pseudo-register addressing

[Henry Cejtin]

If  the  lifetime  is ok, then couldn't you store the pseudo-registers on the
stack?  The point being that this would save one register.

Any way, I would be curious how much space this would save, and also how much
time,  especially  on  programs like self-compile which are large enough that
the I-cache savings might be noticed.  Also, as  Stephen  said,  perhaps  the
register  pointing  to  the  block of pseudo-registers would only be reserved
during chunks of code doing a lot of access to the pseudo-registers  (instead
of all the time).