johnl@esegue.segue.boston.ma.us (John R. Levine) (03/22/90)
Subject: <27842@cup.portal.com>In the following article you wrote... >A 25 MHz 486 can run software written for the 386 specifically, but a >25 MHz 386 would run the same program at about the same speed, ... Nope. A 486 accomplishes considerably more per cycle than does a 386. For example a 386 takes 6 cycles for a memory to register ADD, while a 486 takes only 2. Floating point speedups are similar, a floating add that takes 25 cycles on a 387 takes only 10 on a 486. This means that a 25 MHz 486 is much faster than a 33 MHz 386. The 486 offers few new features, it's mainly a lot faster. The only extensions to the instruction set are a few multiprocessor synchronization instructions (compare and swap), byte swap to make it easier to exchange data with big-endian machines, and some system instructions to handle the on-chip cache. It actually offers the possibility of considerably cheaper systems in the future. The 486 is far more integrated than a 386, it includes not just the function of the 80386, but also an 80387 math coprocessor, an 82385 cache controller, and 8K of static cache RAM, so an equivalent system will use fewer chips. -- John R. Levine, Segue Software, POB 349, Cambridge MA 02238, +1 617 864 9650 johnl@esegue.segue.boston.ma.us, {ima|lotus|spdcc}!esegue!johnl "Now, we are all jelly doughnuts."