1 Mar
2015
1 Mar
'15
7:51 p.m.
On 03/01/2015 05:13 PM, Mark Linimon wrote:
You see ... my *first* machine was a Bendix (although
late
enough to be labeled Control Data) G-15.
Yup, I tried to help bring up a surplus
G-15D, but finally
discovered the drum was
trashed. If you didn't reorganize the instruction stream
around the drum for
best access time, the machine ran at roughly 30 IPS, one
instruction per drum revolution.
(I'm simplifying here, as the drum on the G-15 had
recirculating tracks that were
rewritten continuously between read head and write head.)
if you reorganized
the instructions to just come under the head when the last
instruction was done,
you could get 10K + IPS. A number of other drum-memory
machines used
similar schemes. And, that was the GREAT advantage of core
memory,
RANDOM access!
Jon