[comp.sys.amiga.tech] Memory cycles

pkontkanen@cc.helsinki.fi (01/12/90)

	
	I have read one memory cycle is approximately 280 ns in duration.
 	Now I would like to know how many cycles each machine language
	instruction takes. Is the number of cycles same as the number of
	words the instruction allocates?

					Petri Kontkanen
					The University of Helsinki,Finland	


 

MARKV@kuhub.cc.ukans.edu (MARK GOODERUM - UNIV. OF KANSAS ACS - MARKV@UKANVAX) (01/13/90)

> 	I have read one memory cycle is approximately 280 ns in duration.

Actually it is only 140 ns, but in CHIP RAM 68000 gets every other cycle, but
in FAST RAM it gets every cycle.  However lots of Amiga memory boards are 
1 wait state, which means 280 ns read/write cycle.  (Pretty easy to tell,
if your memory board must have 120 ns chips then you are 0 ws, 150 is 1 ws).

>  	Now I would like to know how many cycles each machine language
> 	instruction takes. Is the number of cycles same as the number of
> 	words the instruction allocates?

Actually it takes a certain number of cycles for the instruction plus
the number of cycles for read and write access.  The number of cycles vary
for each instruction.  The values are available in the Motorola Reference
manuals (which are reprinted in many 68000 assembly language books.)

Note that the specific timings vary for each member of the 68000 family.
Also bus size effects speed since a 68000 would take 2 read cycles to fetch
a LONGWORD, but a 68020 w/32 bit RAM only takes 1.  The situation gets 
even more confusing with the 68020/30 and their caches and pipelining.

The net point is, instruction timings are interesting for curiosities sake
and maybe comparing the relative merit of differnt chips, but on the 
Amiga you should NEVER NEVER NEVER count on the timings of instructions
for anything.  (Motorola doesn't even guarentee the timings will remain the
same for the same chip (for instance some newer 68010/12's have faster 
timings then other 68010)).

> 					Petri Kontkanen
> 					The University of Helsinki,Finland
 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Mark Gooderum			Only...		\    Good Cheer !!!
Academic Computing Services	       ///	  \___________________________
University of Kansas		     ///  /|         __    _
Bix:	  markgood	      \\\  ///  /__| |\/| | | _   /_\  makes it
Bitnet:   MARKV@UKANVAX		\/\/  /    | |  | | |__| /   \ possible...
Internet: mark@kuhub.cc.ukans.edu
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

bdb@becker.UUCP (Bruce Becker) (01/14/90)

In article <1728.25ad4906@cc.helsinki.fi> pkontkanen@cc.helsinki.fi writes:
|	
|	I have read one memory cycle is approximately 280 ns in duration.
| 	Now I would like to know how many cycles each machine language
|	instruction takes. Is the number of cycles same as the number of
|	words the instruction allocates?

	The Motorola MC680x0 proccessor manuals give
	this information in an appendix.

	The number of cycles are dependent on several
	things, too complex to enumerate here - it is
	worth pointing out in passing that as "x" gets
	larger in 680x0, the cycle time for many
	instructions gets shorter due to better processor
	internal architecture and improved microcode.

Cheers,
-- 
  ,,,,	 Bruce Becker	Toronto, Ont.
w \$$/	 Internet: bdb@becker.UUCP, bruce@gpu.utcs.toronto.edu
 `/c/-e	 BitNet:   BECKER@HUMBER.BITNET
_/  >_	 "Money is the root of all money" - Adam