Nothing in these patents implements anything
that wasn't known
already in the 1970s. Why even UNISYS and it's predecessor Burroughs implemented
everything in these patents.
Gee. This morning, I was going to
write up a description of the Burroughs B5500 ca 1965, with its four
(count 'em, 4) switching fabrics in the path between core memory and the disk
surface. I spent too much time in the local ER, though, while the medicos were
deciding whether it was safe to send me to the cardiologist via home or
via inpatient observation. (We were finally able to trigger the failure,
in classic debugging fashion, so that they could see it as sinus tachycardia
rather than a more messy tachycardia or other issue. That's why I'm home, now,
pending cardiologist time. I see beta blockers in my future.)
Anyway, the
B5500 had (1) a switching fabric between the disk read/write heads and the
write-driver/sense-amp electronics: the disk was the head-per-track design,
with lotsa stationary heads per surface. There was (2) another, called
Disk-File Exchange, between the Disk-File Electronics Units and the Disk-File
controllers. Then there was (3) the Peripheral Exchange, between the DFCUs and
the I/O Channels: all the other peripherals connected through this fabric, also.
Finally there was (4) the Memory Exchange, which connected the I/O Channels (and
the processors) to the memory modules.
Fabric (1) was one-by, so it might not
count as a full fabric. The DFX (2) was two-by-ten, with up to two simultaneous
paths from 2 DFCUs to 10 DFEUs. The PX (3) was four-by-32, four independent
paths, between 4 I/O channels and 32 peripheral units; the MX was six-by-eight,
6 requestors (I/Os and Processors) vs eight Memory modules.
Forty-odd
years later, I remember the DFX well: my code was the first that was able to
make full use of that particular switching fabric. I don't recall whether it
was my first or second MCP (Operating-System) assignment. --- --Bill.
NAL: question the answers, especially mine. [ Reply to This | Parent | # ]
|