Michael Jones (mtj++at++isdn-celeste.corp.sgi.com)
Sat, 27 Apr 1996 11:37:16 -0700
Michael Jones
IRIS PERFORMER 2.1
As Sharon explained yesterday, IRIS Performer 2.1 has released and is
now available. Those who noticed that the 2.0 release tool a bit longer
than planned might be surprised that 2.1 was done so soon. There are
significant new features for InfiniteReality users that will amaze
you. This version of Performer (which is the partner of
InfiniteReality) supports the creation of the most sophisticated
applications in the history of visual simulation.
SIMULATION CREDIT AND HISTORY
Speaking of the history of visual simulation, General Electric Co.
invented the computer graphic image generator (IG) in the 1960's and
NASA was the deep-pockets customer sponsoring the work. I have a great
deal of respect for those early GE engineers and want to make sure
they're credited where credit is due. In fact, we've made a point of
listing the contributions of these pioneers (and the work at Evans &
Sutherland, Link/Sunnyvale, Thomson, etc.) in our SIGGRAPH course
"Real-Time Graphics for Entertainment" for the past two years.
What SGI did was to find a way to use aggressive technology, rapid
product cycles, clever approaches, and a commercial business model
(rather than the military contractor style) to reinvent the image
generator industry. Our machines, such as the InfiniteReality, are
faster, more capable, and less expensive by far (in all three areas!)
than those of the Ancien Regime. A trip to a visual simulation trade
show such as the recent ITEC show in Holland, IMAGE, or I/ITSEC makes
the point very clearly: the tables have turned in visual simulation
and it's the early pioneers that are now trying to catch-up. If you
were at ITEC and disagree (or agree) please say so.
COMPETITIVE WORKSTATIONS
Now, as far as HP, SUN, DEC, and IBM, are concerned, it's a very
different story. None have products that pose a challenge to SGI,
E&S, GE (sold to Loral, then Martin, then Lockheed), CAE, or any
other visual simulation provider. Perhaps you might argue that some
non-SGI workstation (or high-end PC) is good enough for your database
modeling needs, but there is no chance of a SparcStation (or Windows
NT Intergraph) providing the visual simulation needs of the F-117
stealth fighter trainer, the German ICE train simulator, or thousands
of the other demanding Onyx/RE/Performer applications.
This is not arrogance. All of these companies have capable engineers
who may be building *future* products to compete in these spaces. Time
will tell (but the same is true for SGI -- companies who aim directly
at our moving target will miss.)
IMAGE GENERATOR BENCHMARKS
John Archdeacon of Gemini Technologies sent out email describing the
portability and openness of his product, OpenGVS. Most interestingly,
he discussed a benchmark suite his company has developed. I think
that this is a great idea! While it's always difficult to reduce
complex ensembles down to a single virtue number (SPECmark, MPG,
and so on), more data can only help. The only problem that I see is
the fact that the metric of such tests is not "Virtue of Machine X"
but rather "Virtue of Machine X running software Y". This is fine
as long as it's marked clearly. The Gemini IG test suite will serve
as an interesting benchmark since it's open, in the following ways:
1. It can be run on multiple machines using OpenGVS. It would be a
way to verify and judge implementation and hardware. This is the
test that a GVS user would want to make, since it is a measure of
virtue for the range of available options.
2. It can be run using multiple software libraries (OpenGVS, dVS,
VistaWorks, SENSE8, Performer, etc.) on the same machine and
database. This is what an SGI user would want to test since it
is a measure of virtue for the range of available options.
An undecided user (not committed to either Gemini or SGI) would want to
examine the results of these tests for the full matrix of results and
the prices of the configurations. Now I see the wisdom of Gemini's
approach. Hardware vendors would shy away from #1 based on experiences
where poor software implementations make their machine look bad, but
this fear is resolved by test #2, which validates the benchmark by
showing how well the machine can solve the stated problem. If there is
a major difference between the results of test #1 for a given machine
and test #2 on that same system, then the meaning will be clear.
Based on this, I'm eager to get the Gemini Technology test suite and
have the Performer team get it up and running. If we can get it here
at SGI in time, we can have numbers for test #2 in time for John
Archdeacon's presentation at the Image Society meeting in June. We'll
post the results here on the mailing list after that. I'll get with
Gemini next week to get this started and see if it also includes both
single and multiple channel and single and multiple processor tests.
PERFORMER .VS OTHER TOOKITS
This is the wrong question. Many other toolkits use IRIS Performer,
some visibly, some not. We believe that SGI visual simulation and
high-speed rendering customers are best served by Performer, but we
have no reason to "slight" anyone who uses or sells an alternate
product on SGI, or who places portability above all else so long as
they know that that choice implies. For example, the Performer team has
helped most non-Performer vis-sim partners tune to SGI over the years
and have sent extracts from Performer source code where that made
sense. SGI appreciates it's partners and customers and has no reason
to engage in a debate about product merits. (Though I must admit that
the Gemini Technology posting has made me anxious to see the results
of tests #1 and #2 on SGI machines, but in a friendly way. :-)
THE DESIRE FOR A PORTABLE PERFORMER
IRIS Performer is much like IRIS GL in that it has grown up along with
increasing hardware capabilities and diverging application scopes. This
all comes from user experiences and the growing performance levels of
new machines, CPUs, and graphics systems. This leads to a very capable
but (perhaps) awkward and eclectic set of features. Moving from this
state to a highly portable situation while retaining the features that
make a product great requires much reevaluation and analysis. It's a
difficult task. This is how OpenGL was created and is what would be
needed to create the open, portable performer that people have been
asking for here in the mailing list and elsewhere.
People like carter++at++eai.com who are interested in openness issues, who
want to understand the SGI software strategy and development road-map
for IRIS Performer, OpenInventor, and the ImageVision library, or who
seek answers to issues such as portability to other workstations, PCs,
Java, and distributed WEB graphics, will find many answers at the
Silicon Graphics Developer's Forum this year (Forum info available at
http://www.sgi.com). Both Jim Helman and I will be making detailed
presentations on these issues and the news should please you (that's
the surprise I mentioned before ;-), but we are not free to disclose
details of our presentations before the Forum.
Be seeing you, Phone:415.390.1455 Fax:415.965.2658 M/S:8U-590
Michael T. Jones Silicon Graphics, Advanced Graphics Division
mtj++at++sgi.com 2011 N. Shoreline Blvd., Mtn. View, CA 94039-7311
"Competition is a by-product of productive work, not its goal. A
creative man is motivated by the desire to achieve, not by the desire
to beat others." -Ayn Rand
This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:52:47 PDT