aceverj@accucx.cc.ruu.nl (Jaap Verhage) (06/10/90)
In article <3076@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: (among many other things): >On a a 286, it's the clock crystal divided by 2 to get the clock speed of the >CPU. On the math coprocessor, it's the crystal divided by three or four. I've often wondered about this: the speed of the math coprocessor may be lower than that of the main processor? E.g., if I've got a 12MHz 80286, I can use a 10MHz 80287 with it, or something? Would a knowledgeable soul please email me with an explanation/clarification? I'll summarize for the newsgroup.
jca@pnet01.cts.com (John C. Archambeau) (06/11/90)
aceverj@accucx.cc.ruu.nl (Jaap Verhage) writes: >In article <3076@crash.cts.com> jca@pnet01.cts.com (John C. Archambeau) writes: >(among many other things): >>On a a 286, it's the clock crystal divided by 2 to get the clock speed of the >>CPU. On the math coprocessor, it's the crystal divided by three or four. >I've often wondered about this: the speed of the math >coprocessor may be lower than that of the main processor? >E.g., if I've got a 12MHz 80286, I can use a 10MHz 80287 with >it, or something? Would a knowledgeable soul please email me >with an explanation/clarification? I'll summarize for the >newsgroup. Note that this ONLY applies to the 286 to my knowledge. On the 386(SX) the clock speeds must be the same. // JCA /* **--------------------------------------------------------------------------* ** Flames : /dev/null | Small memory model only for ** ARPANET : crash!pnet01!jca@nosc.mil | Unix? Get the (*bleep*) out ** INTERNET: jca@pnet01.cts.com | of here! ** UUCP : {nosc ucsd hplabs!hd-sdd}!crash!pnet01!jca **--------------------------------------------------------------------------* */
johnl@esegue.segue.boston.ma.us (John R. Levine) (06/11/90)
You can clock the 287 in two different ways. The default is to use the same
clock as the 286. The 286 divides the external clock rate by 2, and the 287
divides it by 3, so the 287 runs at an effective clock rate 2/3 of the 286.
Alternatively, you can provide a dedicated clock for the 287, in which case
the 287 runs at the full rate of its own clock. There is an input on the 287
called CKM which tells it which way it's being run. This is all from Intel's
286 hardware reference.
The 287 is quite slow compared to a 286: a 286 fixed point multiply takes 21
cycles, but a 287 floating multiply takes about 100. Intel wanted to make it
possible to attach a faster 287 to a 286 in order to have a more balanced
system.
Different PC clones clock their 287s differently. The IBM PS/2 series clock
the 287 directly so that the 286 and 287 both run at 10 MHz.
--
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
Marlon Brando and Doris Day were born on the same day.sonny@charybdis.harris-atd.com (Bob Davis) (06/12/90)
In article <1990Jun11.000613.6589@esegue.segue.boston.ma.us> johnl@esegue.segue.boston.ma.us (John R. Levine) writes: >You can clock the 287 in two different ways. The default is to use the same >clock as the 286. The 286 divides the external clock rate by 2, and the 287 >divides it by 3, so the 287 runs at an effective clock rate 2/3 of the 286. >Alternatively, you can provide a dedicated clock for the 287, in which case >the 287 runs at the full rate of its own clock. There is an input on the 287 >called CKM which tells it which way it's being run. This is all from Intel's >286 hardware reference. > Does anyone know why the 287 chip designers would choose to divide the Coprocessor clock by 3 if they did not have to? Why not run it at the same clock rate as the CPU? ----------------------------------------------------------------------------- Bob Davis \\\ INTERNET : sonny@trantor.harris-atd.com Harris Corporation, ESS \\\ UUCP : ...!uunet!x102a!trantor!sonny Advanced Technology Department \\\ AETHER : K4VNO PO Box 37, MS 3A/1912 \\\ VOICE : (407) 727-5886 Melbourne, FL 32902 \\\ FAX : (407) 729-2537