When Alan Priestley, Intel's European marketing manager, reiterated the company's plans to sell mid-range Xeon and high-end Itanium platforms for the same price, you have to wonder why they would do that. First, be clear what Intel means by a platform: it's the cost of a server's basics -- the combination of motherboard, chipsets and processor -- and the price that an OEM server vendor pays for a 'white box' server to which it adds its own components to customers' specifications.
Until now, Intel has been keen to underscore the difference between the two CPUs in terms of both price and functionality. Xeon is for lower-end, general purpose machines performing tasks such as Web and email serving, while the Itanium is for scientific and compute-intensive tasks, and competes with RISC processors. While they don't run the same software -- yet -- price-matching accomplishes two things.
First it makes clear Intel's intent to drive the 64-bit Itanium into the mainstream, just as it promised back in 1994 when it first announced the product. However, it could be a while before we get there.
OEMs can charge a lot more for Itanium-based boxes than they can for Xeon equivalents, simply because of the different tasks they're performing. You could pay 10 times as much for a finished Itanium server as for a Xeon equivalent, for example.
So in the short term, OEMs will cream off the extra profit at Intel's expense. Further out, however, expect price pressures to bring the price of Itanium-based systems down to the point where, if you're planning a new roll-out, plumping for Itanium makes as much sense as Xeon on a price-performance basis. At that point, the death of Xeon is only a matter of time.
The second result of the move is to put pressure on AMD's 64-bit Opteron. Even though the Xeon's new 64-bit extensions (EM64T) are a temporary technology until the world goes 64-bit, it's going to be a long wait for that to happen. In the meantime, Intel will compete head-on with AMD. Priestley downplayed the importance of EM64T but it needs that technology to compete effectively, despite the smaller company's relatively small market share. Intel is still run on the mantra that only the paranoid survive.
So the practical upshot? Expect Itanium-based systems to get cheaper -- and if they don't, pressurise hardware vendors to make them so, especially since Intel by its own admission hasn't really dented the RISC market; it claims three to five per cent of it, so the company needs to make this new mid-life boost for Itanium work. In the meantime, making head-to-head comparisons with Opteron-based systems could prove profitable.