Actually, best value in purchasing is achieved, not by buying the very fastest components possible under your budget, but by buying whatever components
1) meet your minimum needs
and
2) have greater price/performance ratios than competing products, including products of the same brand.
This is especially true in the computer hardware market, where prices fall so rapidly and technology advances so quickly. Never think of a computer systems as an investment in longevity. If you do, you've already lost by resigning yourself to spending exhorbiantly. Instead, think of a computer systems as a disposable commodity. It's sad but it's the truth -- book depreciation on computer parts is an incredible 30% per year, and real market depreciating is closer to 50%. If you bought, say an Athlon 1.1 GHz for $80 more than the 1.0 GHz, it might be marginally faster today, but just think what that $80 will buy you next year!
In other words, the value of your computer depreciates faster than the value of your money, so you want to change as little of your money as possible into computer parts. You'd be much better off buying a cheap computer every year as opposed to an expensive computer every two years. Now, in real estate or collectibles, it's entirely different. Value of property tends to go up rather than down, so investment is wise.
Another common fallacy is that we ought to compare each component's price with the cost of the whole system. This is easily shown wrong by an example of a $800 system with a $200 video card and consideration being given to a $400 video card that was only 20% faster. By conventional wisdom, you would have to buy the more expensive video card, because it would only add 20% to the cost of the whole system while delivering the required 20% performance boost. But obviously, we must compare components to components to achieve best value.
The best system today is a Duron 750 with 128M of ram, a 20G or 30G 7200 rpm hard drive, and a GeForce 2 MX. No other system comes close in terms of price/performance, and no other system can demonstrate the need for any further speed in today's applications (with the possible exception of professional audio/video/3D work).
Modus