[Info-vax] OT: book from George Dyson.

[Paul Sture]

On Wed, 11 Apr 2012 11:14:58 -0700, John Wallace wrote:

> On Apr 11, 9:16 am, Paul Sture <p... at sture.ch> wrote:
>> On Tue, 10 Apr 2012 10:21:38 -0700, Rich Jordan wrote:
>> > On Apr 10, 12:10 pm, brendan welch <w1... at uml.edu> wrote:
>>
>> >> It is not exactly the same thing, but I used the screen of an early
>> >> 6 Mhz IBM PC, as the memory to generate a pretty large list of prime
>> >> numbers.  For programming convenience, I just let it run overnight,
>> >> using some easy-to-learn DOS-type instructions.
>>
>> >>http://www.setileague.org/articles/primes/index.html if you are
>> >> masochistic.
>>
>> > Good thing you had the turbocharged 6MHz model, not the base original
>> > 4.77 MHz or it would have taken weeks...
>>
>> LOL! Anyone remember the Turbo Button on early 1990s PCs?
>>
>> You switched Turbo off when a DOS program couldn't cope with the speed
>> of the CPU. :-)
>>
> I don't have to remember the Turbo button, there's a system with one a
> few feet away from me at work. Not everything can be virtualised under a
> HYPErvisor, and when support for products needs to be provided for more
> than a couple of decades, sometimes old hardware (and software) hangs
> around for quite a while. [But please don't ask what happens if it
> breaks]

I recall someone hunting for a PC which would run Windows 95 just a few 
years back, I assume for the same reason.

Bringing us back to the VAX world, someone contacted me offline with the 
following:

--- quote ---
>From the VAX-11/780 console:

>>> SET CLOCK FAST  

SLOW and NORMAL were also supported.  

FAST sped the master clock by about 10%, IIRC. 

Setting to SLOW dropped the clock by more than 10%, IIRC.

More than a few 780 boxes were running FAST, where that was stable.

Mentioned (in passing, in DEC WARS) in
<http://archive.hmvh.net/txtfiles/h-mail/H-MAIL16.TXT>

Listed in VAX-11/780 hardware service manual, too.

--- end quote ---

I missed that one when I had a pair of 11/780s to play with.  I am
reminded of an 11/780 customer benchmark which failed to a achieve
the results it had when still at the manufacturer.  The bottleneck
there turned out to be the PDP which was trying to simulate a
bunch of terminals.

-- 
Paul Sture