Re: as/400 / linux / lpar

["Leif Svalgaard" <leif@xxxxxxxx>]

From: Larry Bolhuis <lbolhuis@arbsol.com>
> What WE do not know is what processor instruction, register, microcode
> enhancement etc.... is needed to support fractional Linux LPARs. What we DO
> know is that each turn of the crank produces processors with more
> capability, more capacity, more 'RPMs' (MHz), or some combination, than the
> last generation. If that were not true then V5R2 would run on a model 400
> but it would just be somewhat slower than say an i890.
>
> It may be technically possible to code this capability back to iStar or
> Pulsar processors, I couldn't say. But if it's any significant effor
> whatsoever, do we really want IBM to spend development dollars to support
> fractional Linux LPARs on machines marginally capable of having enough
power
> to support it?  I think not.
>

Now we are talking processor speed and capacity rather than
different (new) instructions. I don't think it costs any development
dollars to support older processors that are instruction-wise
identical to newer ones, so we are back to marketing and business
rather than technical reasons. Not that these other reasons can't be
valid. When talking about spending development dollars, it seems
that IBM is happy to spend such dollars PREVENTING the new stuff
to run on older systems. They did the same when the AS/400 first
came out. S/38 programs could run on AS/400, but AS/400 programs
could not run on S/38 (a major reason was the deliberate change
of some ODP parameters for opening files. The parameter was 72
on S/38 and 72 and 75 on AS/400 - the AS/400 would understand both
but the S/38 not. All code generated on the AS/400 used 75, of course,
but I digress...)