The Development of PA-RISC

PA-RISC proved to be a boon for Hewlett-Packard through the rest of the 80's and into the 90's, as it coincided with the trend to open client/server computing. The reliance of this sort of computing on the UNIX operating system allowed HP to establish its own brand, HP-UX, as a major player in the UNIX world.

Former CEO Lew Platt noted that there would be several opportunities to enhance PA-RISC, but that at some point it would 'run out of steam' and HP would need a next generation architecture. To that end, a team was created to work on this very possibility. Platt notes that the team was made up of people “whom we had acquired from some organizations that were working on very advanced computing techniques. So we brought in some really world class experts who...began work on the next generation architecture.”

Joel Birnbaum, HP's Chief Technical Officer and General Manager of the HP labs at the time, noted in his presentation at the Microprocessor Forum in 1997 that the designers in the lab were able to extend the life of the PA-RISC architecture by adding several new innovations, such as out-of-order execution. Concern over the increasing level of design complexity began to grow again. This was because excessive complexity could prove limiting due to the now large base of installed PA-RISC systems. The solution found was known internally at Hewlett-Packard as 'wide word,' and that turned out to influence the next architectural differentiation at the system-wide level in the following decade.

CEO Platt remembers that at the start, 'wide word'—which would eventually evolve into the Itanium architecture—was one that was thought of as too far ahead of the curve initially. Yet he mentions, in an instance of foresight, that “if you're going to make a mistake, probably a good one to make is to be there with a new architecture ahead of when you need it, instead of after.”

As the project began to develop beyond its initial stages it became obvious that the project's special architecture had tremendous potential. Beyond the synergy that resulted from the partnership with Intel, Platt noted that it would “represent a large step forward from where we were with PA-RISC at that point in time, and it really had a very long future ahead of it.”