An update on Intel Itanium processors

I get asked regularly about Intel’s Itanium strategy and plans.

Having taken over responsibility for Itanium at Intel in 2005, I spent time taking a deep and careful look at the technology, products and positioning of the product in the market place and what was required to make it successful.

The first thing that became apparent was, our execution had been poor. Numerous project cancelations and many product slips. Our credibility was damaged and our customer’s ability to consistently deliver winning products was hampered. Originally we had a complex shared development model with HP which was inefficient and introduced a lot of room for mis-steps. In early 2005 we combined all the silicon teams into a common development group at Intel to have singular accountable engineering ownership with common methods, tools, process files, etc. The results have been dramatic improvements in roadmap execution. In fact, just last week I was pleased to open the Ft Collins design center for Intel, a $60M facility that houses the design team that was formally at HP.

It was also apparent we needed to better marshall the efforts of the numerous industry players in order to drive critical mass around development efforts, ecosystem enabling and market momentum. In late-2005, we launched the Itanium Solutions Alliance which pulls together the interests of all those invested in itanium from OEMs like HP, NEC, Fujitsu to ISVs like Microsoft, Oracle and SAP to System Integrators like EDS. These players collectively committed to delivering $10 Billion in combined industry investments in Itanium through the end of the decade. The results have been a clear shift in both perception and impact in the industry. For instance, ISA has contributed to getting over 12,000 applicatons ported to Itanium. At JavaOne two weeks ago ISA announced that they would be working with Sun to deliver Sun’s Java on IPF.

We often get asked about the roadmap and our commitment to Itanium. The Dual Core Intel® Itanium® 2 processor 9000 (Montecito) has ramped extremely well after being launch in Q2 last year. Itanium in general experienced over 2x unit growth between 2H05 and 2H06 hitting over $1B in system revenue in Q4’06. Montvale, the next stepping of Montecito, will launch and ramp later this year. Tukwila, a new generation of Itanium which includes critical advances in system architecture is planned for late next year. Each generation continues to offer great scalable performance and RAS as critical differentiators for a product designed for mission-critical, data-intensive workloads.

The results are being felt in the market place with Itanium systems competing very effectively with SPARC, Power and other mainframe-based solutions. Itanium system revenue is now over 63% of SPARC system revenue and over 54% of Power system revenue

Finally, I get asked about why not just focus on Xeon? Xeon is well addressing the needs of the $28B volume server market. This is the fastest growing portion of the server market. However, the mainframe and risc replacement market remains about the same size at $28B and is growing, albeit more slowly than the volume server marketing. Itanium plays a crucial role in the our pursuit of this market — with it’s best in class reliability and scalable performance. It competes head to head with Power and SPARC, creating an open system alternative to these other proprietary high end architectures.

In summary, we are incredibly committed to the Itanium roadmap, with multiple generations in development, and we look forward to continued industry momentum in the years ahead.



7 Responses to An update on Intel Itanium processors

  1. dingwei says:

    I think epic is idealism, instrction level parallel is imposible for OLTP or business computing. Now,Multi-Core will solve the problem. Montecito is the first one.

  2. Igor says:

    Nice to see Pat posting here. I apreciate the new info about Itanium even though as a developer I never had a chance to work on one.

  3. We have had spectacular results with the Itanium so far. Used as a integrated environment for hosting the GBM (Banking Database and Datawarehouse) on Oracle 10G running on Windows Server for the Itanium.
    This machine shunts data. It comfortably handles a massive set of ETL processeses, sourcing over 60 systems in the largest banking in southern hemisphere. More that 20 Million Client/Stakeholders, all their related Deals data (80 Million) as well as all the related Accounting, Asset, Risk, Impairment Data.
    So we are extremely happy that this technology has a future.
    It baffles me why the market compares it with the XEON and AMD 64 bit (microprocessor technology). It simply is not the same !!
    This is a wonderful machine !!!

  4. Wim Rombauts says:

    Why people compare it to XEON and AMD64 ? You have to compare to something and the x86 processors are still the “standard” or “default” processors on the market, and it are especially the x86 fans that question the Itanium.
    Despite the fact that the Itanium may have a superior architecture to x86 designs, intel hasn’t succeeded yet in letting it show and leaving the x86 way behind in performance results or price/performance comparisons.
    I hope Tukwila and Poulson will finally do this, or Itanium may yet enter a dead-end road.

  5. Paul C says:

    Xeon and AMD64 are undoubtedly standards, however you really cannot compare them to SPARC and POWER right? – These are the true competitors for Itanium.