Re: SGI vs. WINTEL
> Compared with UNIX, the TCP/IP stuff by MS is simplistic
> but sufficient. In any case, you need a good X emulator.
> "X on Net" I believe it is called works flawlessly with Suns
> and it can be had for about $50/seat on a site license.
The apps are sufficient if communication is only limited to FTP and telnet. WindowsNT has no built-in NFS for file sharing. Printer sharing does not work with all UNIX variants. Even setting up WindowsNT as an FTP server is not as straight forward as our NT administrator would like it to be. But these problems are mangeable with third party products and may not affect all mixed computing environments. As for "X on Net". I must admit that I have never heard of the product. The $200/seat price for X emulation which I quoted was based on eXceed and PC X-ware; the two leading X emulators for Windows.
> I can relate to your demanding simulation app. But that isn't
> SGI's bread and butter anymore. The current annual report
> indicates that >50% of the sales are on desktop systems.
>
> The # of polys/sec per 1$ doubles about every 2 years. So in 2
> years the boxes now doing $1M polys for say $8K (O2 included)
> will be doing 2M polys/sec in 1998 and 4M polys/sec in 2000.
> Pretty hard to imagine.
This debate very much reminds me of the Mainframe vs. PC debate that caused a loss of investor confidence in many mainframe manufacturers, particularly IBM. I choose IBM because I see alot of parallels in IBM's and SGI's situations. When IBM disclosed its first quarterly loss a few years ago, it was blamed on poor hardware sales, particularly from the mainframe unit. The story was the same for many other mainframe manufacturers. The streets rang with a victory cry from the PC community. The PC had finally killed the mainframe! Well, reports of the mainframe's demise were exaggerated. The truth is that many traditional mainframe functions, like electronic publishing, were being off-loaded onto PC's. However, there were still plenty of critical applications that PC's just couldn't handle. The mainframe market slumped due to a number of inter-related reasons. Customers spent their IT budgets downsizing to desktop computers for non-critical tasks. Very little money was left for purchasing new mainframes which the customer did not need anyway because... As the non-critical tasks were off loaded from the mainframe, it actually increased the useful life of the mainframe for those critical tasks it still had to perform. This led to an overall decrease in demand for new mainframes. The mainframe market was suddenly overcrowded and manufacturers suffered.
In a little less than two years, IBM turned a profit due to an increase in mainframe sales! How was this possible? Wasn't the mainframe was dead? It's easy to say that some of IBM's competition disappeared and sales were bound to pick up as a result. Also, IBM did alot of cost cutting. But, IBM's mainframe business has been solidly profitable for a couple of years now. The real factor behind IBM's mainframe revival is its very loyal customer base in the financial community. As financial services grew and transactions became more complex (particularly with new investment options like derivatives), the old mainframe systems started to show their age; and PC's were still incapable of handling the large data load. So, financial companies needed to upgrade their systems, and they bought from a company they have trusted for many years.
I see SGI in the same light. I believe SGI's shifting sales to desktop systems is a temporary phenomenon. It is a sign that companies are currently downsizing from older SGI servers to SGI desktops. The new SGI desktops have the same capabilities as some older servers, yet cost less than the maintenance contracts for the servers. It's a matter of economics. The desktops may be all some companies need. And SGI may have to work hard in the future to keep these customers. But there are other customers whose graphics and information demands will only increase. There are many possibilities for the future of entertainment and information access that only SGI servers could make reality. Some of these could turn into booming businesses. The key is that SGI has already gained the loyalty of many in the entertainment and infotainment community and has the technology to keep it. Frequently, you have mentioned that SGI cannot continue to rely on its graphics lead because PC graphics are catching up and we are close to reaching the point where increased graphics speed won't make a difference to the human eye. That may be true if we are interested in only one pair of human eyes looking at a single screen. But what about applications that provide multiple graphics screens to multiple persons? It might take only 1 million textured polygons per second to provide a realistic image to one screen. But it takes 20M polys/sec to drive twenty displays with different content or different perspectives on the same content. The question then becomes, what is the cheapest, most appropriate solution for this application? Buy twenty networked PC's using a distributed application. Or, buy twenty terminals and one SGI server using a client/server application. Our flight simulators require that four out-the-windows displays are driven, each with a different perspective on the same environment; we also have numerous cockpit displays that supply pilot's information. This is just one example of a graphics intensive application that PC's will not be able to handle for many more years. Our application may never become mainstream. [Maybe flight simulators will replace arcades in the malls :).] But there could be other equally demanding applications that will reach the mainstream. Always beware of taking a single-user, PC centric view of applications. Most applications in use connect with multiple users and multiple computers/terminals. |
