origin of gates quote on 640k memory...

From: Neil Morrison <morrison(a)t-iii.com> Thats so far out left field that I can say there is little basis for it. RS was not a significant vendor for memory cards or extensions. As to the choice of 8088 vs 8080, nobrainer. Everyone knew by 1980 that while the z80 was a great cpu there were an emerging class of apps that really needed 16 (or more) bit wide ALUs and memory to perform well. Graphics was one of the emerging apps and the other was programs like multiplan (math intensive). These and others were the push for the 16bit world. Additionally 8080 was quite passe` by 1980, 8085 maybe as it was available as a 5mhz part (vs 8080 at 3!) and Z80 at 4mhz. There was also Z8000 and other promised parts for the 16bit world to come. I suspect 8088 won as it was quite easy to interface and use. Another point, Multiplan ran under CP/M with 48-60k of available memory and most 8080 class (include z80) didn't have a MMU to manage more nor did the OS (CPM2) have management. I have several systems that still run multiplan (8080 and z80), they used overlays and on disk storage very intensly to get around memory shortfalls. Additionally PC even early on were rift with code bloat (8080 lofted code tended to expand) and it was deemed a requirement to have more than 64k mostly due to the OS eating most of it. At the time 256k was enough and it was felt (just like a few years before when z80 64k was plenty) that was enough and if not 640k would insure enough space. This is true also due to the segmented addressing the 8088 used, it was still largely a 8/16bit cpu with 16bit basic addressing. Apps didn't grow well until the 286 memory style and more aptly the 32bit 386 appeared. An aside, non PC 8088/86 systems used to permit a full 1 MB of ram via rom shadowing or added a mmu to extend beyond 1mb. So Billy boy was really refering to those system that were promoting the full 1mb or larger memory maps. Allison