Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!uwm.edu!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!decwrl!shelby!portia!dhinds From: dhinds@portia.Stanford.EDU (David Hinds) Newsgroups: comp.sys.ibm.pc Subject: Re: Brain-dead 286 - summary Message-ID: <9982@portia.Stanford.EDU> Date: 9 Mar 90 05:01:05 GMT References: <8681@rosevax.Rosemount.COM> <29405@amdcad.AMD.COM> <38299@iuvax.cs.indiana.edu> Sender: David Hinds Distribution: usa Organization: Stanford University Lines: 40 In article <38299@iuvax.cs.indiana.edu>, sl197009@silver.ucs.indiana.edu (Chima Echeruo) writes: > > I have read many "experts" saying that the 286 is far slower than a 386. > ... won't a 25Mhz 286 outperform an equivalent 386 while in real mode? From what I've seen, a 16MHz 286 runs DOS code at about the same speed as a 16MHz 386SX, +/- 5% or so. The instruction timings are about the same, and the bus widths are both 16 bits. A 16MHz 386 should be consistently faster, even running 16-bit code and without a RAM cache. The reason is that even if nothing else uses the 32-bit bus bandwidth, the instruction prefetch unit on the 386 always loads whole doublewords. I think the effect of this is something like a 20% improvement in speed over a same-speed 286. The 286 also does not support memory interleaving systems, so for >16MHz, you will be forced to have wait states on main memory. A 20MHz 386 can still work at "near 0 wait states" with interleaved memory. At higher speeds, a static RAM cache is necessary for avoiding wait states. I don't know if there are cache controllers for the 286, but I'm pretty sure the Intel cache controller only works with 386's. > > I do not think that the 286 is dead, maybe it is not the state-of-the-art in > computer technology but it provides a very *fast* system for the average > computer user. If the SX does not go past 16Mhz, I do not see the benefit for > upgrading to a 386sx. > I agree that fast 286's are perfectly good machines for running almost all DOS applications, and are guaranteed to have a high level of support for many years due to the huge number of systems around. As a sort of souped-up 8086, I think it still has a very legitimate market. I think that the 386SX puts a LOT of pressure on new fast 286's (say 16MHz or higher), because the price differential is quite small, both will run DOS code about equally well, but the 386SX offers a bit more. If I had a fast 286 machine, it certainly would not be worth it to me to "upgrade" to a 386SX, but I wouldn't buy another 286. By the way, Intel has already announced the 20MHz 386SX, and a version of their cache controller for the SX. A 20MHz 386SX with memory interleaving or a cache would be significantly faster than a 20MHz 286 with the unavoidable wait states. -David Hinds dhinds@popserver.stanford.edu