cclloyd@leland.Stanford.EDU (Charles Lloyd) (06/12/91)
(I am new to this newsgroup so forgive me if this is old news.) I heard today that NeXT will be releasing a Motorola 88K-based box later this year. Has anyone heard similar rumors? If so, what else can you tell me/us?
melling@cs.psu.edu (Michael D Mellinger) (06/12/91)
In article <1991Jun12.031302.13645@leland.Stanford.EDU> cclloyd@leland.Stanford.EDU (Charles Lloyd) writes:
(I am new to this newsgroup so forgive me if this is old news.)
I heard today that NeXT will be releasing a Motorola 88K-based box
later this year. Has anyone heard similar rumors? If so,
what else can you tell me/us?
As someone posted earlier today, the latest issue of InfoWorld has a
story on the cover about the new NeXT. It's supposed to contain
Moto's 96002 chip(successor to 560??) as well as the new superscaler
88K, which is rated at b/w 50 to 60 mips @40Hz. That's what the
magazine said but that doesn't quite seem right. A superscaler chip
ought to do more like 80-100 mips at 40MHz(I'm not complaining though :-)).
Anyway, it's not exactly a secret that NeXT is working on an 88K
machine is it? It should be available by this time next year for
sure.
-Mikecrum@alicudi.usc.edu (Gary L. Crum) (06/12/91)
As the InfoWorld article mentioned, an interesting and delicate issue
is architecture migration. Sun moved SunOS from Motorola 68k to SPARC
architecture and dropped 68k, and SiliconGraphics moved from Motorola
68k to MIPS and dropped 68k. HP is still supporting Motorola 68k
while selling HP-PA architecture. Would anyone like to speculate how
NeXT might make a transition, given NeXT's interest in the personal
computer low-cost shrink-wrapped style? Sounds fun to me...
I personally hope that NeXT adds operating system software support for
high-performance processors like the 88110 ("Hurricane") while
retaining an assumption that the main processor continues to be a
Motorola 68k processor (or at least, with the assumption that a 68k
processor is always somehow available to execute old applications and
so current NeXTstations won't be obsolete soon). So for example, the
88110 chips would be introduced this fall as an optional coprocessor
NeXTbus board. Hopefully the price of a NeXT computer with NeXTbus
slots, i.e. the NeXTcube or a new computer with NeXTbus slots, will be
reduced to be more attractive compared to NeXTstation slabs.
In such a scenario, the "hostinfo" command would report that the
kernel can support multiple CPUs, and the myraid of processor set
control system calls introduced in Release 2.0 would start to be
exercised. Given conventions for organizing binaries for different
architectures (which the very general Mach object file format
supports), the operating system could be setup to use 88110 CPUs if
available but revert to using 68k processors if not. That way, NeXT
software buyers would need not be concerned about what architecture
a given software package requires. The would start to be stickers saying
something like "uses 88110 if available".
Alas, NeXT didn't do that type of thing with the Intel i860 processor
on the NeXTdimension board. (That's not to say that such type of
processor support can't be added in a future release.) But, I don't
think development tools (e.g. compilers) for i860 processors are as
appropriate as current tools for 88k processors when it comes to
making a processor available for general purpose use.
Once heterogenous multiprocessing support is introduced and functional
in NeXT Mach, perhaps 3rd parties would be free to design boards using
other processors, and require little operating system change. Lots
of processor types are already defined in the NeXT Mach include file
/usr/include/sys/machine.h, including i386, MIPS, SPARC, HP-PA,
Motorola 88k , but not i860. Those machine type constant definitions
were inherited by NeXT Mach from the CMU release of Mach, however.
I think that Data General is currently selling a multiprocessor
configuration of its Aviion workstations which use the currently
available Motorola 88100 processor. I'm not sure what the
communication and memory architecture is like, though. I wonder if
NeXT can sell a board with two or four 88110 processors on it for less
than $5000. As for I/O bottlenecks, I think NeXT Mach (and UNIX
systems in general) can nicely support multiple I/O interfaces, such
as multiple SCSI interfaces. That's one area where the original
Macintosh and MS-DOS specifications fall very short -- they assume
that there is a small, fixed numbers of I/O ports (e.g. serial ports)
and applications were written to let the user choose only one of those
ports (e.g. two or four serial ports).
I'm having lots of fun learning things like mpadmin(1), sproc(2) and
sysmp(2) on a Silicon Graphics multiprocessor system, and I hope a lot
of NeXT users can start doing this type of thing soon. (Is anyone out
there using BBN or Encore multiprocessor systems running Mach?)
Garymcgredo@prism.cs.orst.edu (Don McGregor) (06/12/91)
>As someone posted earlier today, the latest issue of InfoWorld has a >story on the cover about the new NeXT. It's supposed to contain >Moto's 96002 chip(successor to 560??) as well as the new superscaler >88K, which is rated at b/w 50 to 60 mips @40Hz. That's what the I'd like to take credit for this turn of events. I just took delivery on my new NeXTStation today. The very same day rumors of a newer, faster, better machine come out. Never fails. :-) Don McGregor | Outside of a dog, a book is a man's best friend. mcgredo@prism.cs.orst.edu| Inside of a dog it's too dark to read.
sef@kithrup.COM (Sean Eric Fagan) (06/12/91)
In article <CRUM.91Jun11223911@alicudi.usc.edu> crum@alicudi.usc.edu (Gary L. Crum) writes: >I personally hope that NeXT adds operating system software support for >high-performance processors like the 88110 ("Hurricane") while >retaining an assumption that the main processor continues to be a >Motorola 68k processor (or at least, with the assumption that a 68k >processor is always somehow available to execute old applications and >so current NeXTstations won't be obsolete soon). If NeXT does this, past experience shows that, generally, nobody will use it. What is more likely, and somewhat easier from their standpoint, is to provide an emulator for the 68k; if they ever end up supporting heterogenous multiprocessing, the emulator can come into play only if the 68k-family processor doesn't exist. -- Sean Eric Fagan | "I made the universe, but please don't blame me for it; sef@kithrup.COM | I had a bellyache at the time." -----------------+ -- The Turtle (Stephen King, _It_) Any opinions expressed are my own, and generally unpopular with others.
anderson@dogie.macc.wisc.edu (Jess Anderson) (06/12/91)
In article <1991Jun12.062208.26369@lynx.CS.ORST.EDU> mcgredo@prism.cs.orst.edu (Don McGregor) writes: > I'd like to take credit for this turn of events. I just took delivery > on my new NeXTStation today. The very same day rumors of a newer, faster, > better machine come out. > > Never fails. :-) Heh heh. You gotta share, here. :-) But any fool who has anything that's much fun to do with computers knows they're getting on a treadmill. We should just admit that our motto is: stick it to me, I love it. After all, the alternative is not to get on, and while that would probably render one's life more manageable, certainly saner, it would be *too* dull. Yeah, that's the ticket! <> A mathematician is a machine for turning coffee into <> theorems. -- Paul Erdos -- Jess Anderson <> Madison Academic Computing Center <> University of Wisconsin Internet: anderson@macc.wisc.edu <-best, UUCP:{}!uwvax!macc.wisc.edu!anderson NeXTmail w/attachments: anderson@yak.macc.wisc.edu Bitnet: anderson@wiscmacc Room 3130 <> 1210 West Dayton Street / Madison WI 53706 <> Phone 608/262-5888
tomt%gtenext1.uucp@gte.com (Tom Tucker) (06/13/91)
In article <1991Jun12.104347.12409@macc.wisc.edu> anderson@dogie.macc.wisc.edu (Jess Anderson) writes: > In article <1991Jun12.062208.26369@lynx.CS.ORST.EDU> > mcgredo@prism.cs.orst.edu (Don McGregor) writes: > > > I'd like to take credit for this turn of events. I just took delivery > > on my new NeXTStation today. The very same day rumors of a newer, faster, > > better machine come out. > > > > Never fails. :-) > > Heh heh. You gotta share, here. :-) > > But any fool who has anything that's much fun to do with > computers knows they're getting on a treadmill. We should > just admit that our motto is: stick it to me, I love it. > After all, the alternative is not to get on, and while that > would probably render one's life more manageable, certainly > saner, it would be *too* dull. Yeah, that's the ticket! > > <> A mathematician is a machine for turning coffee into > <> theorems. -- Paul Erdos > -- > Jess Anderson <> Madison Academic Computing Center <> University of Wisconsin > Internet: anderson@macc.wisc.edu <-best, UUCP:{}!uwvax!macc.wisc.edu!anderson > NeXTmail w/attachments: anderson@yak.macc.wisc.edu Bitnet: anderson@wiscmacc > Room 3130 <> 1210 West Dayton Street / Madison WI 53706 <> Phone 608/262-5888 Motto for the 80's: He who dies with the most toys wins. Motto for the 90's: He who plays with the most toys wins. :-) -- Tom Tucker Systems Analyst, Systems Architecture GTE Directories Corp. tomt%gtenext1@gte.com GTE Place (NeXTMail Spoken here) West Airfield Drive, P.O. Box 619810 Voice: (214) 453-7033 D/FW Airport, TX 75261-9810 FAX: (214) 453-7573 -- Tom Tucker Systems Analyst, Systems Architecture GTE Directories Corp. tomt%gtenext1@gte.com GTE Place (NeXTMail Spoken here) West Airfield Drive, P.O. Box 619810 Voice: (214) 453-7033