The Intel and MS wrangle over MMX IP
Not the right time to help Intel make x86 more proprietary - Maritz
Intel and Microsoft were locked in wrangles over NSP and Java in 1995-96, and Microsoft has denied pressuring Intel to drop NSP and downplay Java. But internal Intel and Microsoft documentation from the period points to a situation rather different from what either company has been claiming - basically, they were wrangling over intellectual property. This casts a rather different light on Paul Maritz's suggestion in his deposition (Intel could sue OEMs over Katmai, IA-64 that Intel has been extending its IP control from MMX onwards so that it will be able to sue rival CPU companies. The Microsoft documentation does indeed strongly suggest that this is the case - but it also makes it clear that Microsoft was prepared to help Intel do so, if the terms were right. In May 1995 Paul Maritz wrote to Bill Gates: "It is clear that their [Intel's] strategy is to get as widespread usage as possible of their DSP algorithms via the NSP software layer, and then accelerate it via the 'few SIMD' instructions in the P55C (which they have been careful not to give us or others IP protection on." Maritz therefore sees Intel's NSP technology as a potential Trojan Horse that will allow Intel to spread its technology through the market, and then put a lock on it via P55C (i.e. MMX) IP protection lawsuits. Given the tutting tone of Maritz's deposition on this very subject, we can surely presume that he's agin this. No, not entirely: "given other dynamics this may not be the right time to help Intel start to make the x86 instruction set more proprietary." Which does make it seem awfully likely that there might be a right time, some other time. Note that the earliest PC9x documents, which were jointly devised by Intel and Microsoft, and which were current while OS/2 was a threat, were to a great extent tied to Windows operating software - so locking other people out is a concept not entirely unknown to and unpractised by MS. A year later deals were still possible, and Maritz writes: "I discussed with brasi [Brad Silverberg] whether it might not be wise for us to accommodate Intel in some way wrt [with reference to] the 'process' in return for them getting 100% behind our architecture... The reason for doing this is that I think we need to have some way of slowing down/containing Netscape - I fear that they can use security to screw us in a big way." Microsoft at the time had been pursuing an independent security strategy called WinSec, while Intel wanted to be more industry standard. So a deal over this is perhaps being suggested - Microsoft comes on-side, Intel backs away from Netscape. Tantalisingly, this particular exhibit has been heavily legaled - a whole page and a half from Brad Silverberg which may or may not explain what's going on has been removed. In June of the same year, Gates' thoughts have turned to Merced IP, and he writes to Maritz: "Merced - I told Andy [Grove] that we were quite optimistic things will get resolved with HP in the next few weeks and then we will move quickly to sign the Merced LOI... Intel is disappointed we seem to think that their 98 date has a lot of risk. They don't feel that it does. Andy wanted to understand our DEC 64bit announcement." So here we are still circling warily over IP, and Alpha is being used as leverage. Whatever was being resolved with HP didn't get resolved though. Maritz claims in his deposition that HP and Intel have specifically declined to give Microsoft IP assurances over Merced. Gates' memo here makes it clear that there this was the sticking point in the deal Microsoft was trying to negotiate. ® Complete Register trial coverage
Sponsored: RAID: End of an era?