THESE PRESS JUNKETS are great. You can get to meet all kinds of clever folk and ask them questions. They won't answer them but they will buy you lunch and put you up in a swanky hotel so you can run through the motions of not getting your questions answered.

Here in Barcelona, there were a few questions we didn't get answers to: Why was Barcelona late? It wasn't, apparently. Anyone who thought the chip was to launch back in March or even before must be a bit dim. Or listened to what they were being told and forgot to take it with a pinch of salt.

"It wasn't late," AMD corporate veep Alberto Macchi told me with a straight face.

He claimed the launch was in line with AMD's announcements on the matter. Back in the main hall, AMD's director of process technology, David Greenlaw, had earlier said the chip was late because it brought with it, "complicated design issues". So it was late then? Well, yes. Why is Macchi wasting my time? That'll be because he's a big cheese who obviously feels himself above the need to be frank with members of the press, I suspect.

We'll stick with him, shall we?

How many quad core Opterons will AMD make before Christmas? No idea. Why not? Because Macchi refused to tell me.

Back in the main hall, in the morning Giuseppe Amato sort of suggested there will be a speed bump in November and the clock frequency of the quad-core offering will be increased, maybe to 2.5GHz, from the 2GHz it is at currently.

In the afternoon, I ask how far the frequency can go before it breaks the speed envelope AMD keeps it in, in order for it to be compatible with the current platform, as promised. Amato skipped the question saying AMD wanted to make chips smarter rather than simply bump up the frequency. "We're not looking at the frequency," he said. Oh.

I assume from this that when AMD does up the clock frequency to 2.5GHz it'll bust the power envelope the firm has made so much of, unless it has some other tweaks up its sleeve.

Four cores are better than two, it seems. And if you pull out your old dual-core Opteron and plug in the new four-core chip, you'll get a performance boost of around 30 per cent over what you had before.

What do you do with the old chip? Whatever you like, Alberto Macchi helpfully suggests. He doesn't suggest wiring up a small African village with it and running a few virtual machines off it. Of course, it'll be useless without the board you just pulled it out of, so you'll probably chuck it in the bin, to end up as landfill. Very green. Very 50/15.

It's when these technology firms bang on about how green they are that the whole façade falls over. They don't get it at all. The whole issue of squeezing extra performance per watt is perhaps a useful exercise, but they don't do it to be green, they do it because their customers sitting in their data centres are toasting their nuts off. Their energy bills are going through the roof and not just because their kit draws so much current. It's also because they have to turn up the air con to fight the heat their chips are giving off.

What's more, most of the chips they have ticking away in their boxes are doing bugger all most of the time. They'll just be ticking over. But as Fujitsu Siemens' Bernard Brandwitte pointed out, an idle server consumes 70 per cent of the power of one that is actually doing something useful. (To illustrate, if you never use your microwave oven as anything more than a clock, it'll cost you 50 quid a year to run. 50 quid!)

Anyhow, this is where virtualisation comes in. All these servers sitting around doing bugger all could be doing something useful. If you haven't noticed yet, virtualisation is the next big thing. Instead of buying a new computer, you get your idle server to pretend to be one, or two, or half a dozen. This is fine, except for those times when your server has to do whatever it has to do; run the company payroll is an example used here. The beauty is that you just turn off your virtual machines and concentrate your processing power on the task at hand.

One of the aims - unless I misunderstood totally - is to get servers to use 50 per cent of their CPU capacity using virtualisation. At the moment, the average server CPU use is between five to 15 per cent.

If CPU usage hovers between the five to 15 per cent mark with two cores, what the hell do you need four for? I imagine that'll be because if chip firms don't keep making chips, and firms don't keep buying them, the whole edifice of US capitalism will crumble and the place won't be able to invade places it doesn't like the look of anymore.

On one of Giuseppe Amato's slides was the slogan "No core wars". Amato skipped over it, however. Like the megahurtz madness, core wars are probably here to stay for a while.

Bring on 45nm Shanghai, which is "on track," we were told. After that Montreal, will have more cores (eight) but not to wage war with, obviously. Best thing, though - now the precendent's been set - is that AMD will have to ship us all out to Shanghai and then to Montreal for the launches. µ