Intel is Continuing the Yamhill Project?

by Anton Shilov
08/04/2003 | 08:22 AM

An analyst from American Technology Research Corporation wrote in a recent report that Intel Corporation is proceeding with the development of x86-64-like technology code-named Yamhill. Last year there were some unofficial claims that Intel had cancelled the Yamhill project to fully concentrate on pushing the IA64 technology into the market.

The report written by Rick Whittington for an investment banking firm American Technology Research Corporation and quoted here says that Intel is still performing research and development on the Yamhill project and is planning to implement the technology into its next-generation microprocessors for server and desktop markets.

In fact, there were two contradicting rumours on the idea and the conception of the Yamhill technology. One group thought that the Yamhill is a set of extensions that may be even based on the x86-64 technology; another believed that Yamhill is a powerful IA32 emulation tool for IA64 chips.

As we mentioned a year ago, the implementation of the Yamhill technology depends on the success of the x86-64 architecture of AMD Athlon 64 and Opteron processors. The history already knows examples when Intel implemented AMD-like technologies after the initial success of techniques introduced by the Sunnyvale, California-based firm. For instance, after AMD launched 3DNow! Technology with its K6-2 processor, Intel quickly realised the benefits SIMD-instructions can bring and introduced SSE in order to improve the Pentium III processor and also to compete with AMD’s 3DNow!. The battle of these two innovations is the Intel won: most of ISVs optimised their products for Intel Pentium III CPUs. In general, I expect the same to happen with the x86-64 technology with the exception that there will be compilers and software optimised for AMD architecture.

To sum up, the Yamhill project will not be cancelled at least until Intel has chances to offer an inexpensive IA64 processor for desktop computers provided there is enough software for it on the market. It would be quite realistic to expect Intel to implement the Yamhill in very late 2004 or in sometime in 2005 in the high-end Tejas or even Nehalem processors. By that time it will be clear whether the x86-64 is a success or not for AMD and a necessary or unnecessary feature for Intel’s microprocessors.