johnw@astroatc.UUCP (04/15/87)
In article <518@omen.UUCP> caf@.UUCP (PUT omen's NAME HERE) writes: >While it is conceivable that Motorola put the 256 byte cache in the 68020 >just to help certain benchmarks, it is more likely that the cache >actually improves performance rather inexpensively. > >Does somebody have real data on how the 68020's cache improves performance >on sieve, troff, sort, and other Unix CPU hogs? For a graduate architecture course, I did a paper that studied "locality of reference" for a large program. (I traced the running of a compiler.) My data was a list of address and reference-count pair (sorted highest to lowest ref-cnt). The results were not sruprizing. The distribution was a nice exponectioal decay, with a cumulate dist that approced 100%. As I remember the data: The 8/10/16 most used accounted for a large (but not half) the references. In the 200-300 most frequently accessed locations accounted for most (but definatly under 90%) of the references. I think the 1000 most used were in the low 90's% after 200 you have to add alot more before you see an appreciable increase in the "cumulative distribution function" If anyone wants more details, send mail....I'll sum... John W - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Name: John F. Wardale UUCP: ... {seismo | harvard | ihnp4} !uwvax!astroatc!johnw arpa: astroatc!johnw@rsch.wisc.edu snail: 5800 Cottage Gr. Rd. ;;; Madison WI 53716 audio: 608-221-9001 eXt 110 To err is human, to really foul up world news requires the net!