Intel's battle to ensure a good future for its 64bit Itanium chip suffered several setbacks recently. First, IBM proudly promoted the third version of its x86 server chipset - and announced it would stop making such chipsets for Itanium servers. Shortly afterwards came news that Intel missed its Itanium sales goals for 2004.

Just as we might tolerate and nurture a lunatic uncle, the industry has politely overlooked the occasional outbursts of Itanium hyperbole from Intel. After all, such things usually don't cause much harm, and it's not fair to apportion blame for follies of the mind. Even so, the recent news represents two black eyes

for Intel. Especially since IBM says its new X3 chipset for x86 servers is so good that Dell and HP will give up all hope of selling their own four-way servers. It seems IBM doesn't want to dominate the Itanium server market.

So after such a thorough drubbing, it's unsurprising that Intel bruiser and chief technology officer Pat Gelsinger took the gloves off at this month's Intel Developer Forum (IDF). I call him a bruiser because as well as having a brilliant mind and a powerful personality, Gelsinger has been accused of kicking troublesome journalists. Something I feel should not be forgotten.

Nifty footwork aside, Gelsinger was design manager and chief architect of the original 486 processor, and has led Intel's desktop products group. Although it's not mentioned in his biography, I wouldn't be surprised if he led the work on Intel's 64bit Xeon chips. Then again, I'd be surprised to see anyone shout about that while still working at Intel. Certainly Gelsinger seems to have a bit more invested in Intel than most, and his credibility would be an expensive asset to burn on the Itanium pyre.

And of course, many a good fellow has already fallen in earlier skirmishes. I was saddened to see Mike Fister, one-time top Itanium evangelist, jump ship from Intel to lead chip-design tools firm Cadence. I don't think he ever said he wanted to spend more time with his family, but his move did not exactly underscore confidence in the Itanium.

So, with Fister flown to Cadence, it's left to Gelsinger to step up and fight the good fight. At IDF he argued that IBM is bound to cold shoulder the Itanium because IBM has the 64bit PowerPC to promote. How's that for a U-turn. I seem to remember Fister telling us that Itanium was bound to replace PowerPC chips along with the rest of the proprietary Risc processors. Now it's hard to see who remains committed to Itanium beyond the folk at Intel.

Itanium co-creator HP will continue to foist Itaniums off onto Superdome buyers, but they really are a special case. After all, the other Superdome processor is the PA-Risc chip, with which the Itanium is compatible both at the pin and binary levels.

It's striking that throughout the Itanium debacle the chip has been painted largely as an Intel effort, yet at every stage it seems that PA-Risc replacement has been the actual design goal.

Now HP has transferred its chipset engineers to the employ of Intel. Could it be that even HP can see the writing on the wall?