They're confusing the 486SX/DX difference with the 386SX/DX difference. In the 386, it was the bus. 16bit/32bit In the 486, it was the FPU. NoFPU/FPU Intel re-used the same suffix from the 386's, but changed the meaning. As I can see from this discussion, the confusion continues to this day. There are no bus differences on 486's. The DX can be used in the same mobo's as the SX. The same can not be said of the 386 CPUs. Of course, the LIF/ZIF sockets didn't come about until the 486, so it wasn't like you had to worry about getting the wrong mobo with your CPU on the 386.
rowland wrote: > On Tuesday 15 Oct 2002 2:56 am, Leon Brooks wrote: > >>On Tuesday 15 October 2002 04:18 am, rowland wrote: >> >>>On Monday 14 Oct 2002 11:23 am, J. Greenlees wrote: >>> >>>>Thierry Vignaud wrote: >>>> >>>>>also, the 486sx (at least the first ones) did has a coprocessor; it >>>>>was disabled but was still there (though i don't rember if it was >>>>>missing pins or some silicon hack). >>>> >>>>actually, it was a bad bit of circuit if I remember correctly, the co >>>>pro was completely un-usable because of it and the cpu was a lower >>>>price for that reason. >>> >>>if I remember rightly it was a batch of i486dx's that had this problem, >>>the fpu just couldnt add up properly given the right set of circumstances >>>and intel had to change all the affected chips! >> >>486sx was 486dx sans FPU and on a skinnier buss. To add an FPU, you bought >>a 487sx chip, which was really a 486dx that had failed some factory tests >>and been packaged for the skinnier buss. For a little while, some >>motherboards had an option to run with _only_ a 487sx, because they were >>significantly cheaper than a 486sx and usually worked fine (sometimes >>faster, because a 486sx had no CPU cache at first but many of the 487sxes >>did). >> >>Cheers; Leon > > dont see how 486sx could have 'skinnier bus' seeing as how the same > motherboard socket could take either a sx or dx chip > rowland >
