There’s a short preview of the Atom processor here. The conclusion is that it performs pretty dreadfully, and the gap is so big that one can’t blame it on a low FSB or an alpha BIOS.
Even if it’s possible to overclock the bejeezus out of them, you’re going to have a rough time even with a severely overclocked dual-core Atom to match an Athlon X2 4000+, much less a C2D.
So if you’re looking for performance, look no further. You’re not going to find it here. If you want to read a full coroner’s report, it will come in due time, but they’ll just say the same thing in more words.
This is the first generation of this type of processor, and they’ll no doubt get better, but it’s pretty safe to say that the Atoms will never become what the first Celerons were ten years ago, real competitors to the big boys when overclocked.
Should that leave you “bitter,” as the article suggests? Yes, but not for the reason the article suggests.
You shouldn’t be bitter because the Atom is a poor performer. It wasn’t meant for you, and that’s why you should be “bitter.”
Grandma Stops Paying For Your R&D
Until now, the average computer user has effectively subsidized the high-performance folks. The R&D budgets were mostly geared to generating ever-faster CPUs, but the costs got spread over all the processors made. They got the thrills, the average Sixpack paid the bills. Yes, the high-performers sometimes paid more for it, but nowhere near what they would have paid if all the R&D costs had to be covered by the relatively few high-performance units sold
Until a few years back, this wasn’t too unjust because even the lightest computer user could make real use out of even a fraction of the increased performance available from new generations of processors.
That era ended a few years back. Grandma doesn’t need a quad or even dual-core to send emails and look at webpages. Really, she isn’t playing GTA IV on the sneak when Grandpa isn’t looking.
There are two types of grandmas in the world. There are grandmas in the computerized world who won’t pay a lot for that overkill CPU, and there are grandmas in the uncomputerized world who can’t pay a lot for that overkill CPUs.
Grandmas United aren’t a big enough interest group to change the world, but it turns out they have tons of allies. Whether it’s a smartphone, a notebook, or a first desktop, the remaining new markets consists mostly of potential customers who, for varying reasons, can’t pay a lot for a CPU.
Inevitably, if you build products to capture those markets, those same products will appeal to those in current markets who don’t want to pay more or even the same for what they buy.
What the Atom processor means is that Intel knows that it can no longer run its business as “one size fits all,” and in the long run, it can no longer sell $100 or $200 CPUs to people who need much less.
So, in the long run, say by around 2015, the majority, probably a large majority of Intel processors, will be some descendant of Atom. Intel (and AMD, if they’re still around) will stack ’em high, sell them cheaper, and make them even cheaper. Mind you, the number of CPUs being made will be much higher than the historical trend would indicate; rather than 600-700 million being sold, it will probably be more like 1000-1500 million, and could be even higher than that.
There will still be a “regular” Intel processor line, but it may not be as big in total numbers as it is today, and certainly will be a small percentage of total CPUs made (though a much bigger share of revenues). Those will be the servers, the workstations and the “luxury PCs.”
People who want the power will still be able to get it, but since the R&D required to get it will probably increase just as the costs will have to be spread over a relatively small percentage of CPUs rather than all of them, prices will at best stay the same, or more likely, go up.
By 2015, the situation will look something like today’s desktop/server CPU pricing, or console vs gaming PC. You’ll either pay $26-50 for your tinyCPU, or you’ll pay a number of hundreds, maybe even a couple thousand at the very high-end, for a very high-end CPU.
In other words, the desktop world as we’ve known it will be gone. PC equivalent will be either very high- or low-end, with little inbetween.
This won’t happen tomorrow or next year, probably won’t start really happening until after 2010, but it will happen.