21 Feb
2012
21 Feb
'12
10:56 p.m.
Dave McGuire wrote:
I read somewhere that it's actually an 8-bit
data path, which always
seemed odd to me.
It's the classic time/space tradeoff. The IBM 360 is a
32-bit
architecture, but the 360/30 has 8-bit data paths, and the 360/40 has
16-bit data paths. A 16-bit data path cost a lot more money than an
8-bit data path, even when it was in 7 micron NMOS. It's only fairly
recently that this has ceased to be true, because process geometries are
now so small that less complex chips are pad-limited.
Very interesting stuff. Is the chipset fully
documented anywhere?
Bitsavers has the manual for the KUV11, which is the writeable
control
store option for the LSI-11. There must have been a tech manual from
WD, but I've never seen it, and it would probably be a subset of the
information in the KUV11 manual.
There is a WD manual on the WD16, but it is at the macroinstruction
level (PDP-11-like), and doesn't cover the microarchitecture.