For the past few months, we’ve been seeing a new trend in overclocking.
It’s something I’ll call “cheapie computing.”
Now before you go, “DOH,” hear me out.
Overclocking has usually had a strong economic motive behind it: the desire for cheaper computing, the desire to get the best bang for the buck.
This is something different. Cheapie computing emphasizes the buck over the bang. Really emphasizes the buck over the bang.
Yes, there’s been “bottom-feeders” all along, but the bottom seems to be getting awfully crowded lately.
Let me give two examples to illustrate:
A classic example of cheaper computing is the Celeron 300A (and successors). Overclocked, they didn’t perform quite as well as Pentium II or III, but people saved several hundred dollars using them instead.
An example of cheapie computing are people buying 1700+ “J” Athlons rather than 2100+ “A.” Overclocked, the 1700+ don’t perform quite as well as the 2100+, but people saved several dozen dollars using them instead.
From what I’ve seen, many if not most of those who bought them had very realistic expectations from the cheaper chips. They knew they wouldn’t do quite as well, but saving a couple dozen dollars meant more than getting a couple hundred extra megahertz.
These people aren’t crazy or stupid. Just different, but that difference can have a big impact on the future.
What will cheapie computing mean in the future?
The next obvious manifestation of cheap computing will be overclockers buying 133MHz PIV and Springdales rather than 200MHz PIVs and Canterwoods to save $50-75.
Far more important, though, is what won’t happen. Relatively few people will be buying any kind of PIV system, and there will be relatively little defection from socket Aers. They stick to the $50 wonders and the cheaper socket A boards, and to hell with any performance difference; it’s good enough.
This is good for AMD’s market share if not their profits. It should keep the AMDers in place for most of the year.
Then comes Hammer.
First will come Opterons, and it looks like AMD will charge at least $250 for them. That won’t even show up on the radar screen with this crew.
Later comes Athlon64. Looks like AMD plans to charge a lot for the 1Mb version, and a considerable sum (at least initially) for the 256K version.
I don’t think these folks are going to touch them. They’ll sit on their socket As until they’re good and ready and at least Hammer Jr. gets over 4GHz (equivalent) and into that $50 range.
AMD’s problem is that they need to get paid more for their processors, but the point at which their core constituency pays becomes less and less. If no one in AMD’s core constituency wants to buy Athlon64s until they’re dirt cheap, just who is going to buy them, and what happens in the meantime to AMD?
AMD has tried to keep prices high the last number of months and keep rough price parity with Intel. This has been quite a departure from the past, when AMD chips usually sold at a discount of at least a third less than equivalent Intel processors.
The recent price cuts seem to indicate this hasn’t worked all too well. The upper range of AMD processors are now selling at about a 20% discount from equivalent Intel prices, with the discount increasing at lower levels.
Another indicator on how well AMD is doing will come tonight, when AMD releases its financial results and have its briefing on it.
We’ll talk about this tomorrow.