Enough already, as they say in California, where the vast majority of PC chips are made. The lunatic rate at which these things are changing has gone way beyond the comprehension of anybody - even those who work in the industry. That wouldn't matter if it weren't also clear that these faster chips are doing nobody much good.
I'm speaking specifically about the volume desktop here. In workstations and servers, you can't get too much power and few people ask the price. But when you're on the subject of how to fill in the network seats, the old rules no longer apply.
It was arguable until quite recently that buying the best you could afford was a sensible long-term policy, given the rapid rate of change in software demands. The wildfire success of Windows 3.0 was manna to Intel, encouraging its red-X' campaign to blot out sales of its own 286 processor in favour of the higher-margin 386. It wasn't to be the last time Intel demonstrated its appetite for killing off its first born in favour of the hot new stuff.
Who could blame it? We needed those faster chips to buy into the gewgaws Microsoft and others were feeding us. Iterations of Windows, the associated applications that grew apace and the trend towards multitasking that was engendered, multimedia - these were all grist to the mill. There were plenty of killer apps to shift silicon. All Intel had to do was avoid being too greedy on the amount of light it put between its fastest chips and its mainstream ones. A year after the 486 launched, it was still almost invisible - only when AMD started to cut some 386 market share did Intel realise that the smartest way to move was to build them faster and keep them coming. It never looked back and success story followed success story - millions upon millions of Pentium and Pentium II sales.
I'm sure Intel wishes there were another killer app around the corner, but it doesn't have too much to worry about. The video and 3D demands of PC gaming are still driving it forward, and DVD movies are likely to follow up with another hefty shove. But that doesn't mean the enterprise has to follow.
Business isn't really finding any desktop killer apps that aren't Internet-related and most Net activities aren't CPU-intensive. Indeed, the movement is to thinner clients that rely on the server to handle most processing tasks.
That hasn't stopped Intel trying though, and its latest effort to convince us we need all those CPU cycles is called Constant Computing. It means the ability to run background processes that would usually be slightly pesky foreground processes - the typical instances that Intel cites are virus scanning, file encryption and agents. You could have an agent searching around for updates to your favourite Web sites while you work, for instance.
Frankly, if Constant Computing works we should all head off home and accept that, just as the Microsoft-bashers have always told us, marketing really does run this business. I don't believe that's the case though.
The advantages offered by Constant Computing look thin and that's going to mean that, for almost the first time, ultra-cheap desktops are going to be the smartest choice for enterprises. Intel has spotted this and is now putting plenty of weight behind the new generation of cached Celeron processors. The savings here over full-fat CPU PCs are potentially hundreds of pounds per system. It's time to cash in on those cheap chips.