In a discussion on the merits of various processors, Torvalds wrote that Intel had made the same mistakes "that everybody else did 15 years ago" when RISC architecture was first appearing. Itanium tries to introduce an architecture that is clean and technically pure, something that just doesn't seem to work in the real world. He claims that Intel "threw out all the good parts of the x86 because people thought those parts were ugly. They aren't ugly, they're the 'charming oddity' that makes it do well."
He almost certainly has a point. Although a 1.5GHz Itanium would be faster than a 1.5GHz Pentium 4, that's not the competition it has to face. The Pentium 4 is available at twice that clockrate and the Itanium only gets to keep pace by having huge amounts of cache despite all of its clever architecture.
Clever architecture is something that has trapped others in the past. The Alpha processor team spent years learning that many of the architecturally correct ideals they had held needed to be thrown out when it came to the real world. According to Torvalds, "And all the RISC stuff that tried to avoid it was just a BIG WASTE OF TIME. Because the _only_ thing the RISC approach ended up showing was that eventually you have to do the hard stuff anyway, so you might as well design for doing it in the first place."
He goes on to write, "As far as I know, the _only_ things Itanium 2 does better on is (a) FP kernels, partly due to a huge cache and (b) big databases, entirely because the P4 is crippled with lots of memory". That crippling with lots of memory is due to what many people describe as a major kludge in the Pentium architecture called Page Address Extensions (PAE). According to Torvalds, "the only real major failure of the x86 is the PAE crud".
There are quite a few Xeons in particular that you will see advertised with 8GB or 16GB of memory. Astute observers will have wondered how a 32bit processor can address more than 4GB of memory. PAE is the answer. It allows 36bit addressing using 'pages' of memory. According to Torvalds, the Pentium 4 is crippled "because Intel refuses to do a 64-bit version (because they know it would totally kill ia-64)."
AMD can take some heart at his comments. Reading between the lines, it's obvious that Torvalds thinks x86-64 is the way to go. "Right now Intel doesn't even seem to be interested in '64-bit for the masses', and maybe IBM will be. AMD certainly seems to be serious about the 'masses' part, which in the end is the only part that really matters". It's worth noting that Torvalds' employer, Transmeta, has licensed x86-64 so he is likely to have access to Hammer hardware.
Intel has spent a huge amount of money developing its 64bit processor but the payoff in the real world is likely to take a long time. The problems that are being found now that the processor is out with customers are going to take a lot of effort to smooth over. In his scathing view on the Itanium, Torvalds postulates that "in another 5 years they'll get to where the x86 has been for the last 10 years".
In what could be the best news for AMD, Torvalds summarised his thoughts on Itanium. "Code size matters. Price matters. Real world matters. And ia-64 at least so far falls flat on its face on ALL of these." µ
Linus Torvalds prays Intel will adopt Yamhill
Only 3,500 Intel Itanium servers shipped in 2002
Intel burns AMD-clone Yamhill idea
Intel won't produce AMD clone - Otellini
INQUIRER Hammer coverage
Intel roadmaps page