mw...@ssfcu.org (Ward, Mike S) writes: > Yes, I understand. If I wanted to change processors I would also go for > the fastest cycle times. I remember a company that had a 400 mip(single > engine) machine which then purchased a 600 mip 3 engine machine (200 > mips per engine). (mips and engines are fictitious to protect the > innocent :)) They were sadly disappointed because now the machine was > slower even though they had more mips.
this is been frequent major refrain in the PC industry (almost no progress in parallelizing technology for the past several decades) as they hit the Ghz wall and started moving to multiple cores. There have been comments that nearly all of the embarrasingly parallel applications had already been done decades ago. there was big step-forward with the compare&swap instruction. Charlie had invented it at the science center working on CP67 fine-grain multiprocessor locking (compare&swap was chosen because CAS is charlie's initials). misc. past posts mentioning smp &/or compare&swap http://www.garlic.com/~lynn/subtopic.html#smp initial forey into POK to get compare&swap added to 370 architecture was rebuffed, claiming that the favorite son operating system felt that nothing more than test&set (from 360 multiprocessing) was required. challenge given the science center was to come up with a non-multiprocessor specific use for compare&swap. the result was the examples of multi-threaded use that still in current principles of operation ... where the multi-threaded (aka multiprogramming) operation is independent of whether environment was single processor or multiprocessor. starting at least by the early 80s ... compare&swap saw major uptake in transaction and multithreaded DBMS implementations (with the same or similar construct showing up on all the major hardware platforms) ... aka example is the original relational/sql implementation ... misc. past posts mentioning system/r http://www.garlic.com/~lynn/subtopic.html#systemr -- 40+yrs virtualization experience (since Jan68), online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
