Slashdot Mirror

← Back to Stories (view on slashdot.org)

Digital's FX!32 and Open Source?

Posted by Cliff on from the wouldn't-this-be-nice dept.

Anonymous Coward muttered this question in my ear: "Has anyone thought about approaching Digital to release FX!32 under an open source style license, so that people running Linux/Alpha can take x86 binaries and run them, the same as they can for Windows NT Alpha? What about making the FX!32 backend work on the other non x86 PCs, Sparc, PPC, ARM, etc? Is there a team working on a kernel module to do this? I think it could fall under the Binary format option. I think this would make Linux/some other CPU more attractive then Linux/x86. I use Linux x86 myself, but would like to use an Alpha or PowerPC, however apps like Real Player, Codewarrior, Word Perfect, Star Office, some of which I use, as far as I know, only exist for x86. "

10 comments

  1. The coolness of FX!32 by GoRK · · Score: 1

    I'm sorry I can't say whether or not anyone has approached Digital (Compaq) about this, but I have to say that FX!32 is bar-none the absolute best transparent emulation I have ever seen. It simply executes x86 code on Alpha processors at speeds that used to exceed even the fastest x86 processors. It chews Java up and spits it out. It beats IBCS into the ground.

    That being said, I would support a petition calling for open-sourcing this project or even providing a biary port, possibly as an extention to IBCS After all, FX!32 it only works on Alpha's which could expect a sales increase because of the gesture - God knows the Alpha needs it. Perhaps the whole thing could spark a new slew of 'emulation' kernel modules (Say hello to Transmeta) that would allow anyone with any processor to run applications designed for any other one. Sure it'd be slightly slower, but not *MUCH* slower if done properly, and it would mae Linux *MUCH MORE ATTRACTIVE* to commercial developers.

    ~GoRK

    1. Re:The coolness of FX!32 by Anonymous Coward · · Score: 0

      I agree with getting a petition going. Anyone want host the petition and do a little perl script to log the requests? It should probably ask for (and require): Your name, email address, whether you want it open source or comercial (if the latter, how much will you pay?).

      Another thought, is to get a petition going for Softwindows to be ported to Alpha/Linux and possitbly Tru64 UNIX? (asking for one at the same price as the MacOS version).

      Any other thoughts?


      LONG LIVE ALPHA!!!

    2. Re:The coolness of FX!32 by Anonymous Coward · · Score: 0
      You should check out theFreedows Project They want to do basically the same thing, except they're starting from scratch and writing a whole new (GPL'ed) OS.

      The plan is for the first release to support Linux and part of Windows. In later releases, they want to add more Windows support, and MacOS. Also, there doesn't seem to be anything about it on their new site, but the old site listed the C64 and several consoles, among others as other possible emulation modes.

      Finally, they plan to port GCC, so you'll be able to write native programs, too.

    3. Re:The coolness of FX!32 by Rombuu · · Score: 1

      Oh man, those guys....

      I they are planning on proving Fermat's Last Theorem while simultaneously working out the perfect solution to the traveling salesman problem after they do all of the above.

      --

      DrLunch.com The site that tells you what's for lunch!
    4. Re:The coolness of FX!32 by spinkham · · Score: 1

      Sorry guy, Fermat's Last Theorem has been solved...

      --
      Blessed are the pessimists, for they have made backups.
    5. Re:The coolness of FX!32 by GoRK · · Score: 1

      Fermat's Last Theorem has been proven. The guy that did it thought he proved it, but then someone found an error in his (100+ page) proof. Luckily he was able to fix it within a few weeks and then the thing was done.

      I suppose the Freedows people find the proof extremely bloated and will look to find a more streamlined, modular proof that could be used as parts of other proofs. They're even going to port GCC to it too, so you'll be able to prove things natively!

      ~GoRK

  2. they spent like 5 years on FX32 by johnjones · · Score: 1

    they spent so much time on FX32 !
    they are unlikely to give away the source !

    what they are likely to do if Pressured is give away it for alpha linux to increase sales and PR in linux world just like the alpha math libs

    sorry but you are dealing with compaq and not digital so a move would take @ least a year and everybody else have done it before

    isnt their a emulator for intel to alpha linux ?
    yep just forgot where it is
    if you cant compile it or the vendor wont surplie it for your system at a cost its not that good !!

    cya

    john

    a poor student @ bournemouth uni in the UK (a deltic so please dont moan about spelling but the content)

  3. x86 Emulator for Linux/Alpha by Anonymous Coward · · Score: 0

    There is!! you can download em86 from the following URL. It's been around for a while. ftp.digital.com/pub/Digital/Linux-Alpha /em86 hope this helps all of you out there with those amazing alphas!

  4. em86!=FX!32 by Anonymous Coward · · Score: 0

    em86 is not as good as FX!32. FX!32 and em86 both translate x86 into alpha on the fly. Where FX!32 rules is that it also writes that new code back into the binary. FX!32 also optimizes every time the app is run. You get more native alpha code each time the app is run. em86 just translates it ever time. Why shouldn't compaq give away FX!32. They have already signed the death warrent for Nt on alpha. FX!32 is a great technology. It hopefully wont go the way of NT on alpha.

  5. Oh.. Did I mention what I do? by QuantumG · · Score: 1

    Hey.. I'm a research assistant on the University of Queensland Binary Translator project.. The web page tells you everything I'm about to repeat here. UQBT is a "retargetable" binary translator.. we're actually using the term "resourcable" now.. essentially you take a binary that runs on any architecture that you have a spec file for and run it through uqbt.. what you get out is a bunch of .c and .s files (one .c for each function and one .s file for each section) and a Makefile.. you can then take that and put it on any architecture you want (currently we don't do cross OS's so you have to take, say a linux binary and rebuild it on another linux machine or a solaris binary and rebuild it on another solaris machine) and rebuild it there.

    My research is taking this C output and getting it to compile to java bytecode and then writing the necessary support routines in java to get the binary running.. currently I can translate small example programs and benchmarks (with some of em running 10 times slower, man java sux.. then again, some of em run faster on the JVM with JIT's turned on).

    And that's what I do. We refer to FX!32 in half of our papers cause it is a good example of binary translation.

    --
    How we know is more important than what we know.