November 27, 2020

Reed: LaGrande Another 432?

David Reed has an interesting perspective on Intel’s LaGrande proposal.

Reed likens LaGrande to the Intel 432 processor. Few non-techies have heard of the 432, but in the processor-design community the 432 is a legendary failure. As Reed says, the 432 was “Intel’s attempt to create an ‘object oriented’ processor that would embed all the great ideas of object oriented computing in a revolutionary new architecture.”

The 432 died because it tried to build into hardware ideas that were still under development. Of all the parts of a computer system, the hardware is the most expensive to change, and the most difficult. It follows that you only want to put a particular function in hardware if you know that that function is necessary, and you know exactly how to do it. Because if you decide a year later that you want to do it differently, you’re out of luck. Hardware is much harder to change than software.

The Japanese “Fifth Generation” project from the 80’s is another example of a disaster caused by committing too early to a speculative design approach. Fifth Generation was going to reorganize the computing world around logic-based programming. This seemed like a good idea at first, but when it became evident that the right answer lay elsewhere, it was too late to reorient the project.

Reed has a good point, but I think he goes too far. The 432 and the Fifth Generation were both radical departures from existing practice; they wanted to tear up and redesign the whole processor. LaGrande seems much less ambitious. But Reed is right on target in saying that building security features into processor hardware is a risky engineering decision.

Comments

  1. The iAPX432

    Ed Felten has an interesting piece on Intel’s La Grande processor, and its “predecessor” the iAPX432. I was involved in…

  2. The iAPX432

    Ed Felten has an interesting piece on Intel’s La Grande processor, and its "predecessor" the iAPX432. I was involved in…