Knowing something about the processor in your
computer can tell you something useful about what it can do, relative to other
Just as you don't have have to be an engineer to appreciate
that engine size, number of valves and the type of fuel (or battery) give you
clues about the capability of your automobile, in the same way SPARC processor
names tell you something about the speed or vintage of your computer system.
We know there's confusion about what some of the words connected with "SPARC"
actually mean. That's partly because you, as a user, sometimes have to guess the
meaning from its context in vendor literature, and also because vendors
themselves don't always use the terms consistently. This article is divided
into the following sections, which you can read in order if you're interested in
following the flow. Alternatively please click to go straight to the subject
which interests you.
SPARC versus X86. Why is SPARC nomenclature so
Most people are familiar with the numbering systems for Intel architecture
processors, so why is there more confusion in the relatively smaller world of
Probably for the following reasons
- There is no single dominant supplier of SPARC CPU's. Historically,
the major SPARC chip manufacturer in any system generation may not have been the
same company that was dominant in the earlier one. That's because Sun
Microsystems, the largest consumer of SPARC chips, runs a fabless
business model. That's not a typo, but means that Sun
does not own the semiconductor fabrication plants which manufacture their
processor chips. Although Sun designs many (but not all) of the chips
used in its computers, the Sun designed chips are actually manufactured by other
leading semiconductor companies. For simplicity in this article we're treating
all Sun designed chips as being made by Sun Microelectronics (SME). This part of
Sun was previously called SPARC Technology Business (STB). Sun Microelectronics
also sells SPARC chips and motherboards to other systems oem's.
- SPARC names give you clues about the architecture of the computer they
were designed to work in. Some names tell you that the CPU was optimized for
use in single processor, or multi-processor configurations. In the Intel
computer world, this is not yet a common distinction. Intel processors are
mainly designed for the high volume single processor PC environment, and they
get deployed in multi-processor systems by the use of systems interfaces and
logic which varies according to system manufacturer.
- SPARC chips are marketed more by name than by number, and the name of
each SPARC chip is uniquely assigned to a single vendor. Imagine the confusion there would be in the Intel world if
processors were not labelled by numbers (286, 386, 486 etc) but each had
different names. OK since the Pentium, Intel has stopped
using numbers, but Intel's competitors still use numbers like 586, 686 which
keeps comparisons simple. SPARC vendors add to this
confusion by making comparisons with competing products which may be true at
a single point in time, but are not universally valid. Because words can be
associated with fuzzier concepts than numbers, and it's possible to establish a
brand with a name (but not easily with just a number) marketers make all the
capital they can from these differences.
SuperSPARC is sometimes faster than HyperSPARC and vice versa. It
depends on the clock speeds you use for the comparisons.
microSPARC systems are sometimes faster than SuperSPARC systems. A
110MHz microSPARC single processor workstation (such as Aries Research
Marixx110) will be faster than a 33MHz SuperSPARC based MBus workstation such as
an early model of Sun'sSPARCstation 10. The age of the machine and the clock
speeds are significant factors.
TurboSPARC processors are always faster than microSPARC (because the
TurboSPARC is a go-faster competing X2 upgrade for microSPARC sockets, but
depending on the clock speed it can be slower or faster than HyperSPARC.
- Sun Microsystems also develops a range of Java chips which are intended
for low cost products which have an embedded Java virtual machine. So not
every processor from Sun carries a SPARC label. But
the JavaStation webtop terminal announced by Sun (October 1996) actually does
include a low end 100MHz microSPARC processor.
If you weren't confused before, then maybe you are now. We hope the rest of
this article will help you get a clearer picture.
SPARC... What does it mean?
The main things you need to know as a systems specifier are:-
- "SPARC" was originally an acronym for "Scalable
Processor ARChitecture". The name has lived up to its promise. SPARC
processors can be found today in computer systems which range from portables
(such as Tadpole's
upto supercomputers (such as Sun's
Just as significant from a user's point of view is that all these SPARC
computers run a single operating system Solaris 2.x. So if your application, or
your organization grows it can still be be run viably on a different size
computer without necessarily needing a complete rewrite. Competitors might say the same applies for Unix generally, and
also for Microsoft's NT. The difference between Solaris and other
types of Unix is that Solaris provides binary compatibility for your application
whether it's running on your laptop, or on the datacenter server. The main
differences between Solaris and NT, are that the performance ceiling for Solaris
is higher, and that it's been running in the market longer which reduces the
risk for mission critical applications.
- SPARC is based on an architecture originally developed by Sun
Microsystems. SPARC based computers have been around for over 10 years
(refer to our SPARC
History article if you'd like to see a chronology). Many people who who see
SPARC Product Directory for
the first time say they didn't know that so many companies make SPARC based
- All SPARC processors are RISC type. But not all RISC processors are
We know there's confusion on this point. SPARC is simply a specific type of
RISC processor, just as Intel's Pentium is a specific type of CISC processor.
Other types of RISC processor include DEC's Alpha, and Silicon Graphics
MIPS. Other types of CISC processor include Motorola's 680X0, and the Zilog Z80.
- Many chip manufacturers make SPARC processors. Many computer
systems companies build SPARC based systems. An independent organization called
Inc regulates the licensing of all the SPARC trademarks, and works to ensure
that products from different sources are compatible.
- Has SPARC got a future? As the most commercially successful 64 bit
processor family, we obviously think it's got a bright future.
The table below provides an overview of the popular SPARC processor names
which you are likely to see as a user or specifier of SPARC based computers.
Note:- the list above is intended to be helpful to users of SPARC based
computers which run the Solaris operating system. There are other types of SPARC
CPU which are less commonly seen in this systems environment.
Dates of introduction and current market status.
Another way of looking at these processors, is by the date of introduction
in a commercial system. The tables below shows this information, along with the
first commercial system (according to our records) in which they appeared.
first market appearance in end- user systems and model in which introduced
Current market status*.
Q1 94 as user installable upgrade for SPARCserver 600MP
Sun dropped this CPU from its systems product line when it introduced the
UltraSPARC based systems in Q4 95.
continued the development in terms of speed and performance. These processors
will probably be viable as upgrades for old systems. Some companies such as
Association were still marketing new workstations based on this technology
in Q1 99.
Q4 94 SPARCstation 20
Q4 92 SPARCstation LC "Classic"
Sun dropped the microSPARC from its workstation product line, but used it in
the original JavaStation terminal.
Q4 95 HALstations
This is still a current product, although its future development hinges on
how much volume can be generated by
HAL Computer and
Fujitsu-ICL. However, binary compatibility means that HAL's customers would not
be affected if they or HAL switched to another 64 bit SPARC processor at any
time in the future. Note:- since our original comment 2 years ago SPARC64 has
failed to achieve significant volume in the systems market, and the development
is being underwritten by Fujitsu as a technology process driver.
SPARC power micro-P
Q3 93 as user installable upgrade for SPARCstation 2 and IPX
Obsolete. However, some resellers may still have some of these left, and
it's the lowest cost upgrade for your SPARCstation 2 if you're still using one.
Q2 92 SPARCstation 10
Sun has dropped this from its workstation and server product line, but it
still appears in some embedded board level products from other vendors. Sun's
positioning appears to be that power users should use its 64 bit workstations.
Owners of SuperSPARC based systems have an alternative chip level upgrade path
from hyperSPARC, which has never looked better.
Q2 97 FORCE COMPUTERS CPU-7V
Sun reintroduced its SPARCstation 5 which was given a new lease of life by
TurboSPARC. This is an upgrade option for microSPARC based models. It still
appears in some current systems, such as
Resilient, and the SPARCbook 3 from
Q4 95 Sun Ultra 1 and 2
This is the flagship processor line used in most of Sun's current
workstations and servers. Evolutionary improvements including increasing clock
speed, more cache, and higher levels of integration to include strategic
interfaces will continue for many years.
first market appearance in end- user systems and model in which introduced
Current market status*.
* Current market status - 1996. This is the editor's opinion as an analyst
of this market and is not a definitive fact. Other people may look at the same
data and reach different conclusions.
This article was first published in the 1996 edition of the
SPARC Product Directory.