New York (NY) – In what may be considered the beginning of a third wave of efforts to prove Intel’s IA-64 architecture is viable for high-performance applications, even if it’s not compatible with the x86 architecture that came before it, a coalition of major server vendors, plus some independent software vendors, along with Microsoft, Red Hat, Novell, and Intel itself, have formed today the Itanium Solutions Alliance.
Surprisingly failing to see the irony in the new coalition’s own abbreviation – firms outside of Intel to whom we spoke today were unaware of its historical significance – representatives of the new Alliance told Tom’s Hardware Guide their purpose will be to serve as a “single point of focus” for software companies of all sizes to contact for guidance, training, and even encouragement to port their applications to the Itanium platform. As Stephen Howard, director of enterprise solution alliances for Hewlett-Packard, told us, one of the new Alliance’s first ventures will be the Itanium Solutions Center Network, a group of 19 permanent, physical locations worldwide where Alliance representatives can meet with companies and work with them to support and develop for IA-64.
|Intel’s Itanium processor|
These network stations, said Howard, will be “staffed by experts in software development, for helping ISV developers get on the Itanium platforms that each of our companies build…Instead of fishing around to find all of us, and our capabilities, by registering with the Solutions Alliance, they can be linked up with the geographic location or the expertise that’s going to work best for them, so that they can quickly get themselves onto the platform and get their product out the door.”
The Alliance’s efforts go to the heart of what had been characterized as Itanium’s chief deficiency: not its architecture, which after a rough start, actually has proven itself very capable. It is an altogether different platform, not because it’s truly 64-bit, but because it would have its developers embrace a concept called Explicitly Parallel Instruction Set (EPIC). It is the industry’s first serious attempt at a CISC-based multithreading architecture, based on a simple concept to explain but a difficult one to implement: the idea that when processes fork and parallelism begins, it’s because the code of the program tells it to. It’s this concept which most starkly distinguishes Itanium from x86 (x64) architecture, which actually has no parallelism principles of its own. The ability of multithreaded x86 processors to fork processes into separate cores is based mainly on their ability to ascertain for themselves when such forking is permissible. One of Intel’s most ambitious tests of this capability has been hyperthreading, which is a parallelism technique for single-core processors. But HT is an experiment that will probably come to an end as dual-core and multicore processors become mainstream. As they do, they will undoubtedly bring Intel’s version of 64-bit x86 architecture (EM64T) as well as AMD’s (AMD64) into the mainstream even in high-performance categories.
So what chance does Itanium have, and why are companies such as HP, along with Silicon Graphics, Bull, Hitachi, Fujitsu, NEC, and Unisys, devoting their resources to the platform? It’s because that platform hasn’t failed yet, despite some of us in the press having written it off time and time again. HP’s Stephen Howard revealed to us, “HP’s business has doubled in Itanium over the course of the last year, and we’re looking to do that again. So we think we’re really at the beginning of the ramp of Itanium sales.” Representatives of SGI and Bull also told us today they’re looking to double their revenue from Itanium by 2008 – in SGI’s case, from an estimated $3 billion to $6 billion per annum.
Building a social migration path
Since Intel launched the first generation of Itanium in 2001, the “point of focus,” to coin a phrase, for the platform’s perceived failure wasn’t so much its performance in the workplace as its marketing strategy. As a platform unto itself, even the most reliable and even desirable development tools were not enough to convince software developers to want to support it. Simply put, there wasn’t a migration path – no easy way for ISVs to get from 32-bit ISA support to IA-64.
The Alliance’s strategy is to build the migration path that wasn’t there before, if not technologically, then at least socially. One of its first programs will be a series of Developer Days conferences, beginning in Santa Clara, California, in November, then moving the following month to Tokyo, and to Paderborn, Germany, in February.
One company whose name is prominently missing from the Alliance roster is Dell, which announced two weeks ago it would begin dropping support for Itanium. A place at the table will be reserved for Dell, said Kirk Skaugen, Intel’s general manager for the Server Platform Group, even though, as he characterized it, they may not really belong there anyway: “When we were forming the Alliance, [Dell] chose not to participate. They are a low-end, high-commodity product, where the Itanium chip is [in the] enterprise HPC, high-availability, area.” As a result, added Skaugen, Dell has a different business model from the OEMs forming the Alliance. HP’s Howard was happy to add, “It’s not terribly surprising to us that Dell might choose not to be in the value-add and the business-critical side of enterprise computing…It doesn’t seem to be the area of their expertise.”
Skaugen also told us that in labs in which he’s taken part, where developers first try porting over existing applications from earlier RISC-based architectures to IA-64, “it’s not uncommon for them to see hundreds of percent increase, doubling the performance of their application, or higher, based on performance tuning. So what the Alliance offers is access to the world’s best performance tuning experts, and development tools to get that kind of performance out.” Skaugen noted that ISVs have related examples to him where they think their applications are fully tuned, based on their own internal processes, but after spending time with Itanium experts, realize they weren’t fully optimized after all.