Re: "Double Precison" Keywords?



robin <robin51@xxxxxxxxxxxxxxxxxx> wrote:
(snip)

The DXR instruction didn't exist, so they couldn't have found
that it was used rarely.
More likely., they probably were pushed for time, and didn't
develop the microcode in time for the 370.

IBM has been pretty good at doing statistical analysis
instead of just guessing.

Extended precision floating point was developed for the 360/85.
While designing the 360/85, they used instruction traces, which
keep track of the instructions executed, address of those instructions,
and addresses referenced by those instructions. That data lead
to the cache memory, first introduced on the 360/85 and now
widely used.

IBM has two journals where they publish their research, and there
are articles on the devlopment of extended precicion floating
point and on the development of cache memory. I suppose they
could lie, but when they say that they did the analysis,
I tend to believe them.

-- glen
.