🚀 go-pugleaf

In article <17fb61INN572@agate.berkeley.edu> izumi@pinoko.berkeley.edu (Izumi Ohzawa) writes:
>In article <1992Aug26.060638.10407@ee.ubc.ca>
>  williamb@ee.ubc.ca (william burchill) writes:
>>A floating bus is not going to confuse a memory testing program, that's
>>what it's looking for.  What do you mean "didn't work well"?
>
>Well, at the risk of making myself look totally stupid, the following is a
>simplified version of what I am trying to do.	I do use other test patterns
>in addition to $5a5a5a.  Failure on any one pattern should break the loop.
>..
>
>ERAM_Base	equ	8192		; Start address of external DSP RAM on NeXT
>
>	<some init stuff>
>
>	move	#>ERAM_Base,R0
>TestLoop
>	move	#>$5a5a5a,b		;load test pattern
>	move	b,x:(R0)		;write to a memory loc
>	move	x:(R0)+,a		;read back into A1, post-incr address.
>	cmp	a,b
>	jne	memory_fail
>
>	<same things with different test patterns>

    *NEVER* read back a memory location just after writing it!	If the
    databus is floating it will generally hold on to the previously
    written value.

    I would suggest the following:

    (1) Write pattern to (R0)
    (2) Write  a DIFFERENT pattern to the next address
    (3) Read back the FIRST address and test for the original pattern

    Generally you want to go through memory several times using different
    patterns.

					-Matt

>	jmp	TestLoop
>
>memory_fail
>	<report R0 value to host>
>
>--
>Izumi Ohzawa
>USMail: University of California, 360 Minor Hall, Berkeley, CA 94720
>Telephone: (510) 642-6440     Fax:  (510) 642-3323
>Internet: izumi@pinoko.berkeley.edu (NeXT Mail OK)

--

    Matthew Dillon	    dillon@Overload.Berkeley.CA.US
    891 Regal Rd.	    uunet.uu.net!overload!dillon
    Berkeley, Ca. 94708     ham: KC6LVW (no mail drop)
    USA