It Seemed Like a Good Idea at the Time
The direct technological roots of IA-64 go back to a time when the computing world was eagerly awaiting Intel’s exciting new million transistor processor, the i486. In early 1989 a group was set up in HP Labs to adapting VLIW technology to general purpose computing applications . That effort soon merged with a separate group looking to find a successor to HP’s PA-RISC architecture. The joint program resulted in the development of an instruction set architecture called Precision Architecture – Wide Word (PA-WW). PA-WW was the predecessor of the technology that years later would be named EPIC (Explicitly Parallel Instruction Computing) by Intel’s marketing department.
HP first approached Intel in late 1993, even before PA-WW was finalized. The reason was simple, HP wisely realized it could not stay in the MPU manufacturing business. The cost of process and fab development was growing geometrically with each new process generation and even a well managed and successful company like HP could not keep paying for it from the sales of proprietary hardware. This was the same conclusion that Digital Equipment Corporation (DEC) came to several years later but not before critically damaging its balance sheet and entering a period of decline that ultimately resulted in its acquisition by Compaq.
Ironically, both HP’s and DEC’s chip manufacturing woes were ultimately solved by Intel. In DEC’s case their Hudson fab was purchased by Intel as one of the terms in a complex out-of-court settlement of a patent infringement suit DEC launched against Intel in May, 1997. Reflective of the famous ‘HP way’ of non-confrontational consensus building, HP took the more diplomatic approach of offering its PA-WW technology and IP to Intel in exchange for the inclusion of minor features to ease the transition of its PA-RISC customer base to the new architecture. For Intel, an organization with little demonstrable aptitude for creating praiseworthy processor architectures, HP’s offer of PA-WW in exchange for little more than rights to buy chips from Intel must have been quite appealing. The still open question is, did Intel bet on the right 64-bit horse? Is EPIC the future of computing or an evolutionary dinosaur, unable to thrive amid nimbler and more efficient competitors?
Be the first to discuss this article!