Intel has effectively sounded the death knell for its Itanium processors by saying that its Xeon chips are moving into the high-availability, mission-critical arena.
Talking to thinq_ at Intel's Day in the Clouds event, Dylan Larson, director of Xeon Platform Marketing at Intel, said that the firm was developing its Xeon processors with a "no compromise attitude". When asked what that meant in real terms, Larson said it was to "position Xeon [processors] for the big back end".
The 'big back end' Larson was referring to is currently occupied by the Itanium servers that Intel pitches as ultra reliable, big iron machines. Larson said that Intel has "put significant investment in the Xeon's reliability features" adding that many of its customers use Xeon shod servers for mission-critical services.
Earlier in the day Larson told journalists that Intel will ship a version of its Atom processor under the Xeon brand. Single-socket 'microservers' will, according to Larson, make up 10 per cent of Intel's server sales by 2015.
According to Larson, those 'wimpy' single-socket edge servers would execute code on two or four socket Xeon servers. The hierarchy of servers is due to an admission by Larson that the server processor race, "is not just about improving frequencies and cores, but [having] to look at operational efficiency" adding "the frequency approach to producing products results in chips that are too hot to cool".
Thinq_ asked Larson whether this meant that Xeons would be pushed out, just like the Itanium, to which he responded "We [the Xeon processor group] haven't seen the Xeon move to the niche market yet but other products are disruptors". Larson said that his team "focuses on competitive threats" but when asked whether those threats include products from Intel, Larson quickly said no.
Larson's admission that Intel is planning to launch a modified Atom processor under the Xeon brand is a double-edged sword for the firm. On one hand it highlights just how good a brand Intel has with Xeon, but it raises the possibility that lightweight processors will push out larger monolithic chips, just like the Xeon did with RISC chips and by Larson's own admission, Intel's Itanium.
Usually we would say that Intel's product segmentation would be able to stop the firm cannibalising its own products but so strong is the market force for high-volume, lower-power server processors, Intel may well have to redefine what a Xeon processor looks like. As for the Itanium, it seems even Intel has given up trying to find a unique selling point for its big-budget folly.