Are you saying cost is not a significant factor to a business? I understand your point about the utility of the platform but to deny that cost is not also a significant factor to a business is not an argument that reflects reality. Even when economies scale factors come in you still want to minimize your costs to maximize your profits.
I am not asserting that cost isn't a significant factor to businesses. I'm stating flatly that the way a business calculates cost is different than someone going to a computer store and purchasing a system. This is a constant argument I see people getting wrong on the "it's free!" side when talking about open-source, as well as going too far with the "better TCO" side when talking about closed-source. The reality is somewhere between, and when the proprietary considerations we're talking about are very high-end and individually costly (like the SGI machines), the confusion gets even worse and the argument seems to assume that cost goes up proportionally to individual unit costs instead of several factors.
Yes, maximizing profit is important, but the factors that go into maximizing overall profits are a whole load more complex than the "guy in a computer shop" perspective you're seeming to expect me to take for granted.
If cost was never a factor, All the houses would be still be using SGI workstations (meaning SGI would never have gone out of the workstation business) or hang them all and purchase Cray supercomuters.
You really do have no clue. Why assume that I'm saying cost is never a factor? Arguing against that strawman is easy.
First off, SGI is still selling graphics workstations (as BOXX Technologies), and they sell those workstations with either Windows or Linux installed for use in high-end graphics work, depending on customer needs. SGI didn't get "out of the workstation business" at all, they experienced a period where they were all-of-the-sudden faced with a gradually-declining monopoly on high-end graphics capability due to maintaining an antiquated platform (the MIPS chips) and later building their systems on a marginal platform (Itanium chips), and only recently switching to a more mainstream platform in order to compete. Their lack of competitiveness wasn't the software, it was the hardware, due primarily to x86 chips having adopted the huge improvements AMD brought to the market along with the innovation Intel was able to implement in its chips, both of which blew all other workstation chips out of the water (yes, Apple fans, it blew the PPC chips away as well)-- so much so that SGI, after going through chapter 11 a couple years ago, switched their build platform to... any guesses?
Second off, Crays would suck for the work these 3D shops need.
Third, another company who is a good example of one that sells very costly machines at prices the average joe would balk at would be Sun, and yet their systems are in use all over and you probably don't even know it. The reason why you (or someone else reading) might not know it is because the Sun systems are mostly in use where they actually provide a value despite the sticker price, and that is very much not your desktop at home. While their prices would seem outrageously high for someone to use a Sun Solaris workstation as their home computer (though I'm sure there are people who do), Sun actually does just fine selling its workstations and servers to the markets that get optimal utility from their product, and those who have the scale sufficient to require the Sun systems actually gain value above and beyond the sticker price of the machines themselves. Compare a Sun Solaris system price to that of a Dell system running either Linux or Windows, and dollars to donuts you're going to find the Dell to be significantly cheaper, and while Dell systems certainly sell more volume when you look at NOCs, teleco switching centers, or network datacenters you're very often going to find a large number of Solaris machines chugging away. The reason for this is because you'll find that despite the much greater cost per machine, the value derived from the systems and the life span of operability makes the system an overall benefit in terms of profits, but this is dependent on the scale of the operation for it to work out that way-- while the relatively small company network I manage wouldn't benefit from the cost of a Sun system running things (though there was a point where that was debatable), a large NOC actually gets a much greater benefit from running the Sun systems and will see the benefit much sooner than would a smaller operation.
To make a long story short, scale matters a great deal when determining TCO, and the common mistake of confusing price with value is one you (uruk) have made in your assertions.
