home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Power Tools 1993 October - Disc 2
/
Power Tools (Disc 2)(October 1993)(HP).iso
/
hotlines
/
cshhl
/
sgi3d
/
script.txt
< prev
next >
Wrap
Text File
|
1993-05-10
|
12KB
|
214 lines
SGI01: Recently, Silicon Graphics has been aggressively attacking HP's
traditional mid-range graphics markets. In July, SGI launched
the R4000.
Be assured, that HP is not standing and watching. With the
launch of the 3D Aggressor promotion and other upcoming
announcements, you will see that HP is serious about shutting
SGI out of our traditional markets while making progress in
penetrating markets where SGI has been strong: AEC and Earth
Resources.
The purpose of this presentation is to help you see that HP has
a better story than SGI, especially when it comes to graphics
workstations under $60K. HP offers lower entry prices and
higher performance in almost every price space. Please use
this presentation to help you assemble your sales strategy and
feel free to use slides in your customer presentations where
appropriate.
SGI02: SGI has filled out it's product portfolio to be successful in
their traditional high-end as well as in the mid-range. This
slide is a "periodic table" depiction of SGI's current product
portfolio excluding multiprocessor systems and high end
servers. The left hand column shows CPU characteristics and
rows show graphics options and capabilities supported with
each respective CPU.
SGI03: There is far too much to say about SGI and/or HP on just one
slide. HP strengths can be positioned as SGI weak points and
vice versa.
SGI is currently a 840 M US$ company which grew at about 40%
in 1991/1992. Their worldwide market share reached 6.7%. SGI
acquired MIPS Computer in 1991, whose losses exceeded the
earnings of the new parent, SGI. This places the joint
company in a difficult financial situation. Losses as a
result of the move were reported at 118 M US$. Under the
assumption that SGI will make the same profit for the next
years they need 3 years to just recover from this loss.
Further, it is reported that SGI will layoff about 500 former
MIPS employees. This will also affect SGI's ability to
develop their "free public domain" tools like EXPLORER and
others in the future.
There are two technology white papers available on the hotline
that discuss SGI's graphics technology. The first compares
SGI's distributed 3D library, GL to the X-Consortium offering,
PEX. The second paper discusses the pros and cons of 64-bit
technology.
SGI04: In this slide we compare entry level systems from SGI and HP;
the 710C and 720CRX to the Indigo and the higher performing
Indigo XS.
The 710C competes very nicely with the Indigo 8bit and
customers reportedly mentioned that graphics performance (3D
shading) is about equal.
If you are in competition with an Indigo XS, you would be
better not to discuss performance. In this case it would be
more effective to discuss HP as a company, HP's CPU
development strategy, futures, support etc.
SGI05: In the mid-range SGI has been very aggressive. HP will not be
beaten. Through the 3D aggressor program, we competes very
nicely with SGI. The 720CRX-24Z is very competitive with the
CRIMSON ELAN graphics, especially since we have a comfortable
price advantage. The 720-CRX24 compares nicely with the SGI
XS subsystem, having the advantage of true color. The higher
performance ELAN graphics subsystem is the best comparison.
In real tests of application performance, it doesn't likely
show more than 68K quads/sec. Also note that a CRX24 performs
amazingly well in areas like NURBS thanks to our Powershade
rendering library.
SGI06: The high-end comparison shows that there is a good chance of
beating SGI even if their offering includes a VGX, their high
end system. The slide actually shows some published GPC
performance numbers. GPC benchmarks are standard programs that
emulate application behavior for different markets. This is
one of the rare numbers SGI published and it is quiet useful in
convincing customers to keep a TVRX in mind. If the VGX or
VGXT feature set is important to the customer it is very likely
that you will have problems. On the other hand the TVRX
supports some features SGI doesn't, like contouring in HW,
sectioning/capping and deformation animation. This can be
especially of help for MCAD customers.
SGI07: Please refer to my white paper GL vs. PEX to get more
ammunition should your customer be asking for GL availability
on HP or is indicating that GL is a standard API on the
graphics markets. HP will be offering GL 4.0 emulation from a
third-party software partner. This will help you move SGI
customer over to the HP platform with a reduced impact on
software investment. This is ideal for those customers who
currently have application written on GL 4.0 and who would like
to move to PEX.
SGI08: Xlib/PEXlib is the solution of choice for distributed
graphics. Here we try to show the architecture of PEX/PEXlib
compared to X/Xlib. It also shows how very nicely X and PEX
fit together. Again, the GL vs PEX white paper explains the
details and benefits of PEX/PEXlib.
SGI09: This is a list of 3rd party API's available on HP
workstations. Of particular interest is HOOPS, a very
portable, easy to program and sophisticated object oriented
API. As mentioned above, SGI GL 4.0 is available from Nth
graphics. HP is currently in negotiation to get a GL 4.0
implementation on top of PEXlib. The current implementation
is on top of Starbase.
SGI10: High level API's are very often based on low level API's as
this slide shows. Different markets have different needs. In
the Scientific Visualization market segment GL is very well
known. SciVis usually needs no display lists to manage
complex hierarchical geometric structures. In the ME
application markets, however, hierarchical structures are an
extremely important requirement. Therefore, GL doesn't have
much influence in this market segment. However, it is
possible to base hierarchical and immediate mode APIs on top
of PEX. GL (even OpenGL) will limit the customer to immediate
mode.
SGI11: Read this slide carefully!!! If a customer is currently based
on GL 4.0 and wants to move to GL 5.0 he has to rewrite major
parts of his application and doesn't even get the full
functionality. This is a very strong point for HP and PEX! A
customer could port to Starbase or PEXlib instead which would
give him more flexibility and the use of a REAL standard.
SGI12: Finally this slide gives an overview of the architecture limits
of X, GL and PEX. As you can see, PEX provides distributed
access to a much higher level of graphics functionality.
SGI13: The next four slides provide a glimpse of target markets where
HP competes against SGI. It is intended that these slides
give you some background for slide SGI17 which provides a
competitive positioning matrix.
SGI is targeting the ME-CAD market very aggressively. This
slide provides background on HP's and SGI's respective
positions in this market segment.
SGI14: SGI also tries heavily to penetrate the AEC/FM market segment
where HP was also traditionally very strong. Read this slide
together with slide SGI17.
SGI15: In the Earth Resources market segment, SGI has traditionally a
larger installed base. HP is currently attempting to increase
our presence in this market segment through aggressive market
development activity.. With products like the CRX24 and CRX24Z
we are very competitive, especially if we consider our compute
performance.
SGI16: Scientific Visualization is the traditional market segment for
SGI. Therefore, its number one position is not due to the fact
we can't compete but due to historical reasons. In fact HP is
gaining market share in this segment very rapidly.
Applications like AVS and others run best on HP.
SGI17: This competitive positioning matrix is provided as a tool to
help you bid the most optimal system for each target market
segment. It also will give you an idea of what is the likely
system to be bid by SGI. Please note that this is only a
indication that might not be always true. Customers could
have special requirements like dual head or additional EISA
slots that make these recommendations not suitable for the
given market.
SGI18: The next few slides examine HP and SGI's CPU technology and
and position HP as the most likely to succeed in the future.
HP has been a leader in RISC-based CPU architecture having
introduced the first commercially viable RISC-based system in
1985. Over the years, HP's experience in RISC CPU design
coupled with other advances have positioned HP as the
undisputed leader in CPU performance and architecture.
This slides shows the current and future development of our
PA-RISC technology versus our main competitors. Press reports
indicate that SGI will introduce a 67 MHz R4000 processor this
year which will bring them into the 100 SPECmark range. Other
reports claim that SGI may postpone the development of the
R5000 processor or might even consider alternatives that will
be available on the market to replace its MIPS processors.
SGI19: The R4000 would need to run twice the current speed (200 MHz
internal and 100 MHz external) to keep up with the speed of the
PA7100 chip. With the current technology available this is
very unlikely to happen. A 67 MHz version is probably all they
can achieve.
SGI20: SGI is frequently highlights the 64-bit technology of the MIPS
R4000 processor. This is ironic since SGI's current operating
system has does not support 64 bit addressing. In any event,
64-bit addressing provides some advantages and disadvantages.
64-bit physical addressing isn't needed today because a 32b-it
physical addressing can already support up to 4 GB RAM which
isn't a useful configuration given the current RAM technology
available on the market. A white paper is available that
speaks in detail about the tradeoffs of 64-bit addressing.
SGI21: Virtual addressing can be handled as a flat address space
(addresses are starting with the lowest address and are going
up to the highest in one address space) or as a segmented
address space(segments of 4GB [32bit addressing]) like with the
PA-RISC and IBM RS/6000. The advantages and disadvantages of
each technology are listed on this slide.
SGI22: HP's PA-RISC implementation is 64-bit. In terms of addressing,
64-bit is not needed as you can see on the previous slides.
Please refer to the white paper for more detailed information.