Microsoft killed the PowerPC 615
IBM engineer offers up 615 information
An insider at IBM has responded to the PowerPC 615 story we wrote yesterday with some hard facts -- and corrections --that cannot be ignored.
The insider said that Motorola, although it knew of the programme, was never involved and Somerset and IBM Austin were not in the picture too. The IBM Micro team at Burlington performed the 615 miracle, with 250 engineers involved. Even after the programme finished, he said, 40 or so remained at that site.
Some went to work on the ASIC ARM and TI DSP cores, while others went to Centaur and the Great Stan of Taperecorders (AMD). While the first pass of hardware performed reasonably, the engineer said that pass two was killed and that level of silicon was never tested. Many of the 250 kept the pass one chips as mementoes. A
version of Minix was written to run in PowerPC mode and launch x.86 and PowerPC applications. It proved the concept and a version of OS/2 for the platform also worked. But Microsoft got very cross when it was asked to write a version of Windows to run in the PowerPC mode, it emerged. The chip used .35 micron technology and measured 16.5 by 20 millimeters.
According to the source, the VP in charge of the project was sure that the Pentium II would be an IA 64 design and that prompted the die size. If he had not been so sure of that, the die would have been much smaller, the engineer said. The chip, now presumably in keyrings, pendants, earrings and the like, supported x.86 architecture, 32-bit PowerPC architecture and the 64-bit PowerPc architecture.
The original PowerPC 64 design was not backwards compatible from an OS point of view. Further, a mode switch from PowerPC to x.86 or vice versa took five CPU clocks. That was the time it took for the pipe to empty one architecture's instructions and fetch the new architecture's instructions. The mode switch, despite press reports, did not take an eternity, but the engineer said the US press did not help... The programme was killed because someone realised it would never make money. That was Microsoft's fault, the engineer added. Microsoft would not give any level of support. ®