I just started looking at DotGNU and this statement intrigued me:
"We first convert the CIL bytecode into a simpler instruction set for what
we call the Converted Virtual Machine (CVM). The simpler CVM instructions
are then executed using a high-performance interpreter."
This makes a lot of sense to me because IL code was written specifically to
be JIT'ed, unlike Java's Byte code. So it's inefficient to interpret IL code
directly (like mono's mint).
I realize that this CVM interpreter is going to be much slower than the new
JIT compiler, but this is still an interesting option that can make it
easier to support a new platform. I'm thinking about a smaller embedded
platform that doesn't have a ton of resources, so a lean interpreter sounds
like a good choice.
Is the CVM still a viable model, or has it been super-ceded by libjit? I
understand libjit can fallback to interpretation on a platform that doesn't
have JIT support yet, but I'm fuzzy on the mechanics of this. Can the
current C# compiler still generate CVM code as an option?
Eric
_________________________________________________________________
Talk now to your Hotmail contacts with Windows Live Messenger.
http://clk.atdmt.com/MSN/go/msnnkwme0020000001msn/direct/01/?href=http://get.live.com/messenger/overview
_______________________________________________
Developers mailing list
[email protected]
http://dotgnu.org/mailman/listinfo/developers
