Re: What happened to the ex-Guile VM?

[Rob Browning]

Marius Vollmer <address@hidden> writes:

> As evidence in this direction, I get a speedup of 156 (62 compared
> to the non-debugging evaluator) when I inline comparison and
> addition for fixnums (with proper fallback in case of overflow,
> etc).  Lightning has direct support for this kind of things by
> providing a add-and-branch-on-overflow instruction.

Interesting.  So do you have any idea how long it might be before
anything like this, or the VM would be ready for use.  Even if it only
worked in limited case to start with, or even if we had to write
"scheme assembly" in the beginning, that might be sufficient, and then
we could help improve things as needed.

I'm just trying to figure out what we (in gnucash) should do for both
the short term and the long term (which might be different answers)
with respect to optimization.

Thanks

-- 
Rob Browning <address@hidden> PGP=E80E0D04F521A094 532B97F5D64E3930