Subject: RE: Interesting item on eBay web site item#1233065603: Digital Eq
To: 'Dave McGuire' <firstname.lastname@example.org>
From: Carlini, Antonio <Antonio.Carlini@riverstonenet.com>
Date: 04/28/2001 15:54:21
> Dave McGuire wrote:
>On April 28, Carlini, Antonio wrote:
>> The MicroVAX II was clocked at 40MHz (IIRC) and had a cycle
>> of 200ns (no cache needed since main memory was able
>> to meet this).
> I had to think twice before disagreeing with Antonio...but as I
>the 40MHz oscillator on a MicroVAX-II CPU board is passed through a
>flip-flip to divide it by 2 before being used. All of the docs
>ever seen have always said that the 78032 chip is clocked at 20MHz,
>and the onboard oscillator is definitely 40MHz (because I've
>a few with 50MHz oscillators)...if I get motivated enough I'll go
>a board and try to find that flip-flip, but I'm very sure that the
>processor is actually run at 20MHz.
The 40MHz was just from memory but now I've gone and
looked. However, few of the manuals I have mention the
gory details of the KA630 (the manuals from that era seemed
to oriented to explaining jumpers and suchlike rather than
details of how the modules worked).
(I'm curious which docs you have that mention a clock
frequency ... I had a hard time finding any in the uvax II
tech manuals of the era).
The 78032 User Guide indicated that
the CLKI signal could range from a cycle time of 25ns to
250ns (i.e. 40MHz down to 4MHz).
None of the various uVAX/VS II hardware manuals mentioned
very much about the KA630 board itself. (If only I had
a KA630 UG ...).
I finally found a reference in DTJ No. 6, which deals
exclusively with the MicroVAX II (and is sadly from the period
before they were published online).
In the first article, which discusses the 78032 chip,
the clock input is stated to be 40MHz.
From that the chip produces (internally) eight 25ns clock
phases that control its timing (hence the 200ns CPU cycle).
The third article, which discusses the KA630 module,
states that the CLKI signal is 40MHz. The chip halves
this to produce a CLKO signal for the rest of the
board running at 20MHz.
While I have this journal open, I should clarify that
although the board runs at a microcycle speed of
200ns as I claimed earlier, memory cycles take
two microcycles so the memory timing is 400ns.
Over the time that I have been gathering data about
VAXen, I've found plenty of conflicting information.
Usually this is performance related stuff but clocks
do sometimes fall into that category too. For example,
according to one of the VAX 4000 Technical Information
manuals I've just pulled off the shelf, the KA694
(VAX 4000-705A CPU module) runs at 444MHz !!!
and achieves a cycle time of 9ns. I don't doubt that
there is a 444MHz clock somewhere (possibly inside
the chip after it has multiplied up the supplied
clock) but clearly its just used because it was
more convenient to have a high clock and divide
by four to generate the necessary timing signals.
I suppose marketing wanted to quote the highest
number they could get away with!
CPU cycle times are generally a better measure
of what is really going on.
Oddly enough, if we turn cycle times of 200ns
(UVAX II) and 90ns (VS 3100-30) into
"frequencies" we end up with 5MHz and
11.1MHz so for the VS3100-30 Dennis was
spot on in one sense. Given that the 78032
takes in 40MHz and spits out 20MHz, that
figure could be defended and may have been
quoted in a manual too ... where did you