21 Apr
2016
21 Apr
'16
1:36 p.m.
On Apr 21, 2016, at 3:55 PM, Chuck Guzis <cclist at
sydex.com> wrote:
...
Ten was a number that figured into various aspects. The clock was
nomially 10 MHz;
In serial numbers 1-7 only nominally -- the clock was a ring oscillator, tuned by tweaking
wire lengths. Starting with serial number 8, there's a crystal oscillator (in the ECS
controller if ECS is present, otherwise in the CPU).
...
Initially, the maximum central memory size was 131KW; late in the
series, a 262KW option was added, necessitating extensive code changes,
I thought the 70 series (6000 series) was 131 kW max because the top bit is the "ECS
active" bit. 170 series makes it 262k.
...
For a couple of years, I worked on the development a system using up to
4 CPUs with a common 4MW of bulk core (ECS). Since ECS transfers, after
an initial startup overhead ran at full memory speed, a model was
contrived that divided programs up into modules to create "chains", with
inter-module communication, each module resident in either ECS or in a
CPU.
Neat. PLATO made extensive use of ECS, swapping per-terminal state and programs in and
out of ECS for fast interactive service. ECS was also where most I/O buffers went, with
PPUs doing disk and terminal I/O from/to ECS rather than central memory. A dual mainframe
6500 system (4 "unified" processors total) did a decent job supporting 600
concurrent logged-in terminals, out of a total of 1008 connected. That was around 1977
when I worked on that system at the U of Illinois.
paul