[Simh] Revised FP11 debug

[Bob Supnik]

Thanks, Don. Good to know who wrote the tests!

Bitsavers has listings for 3 of the 5 11/60 FP tests, so I'm not sure 
whether it's usable. I'll try the 11/34 version, as you suggested, 
although I don't have paper tapes for that, just an XXDP pack (somewhere).

/Bob

On 3/25/2015 4:49 PM, Don North wrote:
> On 3/25/2015 8:52 AM, Bob Supnik wrote:
>> I'm looking for suggestions on how to test the revised FP11 
>> efficiently. I wrote it so long ago that all my hand tests have 
>> vanished.
>>
>> Bitsavers has a set of paper-tapes for the 11/60's FP unit, but no 
>> listings.
>>
>> XXDP FP diagnostics are "white box" tests, intended to check out the 
>> logic paths of the actual implementation. I need "black box" tests 
>> that work from the specification, without knowledge of the 
>> implementation.
>>
>> Thanks,
>>
>> /Bob Supnik
>> _______________________________________________
>> Simh mailing list
>> Simh at trailing-edge.com
>> http://mailman.trailing-edge.com/mailman/listinfo/simh
>
> Hi Bob long time,
>
> The 11/60 FPU diagnostic listing are slightly mis-filed on bitsavers, 
> they are under the CPU directory:
>
>   http://bitsavers.trailing-edge.com/pdf/dec/pdp11/1160/
>
> Rather than in the diagnostics directory as they are for the 11/34:
>
> http://bitsavers.trailing-edge.com/pdf/dec/pdp11/xxdp/diag_listings/1134/
>
> That being said, the diagnostics (I authored the 11/60 FPP diagnostics 
> a long time ago...) are by their nature
> implementation and hardware specific, especially the 11/60 FP11-E 
> specific diagnostics. Some of the system
> exerciser diagnostics take a bit more abstract architectural view, but 
> still are tuned to the specific hardware
> implementations, as that is what manufacturing/field service wanted 
> (and they paid for the diagnostics).
>
> It would have been nice to have architectural test suites for CPU, 
> FPU, MMU subsystems that would validate
> that engineering did indeed build a new PDP-11 that behaved like all 
> other PDP-11s, but those were not tests
> that existed back then (late 70s/early 80s when I was at DEC) but may 
> have come along later in the 80s/90s
> with the advent of the F11/J11 silicon. I don't know about that, after 
> my time.
>
> The CIS diagnostic test CZKEExx was a bit different; it really was a 
> white box test that only looked at the spec
> and not the specific implementations. It had an instruction emulator 
> and compared expected results from
> software simulation vs real hardware. It also built in a large number 
> of tabled test cases of known results.
>
> Based on the diagnostics info available I would suggest that there is 
> more detailed documentation available
> on the 11/34 FP11 implementation and diagnostic listing than those 
> that exist for the 11/60, so using the
> 11/34 as a baseline may be more productive.
>
> Don North
>
>
>