|By Maureen O'Gara||
|February 15, 2013 10:29 AM EST||
Remember back almost two years ago now when Oracle said it wasn't going to write software anymore for the Itanium chip that HP depended on for its high-end, high-margin Integrity and Superdome servers with their lucrative service contracts because the Itanium chip was secretly nearing end-of-life?
Well, maybe Oracle was righter than Intel and HP's combined and frequently repeated denials, even the protestations HP made in court to force Oracle to go back to work, which it did late last year.
See, on this past January 31, Intel - quiet as a church mouse - posted a terse notice on its web site saying that the Itanium roadmap had changed and that the next, new Itanium, the so-called Kittson chip, won't be manufactured as a 22nm shrink in Intel's newfangled TriGate process and won't share a common socket with the x86 as planned.
What it said exactly was that:
"Intel has updated the definition of the next-generation Itanium processor, code name "Kittson." Kittson will be manufactured on Intel's 32nm process technology and will be socket-compatible with the existing Intel Itanium 9300/9500 platforms, providing customers with performance improvements, investment protection, and a seamless upgrade path for existing systems. The modular development model, which converges on a common Intel Xeon/Intel Itanium socket and motherboard, will be evaluated for future implementation opportunities."
Well, one can assume that HP isn't funding the development like it's been doing for years now either because it can't afford to anymore - having blown the kitty on Autonomy - or because it isn't worth it because the HP-UX market has evaporated, blown away by Oracle.
Gartner says shipments of Unix servers dove 31.1% in 3Q12 on revenues down 16.4%. HP's shipments sank 38.1% on revenues down 28.2%.
Intel has been talking about x86-Itanium convergence for 10 years but only said it would happen with Kittson, which is theoretically due next year, in November when it unveiled the belated eight-core 32nm Poulson model of the Itanium, now the Itanium 9500, which is socket-compatible with the previous Tukwila Itanium, not the x86 as originally promised.
HP was also depending on convergence for the smooth transition of its Integrity and Superdome lines under its Project Odyssey that imagined a single high-end server platform that could accommodate either chip and benefit from the technology in both.
Aside from blades or motherboards that share common components and can be used with either Xeons or Itaniums, the performance, core count and clock speed enhancements and cache memory boosts to be expected from a 22nm shrink are being sacrificed, leaving just a gussied-up Poulson masquerading as a Kittson and supporting Oracle's contention that Itanium is staring at end-of-life.
The Intel-produced chart above shows the commonality that was supposed to happen and now won't. Beside a common processor socket and packaging, it was supposed to involve common chip elements, shared memory, RAS and I/O controllers. There was no precise plan, at least not a public one, but Intel in November referred to it as a "sustainable path forward" and it did involve HP's OpenVMS and NonStop servers too.