Conference 7.286::digital

From the sounds of Bob Palmer's DVN today FX!32 is NOT dead.

/Mario

PM for FX!32 is Pippa Jollie

CHIPBZ::JOLLIE

    I would say the rumor in .0 is rubbish, based on what's going on
    in the FX!32 notes file.  They just released another beta about a
    week ago.  Plus in a DS company strategy presentation a short time
    ago FX!32 was right up there.
    

	I attended the taping of the Bob Palmer DVN in Greenbelt, MD and he 
made a very specific reference about providing additional funding for the FX32 
project. He indicated that this was one of his key areas that he was going to 
invest in, because he saw this technology as a way of getting Alpha Chip 
technology accepted into the Intel market.

Based on Bob's words, IMHO would have to say that FX32 is not dead.

Regards

    	Thanks for the imput so far.  Unfortunately the word out on the
    streets is that it's dead or will be soon.
    
    	Hopefully Bob's message will spread to the "real world" too...
    
    Regards,
    
    TL  
    	
    

4.0 or both 3.51 and 4.0?  I don't consider FX!32 useful since
I'm still using 3.51.  Getting rid of it made my netview problems
go away as well... YMMV...

    
    I'm sure some competitors would want FX!32 dead, but its very much
    alive, as some people have already noted. There are ~500 internal field
    test sites and ~170 external sites. Info is available on the DS web
    page and we communicate with external customers via the
    www.services.digital.com web site. If you have an Alpha running NT I
    invite you to  join the field test and see for yourself. The latest
    Alpha systems are running some X86 benchmarks (ie Intel code) faster 
    than the fastest P6/200 systems - something that has never happened 
    before. So see for yourself.
    
    /maurice marks
    (DS technical director and group manager of the Alpha Migration Tools
    group that developed FX!32).
    

    FX!32 is definitely not dead.  I asked Pippa Jollie for a "suitable for
    public consumption" estimate of a release date and she replied "the
    week of September 30" as being the plan.
    
    					Steve

    Thanks again for the positive feedback.  I'll be contacting the VAR
    today with the news.

    I looked for the FX!32 Notes Conference by doing a key word search in
    Easy...  here is what I learned
    
    1) the FX!32 Notes Conference has restricted access, so that product
    management can monitor who is using the Internal Beta
    
    2) Non Restricted info about FX!32 is available in
    TALLIS::Alpha_Migration_Tools     
    
    	a good pointer note is 930.1 in the above Conference.
    
    	907. has a real good 150 line or so technical overview 
    
    Long live FX!32, long live Alpha, get that stock price up!

    
    Check the WWW @:
    
    Technical introduction to Digital FX!32
    http://www.digital.com/.i/info/semiconductor/amt/fx32/fx.html
    
    
    Digital FX!32 Field Test
    http://www.service.digital.com:80/fx32/

    	I am astounded that anyone would suggest or believe it is dead.
    	It is one of the big keys for us that will soon start its ramp.
    
    	I happen to run it at 500MHz and am just silly over it.
    
    			RJD

    	OK - on second thought, competitors would lie about it.

Will we announce that 500 MHz Alpha before the competition
announces their next faster machine?

:Will we announce that 500 MHz Alpha before the competition
:announces their next faster machine?

   We've already announced it.  AlphaStation 500/500.  Shipping September.

    I think it's even mentioned elsewhere in this notesfile, though the
    fact that you weren't aware that we're in the lead again is an
    interesting sign of the effectiveness of our internal communications.

    Re: 14
    
    	See note 4770.9
    
    		RJD

    There is a double-page add in the July edition of PC Computing
    wherein Polywell Systems advertises their Alpha-based PCs as
    being able to "run almost all PC software utilizing DEC's FX32
    translation program."
    
    It's a good ad, with some nice words for "DEC's high intensive 
    Alpha processor" - I hope our Alpha PC and FX32 ads aren't far 
    behind.
    
    And note the reference to DEC not Digital.
    
    Barrie 
    
    

    re "run almost all PC software"
    
    Hmmm - shame we can't get closer than that. Smacks of incompatibility
    and potential problems. What is the real scoop here?  i.e. what type
    of software is FX32 definitely not good at?  Can we say for example
    that it is guaranteed to run ANY true 32-bit Windows application?
    
    /Chris/

    Without any first-hand knowledge of the product, a goal of running
    100% of PC software is impossible: not even Windows NT on Intel can
    do that, nor Windows 95, nor Windows 3.11.
    
    Certainly, software that expects direct access to specific hardware,
    such as many games and utility programs, wouldn't be expected to
    work under even the best emulation.

One class of software FX!32 won't run is drivers.

			Steve

    I am running FX!32 internally and have tested for some customers with
    their software. The product is a good x86 emulator, but, it does not
    run all x86 software "out of the box". I have had to add such things as
    registry entries on my ALPHA that would only be found on an x86 system
    running NT in order for a given piece of software to work. There is a
    list of qualified software, but if you work at it you can get much more
    out of FX!32. The danger is in promoting it as running "all" x86
    software out of the box.
    The big picture is that it beats WABI by a country mile. The upcoming
    Intel/HP whateverhybirdchip may, probably, will have some
    incompatibility problems with legacy x86 apps and they will have to
    address that at a time when FX!32 will have matured somewhat (if there
    is a digital or alpha in two years). 
    My problem is that many of my customers were exposed to the FX132 story
    early last year, were ready for it, called looking for it, and now
    accept it as marketing vaporware since it has taken so long. I hope
    they are so dazzled with what we have accomplished in management reorgs
    that the technical stuff, or lack of, won't be an issue.

My understanding is that there will be an "open beta" copy made available 
within a week or so - it will require filling out a web form to gain access.

				Steve

Paul:

> Certainly, software that expects direct access to specific hardware,
> such as many games and utility programs, wouldn't be expected to
> work under even the best emulation.

  I'm just picking a nit here, but we both know that an emulator
  *COULD* be designed to run any and all software that actually
  runs on a given hardware configuration. That is, the emulator
  would be sufficiently comprehensive that it also emulates all
  the I/O devices including the graphics frame buffers and such.

  But the emulator is also likely to be massive and slow, so it
  probably wouldn't be practical.

  I think by "best emulation" you probably meant "best practical
  emulation." :-)

  Like I said, I'm just picking at that nit.

                                         Atlant

>  But the emulator is also likely to be massive and slow, so it
>  probably wouldn't be practical.
    
    I think that's compatible with my "wouldn't be expected" comment.

	I think the word "emulator" when talking about FX!32 is mis-used.
	Although there is some emulation code in there, it's primarily
	a translator, translating Win32 Intel calls to Win32 Alpha calls.
	By doing this, the speed at which an application runs is
	significantly faster than if run thru a traditional emulator.

							mike

Mike:

  Well, it could be that we're splitting hairs, but I think you'll
  find there are both components. There's a binary translation
  process (which due to licensing restraints on the 'x86 binaries
  may be done at any of several different stages) and then there's
  the run-time environment emulation that handles the WIN32 calls.

  I *DON'T* know how "FX!32" handles run-time-modifiable code, whether
  it handles it at all, and whether the 'x86 world uses such ugly stuff.

  I think most people will think of "FX!32" as an emulator regardless,
  and I think the dictionary would support that usage.

                                             Atlant

re Note 4789.24 by ATLANT::SCHMIDT:

        I seem to remember that in the old days of IBM's CP/CMS
        operating system, it could run multiple copies of entire
        different operating systems (for the same architecture, of
        course).

        Yes, access to I/O hardware was emulated.

        Bob

        The Aug 19th EE times has an article titled 

         "x86 emulation: IBM drops it, Digital ramps it"

    -Bruce 

    >> I *DON'T* know how "FX!32" handles run-time-modifiable code, whether it
    >> handles it at all, and whether the 'x86 world uses such ugly stuff.
    
    If it's ugly, you can bet that the 'x86 world uses it :-)

FX!32 handles it.

	Steve

    As Steve says, FX!32 does indeed handle run-time generated code,
    i.e. code generated on-the-fly into buffers, heap, or on the stack.
    
    Note that to date, we have not encountered any truly "self-modifying"
    x86 code. But the FX!32 interpreter would also handle that.
    
    So the short answer is yes.
    
    /George

    re .28,
    
    You can still 'do' this with VM systems on ES9000 (plus clones) today.
    
    Ah.. the joy of second (and third) level Virtual Systems..
    
    Hmm.. if we had a VM like hypervisor on Alpha, then you could run NT,
    VMS and Unix all at the 'same' time. Rathole opened....
    
    
    AW
    
    (an ex-VM Systems Programmer)

AW:

> You can still 'do' this with VM systems on ES9000 (plus clones) today.

  No, no, no, you've got it all wrong. It *CAN'T* be any good if
  it wasn't (re-)invented by the PC generation! :-)  And if they
  can't do it on their puny, half-baked 'x86 architectures, it
  can't be done, don't you know?


  Amazing, isn't it?

                                          Atlant

> Hmm.. if we had a VM like hypervisor on Alpha, then you could run NT,
> VMS and Unix all at the 'same' time. Rathole opened....

Rathole entered: this was the original plan for Prism, Alpha's predecessor.
One of Cutler's ideas, I believe.

___Pete

    Tried to put this in the FX!32 notes conf., but it's restricted.  Also
    have calls into Pippa Jollie.
    
    Anyway...
    
    I have a nervous client (IBM shop) of my VAR account.  My account has
    pushed an Alpha solution for a - mostly to be developed NT-based application
    suite for order management.
    
    We had it *sold* until the client called IBM and asked if their MQ
    middleware port plans included Alpha.  Answer came back negative.
    
    We discussed the capabilities of FX!32 to the customer, but they don't
    want "an emulator".  Tried to correct that terminology also.
    
    Anyway, is there anyone with (ideally) MQ middleware experience of
    having run it through FX!32 on Alpha NT (4100), or who can volunteer to
    get on a call with my VAR (and/or the client) to give a convincing
    explanation of why MQ will work via FX!32?
    
    Time is of the essence.  A win here will save the sale of a 4100
    development machine, leading to at least 1 Turbo production machine in
    a few months.  BTW, the client is in Chicago.
    
    Bill Thomas
    264-0018 

    
    Sorry to rathole...
    
    I havent tried it, and would be surprised if anyone had.
    
    You havent said what they are using the MQ Series for. If its Notes
    integration to IBM backends you are probably shot. However, if the
    application is custom or not pre-existing, have you thought of the 
    DECmessageQ to MQ Series bridge? You can run DECmessageQ native on NT
    Alpha and if they insist on MQ Series on the backend do it through the
    bridge.
    
    If the integration to the back is not glued into MQ Series yet
    DECmessageQ has some very interesting properties in terms of
    recoverability that may be a better option for them. Again, people need
    to understand what your VARs customer is trying to achieve and what they
    currently are locked into.
    
    This is probably better taken up in the DECmessageQ conference
    (PAMSRC::DECMESSAGEQ), along with your FX!32 query. Someone may even be
    able to try it out for you.
    
    Regards,
    John
    
    

.35:

Or SAFE, the predecessor of both Alpha and Prism.  Alan Kotok's idea, for
which Dave Cutler took credit.

    Re: .35
    
    Yes, PRISM was the hardware, but Mica was the Hypervisor (should you
    want to call it that).
    
    

    
    Re .33 ahh yes you can bring up another MVS host under VM, or as
          many as you had disk, memory, etc. I could bring up the
          development system right next to the production system. Install
          patches do some testing, run paralell to the production system
          crash, do anything, and both Virtual machines running under VM could
          talk to one another.
    
       
    

    
    This may surprise some, but Digital has had this for years.
    A secure "hypervisor" on top of which runs the operating system(s),
    is a government requirement (for those sites that you don't hear
    much about).
    

> This may surprise some, but Digital has had this for years.
> A secure "hypervisor" on top of which runs the operating system(s),
> is a government requirement (for those sites that you don't hear
> much about).

  Digital *HAD* it. It consisted of special software and micro-
  code for the Nautilus/Polarstar family of VAXen. I knew folks
  who were working on it and I occasionally helped on the very
  outermost edges of the project.

  But we canned the project and most or all of the worker bees
  left or moved on.

  I don't believe we have any such product today.

                                   Atlant

    Normally announcing new notes conferences in the Digital notesfile is a
    no-no, but since FX!32 is uniquely important to sales, note that there
    is now an available-within-Digital new conference TALLIS::FX32.  This
    is not a restricted conference.