scott.r...@joann.com (Scott Rowe) writes: > If it bothers you that much you are going to have to go back to running on > older (slower) hardware. It is just not possible without throwing out all > the innovative features that allow current processors to run at the speeds > they do. If you think it's bad now, it will get a lot worse if/when IBM > introduces SMT in mainframe CPUs.
there was project to do threading for 370/195 (that never shipped). 195 had pipeline and peaked around 10MIPs for carefully crafted code. However, branches stalled the pipeline (modulo special case looping within pipeline); no speculative execution, etc. ... so most codes ran about 5mips (still slightly faster than 3033, which was about 1.5 times 168-3 or 4.5mips). the effort was to duplicate psw and registers to simulate 2-way multiprocessor w/o actually replicating any other hardware (instructions in pipeline would have one bit flag to identify which instruction stream instruction/regs/etc belonged to) ... two 5mip instruction streams utlizing peak 10mips with only modest additional hardware have to settle for 360/195 functional characteristics from bitsavers (very similar to 370/195) http://www.bitsavers.org/pdf/ibm/360/funcChar/A22-6943-0_360-195_funChar.pdf 3081d was supposedly approx. 5mips ... but for some things it ran 20% slower than 3033. 3081k doubled the cache size and was supposedly approx. 7mips ... but for some number of things ran nearly same as on 3033. it isn't until 3090 that you really start to match 195, more than 15 yrs earlier; announce aug1969 http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_PP2195.html 370/195 http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_2423PH3195.html 3090 announce feb1985: http://www-03.ibm.com/ibm/history/exhibits/mainframe/mainframe_PP3090.html big part was the side trip into FS and the disastrous failure (including shutting down lots of 370 development during the FS years). With the failure of FS, there was mad rush to get products back into the 370 pipeline. 3033 started out as 168-3 remapped to 20% faster chips (chips also had 10 times as many circuits/chip, but the extra circuits started out going unused). During 3033 development, some of the critical 168-3 parts was redone to leverage higher circuit/chip density eventually reaching 1.5times 168-3. In parallel with 3033 there was 3081 which was basically the FS 370 emulator ... some amount of details here: http://www.jfsowa.com/computer/memo125.htm from above: The 370 emulator minus the FS microcode was eventually sold in 1980 as as the IBM 3081. The ratio of the amount of circuitry in the 3081 to its performance was significantly worse than other IBM systems of the time; its price/performance ratio wasn't quite so bad because IBM had to cut the price to be competitive. The major competition at the time was from Amdahl Systems -- a company founded by Gene Amdahl, who left IBM shortly before the FS project began, when his plans for the Advanced Computer System (ACS) were killed. The Amdahl machine was indeed superior to the 3081 in price/performance and spectaculary superior in terms of performance compared to the amount of circuitry.] ... snip ... misc. past posts mentioning FS http://www.garlic.com/~lynn/submain.html#futuresys -- virtualization experience starting Jan1968, 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
