re: http://www.garlic.com/~lynn/2014h.html#2 Demonstrating Moore's law http://www.garlic.com/~lynn/2014h.html#4 Demonstrating Moore's law http://www.garlic.com/~lynn/2014h.html#5 Demonstrating Moore's law
part of the issue of applying Moore's law to mainframes was that they had more like 7-8yr cycle ... rather than 18m ... as well as the FS period in the first half of the 70s when all 370 work was being shutdown ... and then again in the first half of the 90s when the company had gone into the red (email from people in POK would include tagline asking the last person to leave POK, to please turn out the lights). In the wake of the FS failure ... 3033 & 3081 were both kicked-off in parallel, the really fast effort to remap 168-3 logic to 20% faster chips (for 3033) and longer effort for 3081 (but both using warmed over FS technology) ... more detail http://www.jfsowa.com/computer/memo125.htm nearly as soon as the 3033 was out the door ... ships spring 1978 http://www-03.ibm.com/ibm/history/exhibits/3033/3033_CH01.html the 3033 processor engineers start on 3090 ... spring 1985 (1st really "new" 370 effort since death of FS in the mid 70s) http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_PP3090.html 1990, the us auto industry has the C4 taskforce to completely remake the industry ... because they plan on heavily leveraging technology, several technology vendors are asked to participate. they explain the US auto industry has been on 7-8yr product cycles, running two in parallel offset by 3-4yrs ... with minor cosmetic changes in intervening years. the trouble is that foreign imports had cut their product cycle to 3-4yrs in the early 80s and by 1990 were about to cut to under 2yrs (providing enormous competitive advantage). Offline at the C4 meetings I would kid the mainframe participants from POK how they were going to help since they had the same problem. acs/360 had been killed in the 60s because management was afraid that it would advance the state-of-the-art too fast and they would loose control of the market. http://people.cs.clemson.edu/~mark/acs_end.html finally over 20yrs later, acs/360 features start showing up with the es/9000 in 1990. http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_FS9000.html the approx equivalent Intel process is now called tick-tock (every 18m one tick or one tock) http://en.wikipedia.org/wiki/Intel_Tick-Tock trivia: the auto import quotas from 1980 were put in place to significant reduce competition and significantly improve US industry profit that they would use to completely remake themselves ... however they just pocketed the money and continued business as usual. the 1990 c4 was another attempt to completely remake themselves ... but there was so many stakeholders in the status-quo that little changed. recent references are that even with the bailouts, nothing has significantly changed. other trivia: I'm part of a group that visit POK in the mid-70s with design for 16-way 370 multiprocessor and get the 3033 processor engineers to work on it in their spare time (lot more interesting than 3033) and everybody really thinks it is fantastic. Then somebody tells the head of POK that it could be decades befor the POK favorite son operating system has useful 16-way support. He then asks some of us to never visit POK again and tells the 3033 processor engineers to keep their nose to the grindstone (and never be distracted again). they are up to 12 in 1999 http://en.wikipedia.org/wiki/IBM_ESA/390 and z900 finally hits 16 last month of 2000 (almost 25yrs later) -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
