SCO's 64 bit Gemini with Merced will topple WinNT

SCO's 64 bit Gemini with Merced will topple WinNT

Post by s.. » Sat, 07 Dec 1996 04:00:00



   Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
   long-term commitment to SCO UnixWare as the industry-standard,
   enterprise-class UNIX operating system for volume enterprise
   servers based on Intel architecture-based processors.
   Intel, the OEMs and SCO will align their Enterprise ISV recruitment
   efforts for SCO UnixWare, starting with SCO UnixWare 2.1. This
   includes special business and technical support programs for ISVs
   to assist them in porting their business critical applications to
   SCO UnixWare.
   Key ISVs have already endorsed this agreement to standardize on SCO
   UnixWare 2.1 which will support single binary versions of their
   application across multiple OEM enterprise hardware platforms.
   SCO, Intel and the OEMs are working together to bring additional
   enterprise class capabilities to SCO UnixWare.  SCO UnixWare
   2.1 is also the first step toward SCO's project Gemini (merger of SCO
   UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
   UNIX technology (announced by SCO and HP) for the forthcoming Merced
   64-bit architecture.

   ?1996 The Santa Cruz Operation, Inc. All Rights Reserved.

    http://www.sco.com
     _________________________________________________________________

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Sean T. O'Conn » Sat, 07 Dec 1996 04:00:00


NT's simply doomed.

--
-Sean T. O1Connor

________________________________________
"Everybody knows that there's a liberal, that there's a heavy liberal persuasion among correspondents."
-- Walter Cronkite at the Radio and TV Correspondents Association dinner, March 21, 1996.

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by John Nag » Sun, 08 Dec 1996 04:00:00



>   Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
>   long-term commitment to SCO UnixWare as the industry-standard,
>   enterprise-class UNIX operating system for volume enterprise
>   servers based on Intel architecture-based processors.

     Except for Compaq (and SNI. whoever they are), that's a list of
real losers.  Those are all mainframe and mini companies that blew
the transition to the microprocessor era.

                                                John Nagle

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by David E. Merchan » Mon, 09 Dec 1996 04:00:00



>    Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
>    long-term commitment to SCO UnixWare as the industry-standard,
>    enterprise-class UNIX operating system for volume enterprise
>    servers based on Intel architecture-based processors.
>    Intel, the OEMs and SCO will align their Enterprise ISV recruitment
>    efforts for SCO UnixWare, starting with SCO UnixWare 2.1. This
>    includes special business and technical support programs for ISVs
>    to assist them in porting their business critical applications to
>    SCO UnixWare.
>    Key ISVs have already endorsed this agreement to standardize on SCO
>    UnixWare 2.1 which will support single binary versions of their
>    application across multiple OEM enterprise hardware platforms.
>    SCO, Intel and the OEMs are working together to bring additional
>    enterprise class capabilities to SCO UnixWare.  SCO UnixWare
>    2.1 is also the first step toward SCO's project Gemini (merger of SCO
>    UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
>    UNIX technology (announced by SCO and HP) for the forthcoming Merced
>    64-bit architecture.

>    ?1996 The Santa Cruz Operation, Inc. All Rights Reserved.

>     http://www.sco.com
>      _________________________________________________________________

Huh?  What'd he say?

In their brief, SCO says, to misquote, "if you programmers will use our
API's, your code will be compatible with our API's."

If I remember from school a couple of centuries ago, a 64-bit OS on a
Pentium class machine will incur MUCH more memory and cycles overhead
than any efficiencies gained.  In fact, the only advantage to a 64-bit
OS in the Pentium class machine is that a float can be loaded in a
single cycle (which ignores long doubles).  If the OS developer can
utilize a variable-size opcode strategy, which is EXTREMELY difficult,
then the 64-bit OS will be optimal.

Just my $0.00.

David

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by rbagw.. » Tue, 10 Dec 1996 04:00:00




>>    Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
>>    long-term commitment to SCO UnixWare as the industry-standard,
>>    enterprise-class UNIX operating system for volume enterprise
>>    servers based on Intel architecture-based processors.
>>    Intel, the OEMs and SCO will align their Enterprise ISV recruitment
>>    efforts for SCO UnixWare, starting with SCO UnixWare 2.1. This
>>    includes special business and technical support programs for ISVs
>>    to assist them in porting their business critical applications to
>>    SCO UnixWare.
>>    Key ISVs have already endorsed this agreement to standardize on SCO
>>    UnixWare 2.1 which will support single binary versions of their
>>    application across multiple OEM enterprise hardware platforms.
>>    SCO, Intel and the OEMs are working together to bring additional
>>    enterprise class capabilities to SCO UnixWare.  SCO UnixWare
>>    2.1 is also the first step toward SCO's project Gemini (merger of SCO
>>    UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
>>    UNIX technology (announced by SCO and HP) for the forthcoming Merced
>>    64-bit architecture.

>>    ?1996 The Santa Cruz Operation, Inc. All Rights Reserved.

>>     http://www.sco.com
>>      _________________________________________________________________

>Huh?  What'd he say?

>In their brief, SCO says, to misquote, "if you programmers will use our
>API's, your code will be compatible with our API's."

>If I remember from school a couple of centuries ago, a 64-bit OS on a
>Pentium class machine will incur MUCH more memory and cycles overhead
>than any efficiencies gained.  In fact, the only advantage to a 64-bit
>OS in the Pentium class machine is that a float can be loaded in a
>single cycle (which ignores long doubles).  If the OS developer can
>utilize a variable-size opcode strategy, which is EXTREMELY difficult,
>then the 64-bit OS will be optimal.

He isn't talking about a 64-bit OS on a Pentium class machine.

Quote:>Just my $0.00.

>David

///////////////////////////////////////
//
//  Randy Bagwell

//
//
 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Bryan Seigneu » Tue, 10 Dec 1996 04:00:00



> NT's simply doomed.

SCO could have killed MS 12 years ago.
Why didn't they?  They're either losers or
the MS ownership of them is too much.
(What is it? 15%?)  They can't do it now for
the same reason (whichever it is) they didn't
do it long ago.  The effort will die a *y
(lots of casualties, messy, inefficient) death.

I suggest all SCO supporters take a long
hard look at their position.  Think about
the long term, folks.  It can't be only MS
that thinks about the longest term.

Find out which alternative has the best position.  
I think Linux looks pretty good.  Plenty of
commercial support now, and it's had a large
base of users and developers for quite a while
now.  I think that's enough.

> --
> -Sean T. O1Connor


--
Bryan Seigneur
  / / inux  
 / /__ink  ==> http://www.veryComputer.com/~bry/linmarks.html/____/ist
 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Carrer Yuri 341353/ » Tue, 10 Dec 1996 04:00:00





>>    Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a

>If I remember from school a couple of centuries ago, a 64-bit OS on a
>Pentium class machine will incur MUCH more memory and cycles overhead
>than any efficiencies gained.  In fact, the only advantage to a 64-bit
>OS in the Pentium class machine is that a float can be loaded in a
>single cycle (which ignores long doubles).  If the OS developer can
>utilize a variable-size opcode strategy, which is EXTREMELY difficult,
>then the 64-bit OS will be optimal.

>Just my $0.00.

 I agree. They should choose an Alpha :-)

--
   ol     Carrer Yuri   Pubblicita': usa \documentstyle{athenaslide}  
   |--    Tel: 0421/211239 *new*     e non usare Power Point :-)
  000 *** WARNING ***   http://www.dei.unipd.it/~carrery/
http://www.geocities.com/TheTropics/1797/  http://www.alfa.it/ :) *new!*

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Ketil Z Mald » Tue, 10 Dec 1996 04:00:00



Quote:> If I remember from school a couple of centuries ago, a 64-bit OS on a
> Pentium class machine will incur MUCH more memory and cycles overhead

Uh...how you run 64 bit OS on only 32 bit machine?  Me no grok!

~kzm

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Justin Co » Tue, 10 Dec 1996 04:00:00



: >
: >    Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
: >    long-term commitment to SCO UnixWare as the industry-standard,
: >    enterprise-class UNIX operating system for volume enterprise
: >    servers based on Intel architecture-based processors.
: >    Intel, the OEMs and SCO will align their Enterprise ISV recruitment
: >    efforts for SCO UnixWare, starting with SCO UnixWare 2.1. This
: >    includes special business and technical support programs for ISVs
: >    to assist them in porting their business critical applications to
: >    SCO UnixWare.
: >    Key ISVs have already endorsed this agreement to standardize on SCO
: >    UnixWare 2.1 which will support single binary versions of their
: >    application across multiple OEM enterprise hardware platforms.
: >    SCO, Intel and the OEMs are working together to bring additional
: >    enterprise class capabilities to SCO UnixWare.  SCO UnixWare
: >    2.1 is also the first step toward SCO's project Gemini (merger of SCO
: >    UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
: >    UNIX technology (announced by SCO and HP) for the forthcoming Merced
: >    64-bit architecture.
: >
: >
: >    ?1996 The Santa Cruz Operation, Inc. All Rights Reserved.
: >
: >     http://www.sco.com
: >      _________________________________________________________________
:
:
: Huh?  What'd he say?
:
: In their brief, SCO says, to misquote, "if you programmers will use our
: API's, your code will be compatible with our API's."
:
: If I remember from school a couple of centuries ago, a 64-bit OS on a
: Pentium class machine will incur MUCH more memory and cycles overhead
: than any efficiencies gained.  In fact, the only advantage to a 64-bit
: OS in the Pentium class machine is that a float can be loaded in a
: single cycle (which ignores long doubles).  If the OS developer can
: utilize a variable-size opcode strategy, which is EXTREMELY difficult,
: then the 64-bit OS will be optimal.

I Quoted the entire thing on purpose..

umm, Dave, where, in that ENTIRE press release, does it mention it's
destination being on PENTIUM based machines?

to quote more directly: "...next generation 3da UNIX technology... for the
forthcoming Merced 64 bit architecture"

Intel, yes, Pentium, no.

--
__________________________________________________________
          _____       __        __ _____  __  __ ___ ___
    ||  ||_  | ||\ | |_ |  |  ||  | | |  |   |  ||__||__|
 |__||__|__| | || \| |__|__|__||__| | |  |__ |__||__||__|
==========================================================
  - - Homepage: http://www.netsync.net/users/jecobb  - -

__________________________________________________________

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Ling Wan » Tue, 10 Dec 1996 04:00:00



>    Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
>    long-term commitment to SCO UnixWare as the industry-standard,
>    enterprise-class UNIX operating system for volume enterprise
>    servers based on Intel architecture-based processors.
>    Intel, the OEMs and SCO will align their Enterprise ISV recruitment
>    efforts for SCO UnixWare, starting with SCO UnixWare 2.1. This
>    includes special business and technical support programs for ISVs
>    to assist them in porting their business critical applications to
>    SCO UnixWare.
>    Key ISVs have already endorsed this agreement to standardize on SCO
>    UnixWare 2.1 which will support single binary versions of their
>    application across multiple OEM enterprise hardware platforms.
>    SCO, Intel and the OEMs are working together to bring additional
>    enterprise class capabilities to SCO UnixWare.  SCO UnixWare
>    2.1 is also the first step toward SCO's project Gemini (merger of SCO
>    UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
>    UNIX technology (announced by SCO and HP) for the forthcoming Merced
>    64-bit architecture.

But whats the point, every Intel box that boots up will
have the M$ logo pop up as the first thing(As per their licensing
agreement).  No matter what OS you are running, the users will
thing it is M$ NT.
 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Joe Ragos » Wed, 11 Dec 1996 04:00:00




Quote:> But whats the point, every Intel box that boots up will
> have the M$ logo pop up as the first thing(As per their licensing
> agreement).  No matter what OS you are running, the users will
> thing it is M$ NT.

Very clever.

Microsoft refuses to pay for tech support on their OS, but insists that the
vendor allow their splash screen to show unattended.

I enjoyed the MS spokesperson's quote in InfoWorld. He stated that they had
to put that into their licensing agreement because the boot process is so
fragile that it must not be interrupted for any reason.

--
Regards,

Joe Ragosta

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Steven Gavet » Wed, 11 Dec 1996 04:00:00



Dec 1996 08:24:29 -0500, Joe Ragosta regaled us with...



>> But whats the point, every Intel box that boots up will
>> have the M$ logo pop up as the first thing(As per their licensing
>> agreement).  No matter what OS you are running, the users will
>> thing it is M$ NT.

>Very clever.

This doesn't prevent the use of a boot manager. It simply means that once the
selected OS has started the boot process, it shouldn't be interrupted.

Quote:

>Microsoft refuses to pay for tech support on their OS, but insists that the
>vendor allow their splash screen to show unattended.

MS does no such thing. They will be happy to offer free support. *IF* the
*OEM* loads a retail version of the product. *IF* the *OEM* chooses to use the
discounted OEM version, the OEM agrees to provide support.

Steve

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by s.. » Wed, 11 Dec 1996 04:00:00


David E. Merchant <dmerc...@mail.HiWAAY.net> wrote:

>s...@cyberspace.com wrote:
[snip]
>>    SCO UnixWare
>>    2.1 is also the first step toward SCO's project Gemini (merger of SCO
>>    UnixWare and SCO OpenServer) and the forthcoming next-generation 3DA
>>    UNIX technology (announced by SCO and HP) for the forthcoming Merced
>>    64-bit architecture.

>Huh?  What'd he say?
>In their brief, SCO says, to misquote, "if you programmers will use our
>API's, your code will be compatible with our API's."

>If I remember from school a couple of centuries ago, a 64-bit OS on a
>Pentium class machine will incur MUCH more memory and cycles overhead
>than any efficiencies gained.  In fact, the only advantage to a 64-bit
>OS in the Pentium class machine is that a float can be loaded in a
>single cycle (which ignores long doubles).  If the OS developer can
>utilize a variable-size opcode strategy, which is EXTREMELY difficult,
>then the 64-bit OS will be optimal.

One this one, I would not bet against the top engineers at HP and Intel.
The following article should help make my point a little more clear:

Hewlett-Packard: What Is It Doing?

   UNIX Review, June 1996

   Has Hewlett-Packard lost its collective mind? It seems like HP is
   abandoning both its highly successful PA-RISC microprocessor family
   and its HP-UX variant of UNIX. What is happening? HP is one of the
   most successful vendors in UNIX. The classic Silicon Valley garage
   start-up was a reasonably sized minicomputer vendor when it decided
   that RISC and UNIX represented the future. HP turned foresight, good
   engineering, and superb market strategy into a leadership position in
   the UNIX market (analysts place HP first in revenue and second in unit
   volume).

   In the workstation market, the generally high performance of the
   PA-RISC microprocessor and the workstations HP built around it have
   allowed HP to challenge Sun's historical dominance in that category a
   feat no other company has matched.

   HP was one of the earliest vendors to recognize the great opportunity
   in the business world for UNIX-based systems. It worked hard to get
   business applications (such as accounting, inventory, and order entry)
   onto its systems and often had the first port from vendors of large
   applications. Its HP-UX evolved to handle business users' needs for
   reliability, manageability, and connectivity. HP's "mainframe
   alternative" marketing program was often a catalyst in a mainframe
   user's decision to downsize.

  An Enviable Record, Then. . .

   In late 1994, Intel and HP announced an agreement to design and
   develop jointly a next-generation microprocessor that would be
   compatible with both the Intel x86 and the HP PA-RISC instruction
   sets. The resultant chip would be manufactured by Intel. Thus, HP
   declared its plans to discontinue the evolution of the PA-RISC, or at
   least steer its development in a different direction.

   Last year, Novell, SCO, and HP announced that Novell would transfer
   UNIX to SCO, and SCO and HP would lead the charge toward 64-bit UNIX
   (see this column in the January 1996 issue, p. 9). This HP-SCO effort
   would lead to a merged operating system available from both vendors.

  What Is Going On?

   HP seems to be crazy like a fox. On the chip front, estimates show
   that each new-generation microprocessor design costs more than $100
   million and an up-to-date fabrication plant costs more than $1
   billion. Even if a company can afford the up-front costs, these costs
   must be amortized over each chip. If a million chips of that
   generation are produced, the design costs alone would add more than
   $100 per chip. The cost of chips these days is influenced as much by
   the start-up costs as by the size of the die and the technology used.

   HP decided the competitive benefits of raw chip technology were not
   worth the cost, and its deal with Intel is designed to provide HP with
   a supply of relatively inexpensive yet high-performance
   microprocessors. Intel, of course, has both the production technology
   and the chip volume to make it the envy of the industry.

   Intel's latest announcements introduce its Merced family of
   microprocessors. Merced is a 64-bit architecture that will be
   compatible with both the Intel x86 and the HP PA-RISC instruction sets
   and will be "very fast." Competitors claim that any microprocessor
   that tries to be such a hybrid will be neither as fast as their
   contemporaneous RISC processor designs nor as inexpensive as a
   straightforward x86-architecture processor. While no one will say when
   it will appear, the Merced family is expected to become available
   initially in late 1997 or early 1998, and we will just have to wait
   for actual silicon to see who is right.

   On the operating-system front, SCO and HP have announced a joint
   architecture and development effort for their next-generation UNIX
   technology and a business model for a set of partnerships built around
   this technology.

   The joint effort has both technical and political aspects. On the
   technical front, the team announced several already-available
   capabilties, such as enhanced fault tolerance, built-in LAN and WAN
   networking, support for symmetric multiprocessing (SMP) and clusters,
   and a 64-bit-enabled system. HP is saying clearly that customers no
   longer care which kernel is used and that they will use a
   best-of-breed approach (a signal that they intend to keep some
   SVR3-based parts of HP-UX and SCO OpenServer).

   One new idea is plug-and-play kernel modules. If implemented, this
   technology could provide better fault isolation, system upgrades
   invisible to the user, and an easier way for system vendors and ISVs
   to customize the basic system without creating incompatibilities
   visible at the user or application levels.

   On the political side, several other system vendors have endorsed the
   effort; these vendors might become OEMs of SCO's version of this
   system. The endorsers include low-end Intel-based vendors one would
   expect (who previously sold SCO OpenServer or UnixWare), high-end
   Intel-based vendors (including NCR, Sequent, and Unisys), and others
   such as the membership of the Mips ABI Group (including NEC, Siemens
   Nixdorf/Pyramid, Silicon Graphics, Sony, and Tandem). Why would HP
   want these competitors to join in, and why would these companies want
   to?

   For potential OEMs, the cost of keeping an operating system up-to-date
   and competitive has been estimated at about $100 million per year, and
   it is becoming clear that basic operating-system functionality does
   not provide much competitive benefit. In fact, users would rather have
   standardization at this level. The areas for competitive
   differentiation continue to move upscale and currently are related to
   system management, large-scale connectivity, cluster/MPP
   architectures, and general reliability, availability, and
   serviceability (RAS) features. Thus, the OEMs would prefer to put
   their development dollars into these areas rather than just keep up
   with the changing base functionality.

   The potential high-end OEMs, however, seem somewhat wary. They say
   they will keep their current variants of UNIX for the near term and
   will look for better application compatibility in the future. They are
   watching HP-SCO developments closely; some are involved in creating
   the new UNIX and certainly will consider making their operating
   systems compatible (if not simply converting to the merged core
   operating system) at some point.

   NCR's partnership with SCO should ensure that NCR's current RAS
   capabilities and middleware software will not be compromised. However,
   this merged operating system represents NCR's strategy for the 64-bit
   future. The Mips ABI Group members have established a "Golden Gate
   Operation" to work jointly on development and communication with the
   HP-SCO team but have left decisions about adopting the technology to
   individual members. Sequent says it always looks for reliable sources
   of technology and strongly supports SCO's "role as a
   level-playing-field source of UNIX technology." (In other words,
   Sequent really does not want to adopt the operating system but might
   license some pieces.) Expect all these vendors to move to the HP-SCO
   core technology not in two to three years but certainly in five to
   eight years.

   Other vendors also endorsed the effort, but only for their Intel-based
   systems. Included in this category are DEC, ICL, and IBM. DEC remains
   committed to Digital UNIX (which it claims is already two years ahead
   of the HP-SCO system) for Alpha-based systems. ICL will put the new
   system on its Intel-based computers (along with its own operating
   system) but will have only its own system on its SPARC-based machines.
   IBM remains fully committed to AIX for its RS/6000 family.

   HP's motivation is more interesting. In industry-analyst circles,
   there is a notion of computer-industry ecosystems, that is, a set of
   companies whose success in the same area depends on everyone helping
   each other. For example, Microsoft's Visual Basic (VB) is helped
   enormously by the existence and energy of the companies that build OCX
   components and add-on VB tools. HP strategists have decided that
   creating such an ecosystem is better in the long run than any
   competitive advantage they might get from going it alone. The view is
   that HP really does not compete on the basis of raw chip technology
   nor on the basis of operating-system fundamentals. It competes with
   customer relationships, total system solutions and services, and
   generally having the right partnerships to satisfy the customer's
   needs. This is a powerful view and one that
...

read more »

 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Raymond N Shwa » Thu, 12 Dec 1996 04:00:00




>>Microsoft refuses to pay for tech support on their OS, but insists that the
>>vendor allow their splash screen to show unattended.
>MS does no such thing. They will be happy to offer free support. *IF* the
>*OEM* loads a retail version of the product. *IF* the *OEM* chooses to use the
>discounted OEM version, the OEM agrees to provide support.

        Get real. Ever tried to get support for the shrink-wrap (retail)
version of Windows NT (workstation or server) from Microsoft? ZERO support.
Ever tried to get support for network-related issues for Worgroups or
Win95? ZERO support! ...unless, of course, you've got a support contract or
are willing to shell out for the support call.
 
 
 

SCO's 64 bit Gemini with Merced will topple WinNT

Post by Daniel C. Benton Jr » Thu, 12 Dec 1996 04:00:00




> >   Compaq, Data General, ICL, NCR, Olivetti, SNI and Unisys have made a
> >   long-term commitment to SCO UnixWare as the industry-standard,
> >   enterprise-class UNIX operating system for volume enterprise
> >   servers based on Intel architecture-based processors.

>      Except for Compaq (and SNI. whoever they are), that's a list of
> real losers.  Those are all mainframe and mini companies that blew
> the transition to the microprocessor era.

>                                                 John Nagle

Well, the SNI company listed is *NOT* SuperNet, Inc. (sni.net). We
have Solaris, Irix, Linux, BSDI, and HPUX systems, but not a
single SCO box.

--
Daniel C. Benton Jr.

http://www.sni.net/dcbenton
--
"Any technology sufficiently advanced is indistinguishable
from a Perl script."   -  Larry Wall

 
 
 

1. News Sun's 64-bit Gemini Chip

I think this is a good idea, but I would like sun to also introduce a CICS
chip to compete with Pentium. Most consumer just look at the clock speed
anymore and don't really understand that RISC is superior chip design.

-------

News Sun's 64-bit Gemini Chip
Chip Details Released 2003-08-25

Back To Top Current Issue Fullsource Subscribe
At the recent Hot Chips conference, Sun released the details of its new
64-bit processor designed for blade and thin rack-mount Web and Application
servers. The Gemini chip is the company's first generation CMT (Chip
Multi-threading) UltraSPARCR processor and is capable of running dual
threads. The chip runs at about 1.2 GHz in a normal mode and combines two
4-issue superscalar UltraSPARC cores with a 1 MB integrated L2 cache, a DDR1
memory controller and a JBus system bus interface.

The new solution is based on the SPARCR V9 64-bit ISA (Instruction Set
Architecture) and is part of Sun's Throughput computing initiative along
with the next-generation UltraSPARC chip code-named Niagara. Gemini will be
pin-controllable with its UltraSPARC IIIi processor and Sun estimates a
32-watt power dissipation for the processor. The processor will be binary
compatible with previous SPARC processor generations, and supported by the
SolarisTM Operating System (Solaris OS). Texas Instruments is manufacturing
the chip for Sun, and the company expects the chip to debut in products in
early 2004.

"Based on internal projections on throughput-oriented metrics, Gemini is
expected to provide 3x the throughput compared to a UltraSPARCR IIIi
processor running at 1 GHz, and a 6x throughput boost relative to the
UltraSPARCR IIi 650 MHz processor, the currently shipping UltraSPARC
processor for blades," Sun said in a statement.

2. Help with configuring ...

3. Linux on 64-bit merced?

4. LXNY LUNY NYLUG New School Install Fest Wednesday 28 July 1999

5. Intel's IA-64 processor (Merced) and Linux

6. : RTF Editor ?!

7. 64 bit Unix (esp. 64 bit IO)

8. driver hp1100?

9. host mastered 64 bit wide transfers to 64 bit PCI slot?

10. performance of runing 32/64 bit program in 64 bit kernel.

11. HALstation (64-bit processor running 64-bit Solaris) as webserver

12. Is 64-bit Linux "true" 64 bit thru-and-thru??

13. IBM announces 64-bit mainframes and 64-bit Linux for S/390