Let me add to one point that Misha made (just to show I actually read these messages!)...>> >> Any ballpark ideas on when an alpha version could be available? Are we >> talking months or years here? > > First of all, I would like to point out that the goal is *NOT* to > compile Linux > to run natively on your favorite architecture; instead, we aim to > compile Linux > to a bytecode file which can then be run in userspace via an execution > environment (i.e., code generator + implementation of OS services for > a given > target). The goal here is to abstract away the hardware from the OS, > using > these magical `intrinsics' that were mentioned. This means that Linux > is ported > to run on a fundamentally new architecture, which we call LLVA. You > can see > our MICRO paper on this topic: > > http://llvm.cs.uiuc.edu/pubs/2003-10-01-LLVA.htmlThe above paper describes LLVA as a virtual instruction set for new processor designs. But in the context of Linux, LLVA can be a way to virtualize the processor even for ordinary processors. I thought I should clarify this because our MICRO paper does not try to make this point. LLVA is essentially LLVM extended with enough low-level details to act as a hardware instruction set, as opposed to "just" a compiler and user-level code representation.> This implies that it is not intended for end-user consumption but more > for a proof-of-concept and ongoing research potential.You're being too modest, Misha --- this is going to be the Linux of the future :) --Vikram http://www.cs.uiuc.edu/~vadve http://llvm.cs.uiuc.edu/> > That said, we don't really have an estimate for when this will be > functional > and available. It's ongoing at this time, but as John said, in very > early > stages. > > -- > Misha Brukman :: http://misha.brukman.net > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev
On Wed, 2003-11-12 at 11:45, Vikram Adve wrote:> > This implies that it is not intended for end-user consumption but more > > for a proof-of-concept and ongoing research potential. > > You're being too modest, Misha --- this is going to be the Linux of the > future :) >I'll add my pragmatist $0.02 worth here. Take it with a grain of salt. If LLVA isn't "real world" or "intended for end-user", how valuable is it as a research tool? While I agree that it might be useful as a proof-of-concept, research needs to founded in reality (i.e. how real users would use it). Since you've decided to open this up as source code, don't think that you're alone in making it "real". I for one plan to base a very real product on LLVM and possibly LLVA. Because I'm not aiming for a research tool, you'll get the benefit of open source (i.e. making it "real"). That should only make your research better. So, I tend to agree with Vikram .. shoot for the moon! Even if you only get half way there, its better than only getting halfway down the street around the corner. -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20031112/2ffb8a81/attachment.html> -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: This is a digitally signed message part URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20031112/2ffb8a81/attachment.sig>
On Wed, Nov 12, 2003 at 04:05:20PM -0800, Reid Spencer wrote:> On Wed, 2003-11-12 at 11:45, Vikram Adve wrote: > > > This implies that it is not intended for end-user consumption but > > > more for a proof-of-concept and ongoing research potential. > > > > You're being too modest, Misha --- this is going to be the Linux of > > the future :) > > I'll add my pragmatist $0.02 worth here. Take it with a grain of salt. > > If LLVA isn't "real world" or "intended for end-user", how valuable is > it as a research tool?Perhaps I should say what I really meant to say instead of vaguely generalizing. :) The thing that will exist at some point (hopefully soon) will be essentially an emulator for LLVA. *THAT* is not intended for an end-user, an end-user here being someone who is not a compiler/architecture researcher or developer. This is because unlike something such as VMware, it does NOT have off-the-shelf software it can run, unless we make an effort to port it, or make LLVM able to compile it. THAT is what I meant. LLVA is a definitely for everyone, but as an architecture implemented in hardware, not as an emulator. -- Misha Brukman :: http://misha.brukman.net
On Wed, 12 Nov 2003, Reid Spencer wrote:> If LLVA isn't "real world" or "intended for end-user", how valuable is > it as a research tool? While I agree that it might be useful as a > proof-of-concept, research needs to founded in reality (i.e. how real > users would use it). Since you've decided to open this up as source > code, don't think that you're alone in making it "real". I for one plan > to base a very real product on LLVM and possibly LLVA. Because I'm not > aiming for a research tool, you'll get the benefit of open source (i.e. > making it "real"). That should only make your research better.I, for one, completely agree. My intention is for LLVM to be fully robust and "commercial quality". This takes time, of course, but that's the goal. LLVA will start out simple and not-quite ready for prime-time, but will, over time, becomes something generally useful. World domination can only be achieved one step at a time. :) -Chris -- http://llvm.cs.uiuc.edu/ http://www.nondot.org/~sabre/Projects/