SGI's first MIPS workstation

From: r. 'bear' stricklin <>
Date: Sat Jul 6 12:08:11 2002

I recall this was the subject of some debate earlier. I dug up a primary
source last night, and offer the following article from the Summer 1987
edition of IRIS Universe:

"announcing the IRIS 4D/60 turbo option"

In March, Silivon Graphics demonstrated its commitment to being the
undisputed leader in the superworkstation market by introducing the IRIS
4D/60, the industry's first RISC-based superworkstation. This new product
introduction generated more interest and excitement than any other single
event in Silicon Graphics' history. Certainly a major reason for the
industry's excitement is the IRIS 4D/60's premier graphics performance.
Equally important, however, is the new RISC and floating point technology
incorporated in the IRIS 4D/60. Users' response to this level of computing
performance from the RIS 4D/60 continues to be very enthusiastic.

Now, four months after the introduction of the IRIS 4D/60, Silicon
Graphics once again leads the industry by introducint the IRIS 4D/60 Turbo
Option, a high-performance computing upgrade for the IRIS 4D/60
Superworkstation. By integrating the latest in RISC and floating point
technoloyg, the IRIS 4D/60 Turbo Option breaks both the 10 MIPS and 1
MFLOPS performance barriers! While the IRIS 4D/60 is already the best
all-around performer in its price range, the integer and floating point
performance of the new Turbo Option make this upgrade truly phenomenal.

The IRIS 4D/60 Turbo Option incorporates a 12.5 MHz RISC CPU, 50 percent
faster than the 8 MHz CPU introduced with the IRIS 4D/60. This CPU is
tightly coupled with a new single-chip floating point coprocessor. By
using custom VLSI technology, the the [sic]Turbo Option increases the
floating point performance of the IRIS 4D/60 by a factor of three. The
Turbo Option's new CPU board also includes cache memory, up to 16 Mbytes
of on board CPU memory, and an interface to SCSI peripherals. By having
all of the major components of the CPU architecture on a single board, the
rate of data transfer between components is significantly improved,
thereby maximizing system performance.

The IRIS 4F/60 Turbo Option employs a dual-cache architecture
incorporating 64 Kbytes of instruction cache plus 32 Kbytes of data cache.
The dual cache architecture increases system performance byholding large
portions of the application program in efficient cache memory, and ensures
that instructions and data will not compete for the same cache locations.
This architecture improves performance of large applications by reducint
the time required to access instructions and data.

How fast is fast? Using the Dhrystone benchmark, the industry standard for
measuring integer computing performance, an IRIS 4D/60 with the Turbo
Option leads the superworkstation pack at 10 MIPS (ten times a VAX
11/780). Using the Linpack benchmark to measure floating point
performance, the Turbo Option performs 1.1 million double precision
floating point operations per second (1.1 MFLOPS).

In addition to theis significatn boost in computing performance, Silicon
Graphics announces support for advanced lighting models on the IRIS 4D/60,
with or without the Turbo Option. Both infinite and local light sources
are supported, including attenuation of local light sources with distance
from the source. Diffuse and specular (Phong) lighting are also supported,
and up to eight colored light sources may be modeled simultaneously.

The Turbo Option may be purchased as a field installable upgrade for the
IRIS 4D/60. Please contact your local sales representative.

typos solely my responsibility.
Received on Sat Jul 06 2002 - 12:08:11 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:35:01 BST