Michael Haupt [EMAIL PROTECTED]
Software Architecture Group Phone: ++49 (0) 331-5509-542
Hasso Plattner Institute for Fax: ++49 (0) 331-5509-229
Software Systems Engineering http://www.swa.hpi.uni-potsdam.de/
Prof.-Dr.-Helmert-Str. 2-3, D-14482 Potsdam, Germany
Hasso-Plattner-Institut für Softwaresystemtechnik GmbH, Potsdam
Amtsgericht Potsdam, HRB 12184
Geschäftsführung: Prof. Dr. Christoph Meinel
_______________________________________________
fonc mailing list
[email protected]
http://vpri.org/mailman/listinfo/fonc
