On Sun, Sep 6, 2015 at 9:35 PM, Thomas Munro
<thomas.munro@enterprisedb.com> wrote:
> Running various contrived aggregate queries on a large low cardinality
> dataset in a small range (therefore frequently the same weight & size), I
> managed to measure a small improvement of up to a few percent with the
> attached patch. I also wonder whether numeric_cmp could be profitably
> implemented with memcmp on big endian systems and some unrolled loops on
> little endian systems when size & weight match.
I think that setting up numeric SortSupport to have scratch memory
used across authoritative numeric comparator calls would also help.
We should prefer to do this kind of thing in a datatype independent
way, of course. I'm not opposed to doing what you outline too, but I
don't think it will be especially helpful for the cases here. I think
that what you're talking about would live in the SortSupport
authoritative comparator, and would benefit non-leading-attribute
comparisons most.
> Of course there are a ton of other overheads involved with numeric. I
> wonder how crazy or difficult it would be to make it so that we could
> optionally put a pass-by-value NUMERIC in a Datum, setting a low order tag
> bit to say 'I'm an immediate value, not a pointer', and then packing 3
> digits (= 12 significant figures) + sign + weight into the other 63 bits.
That seems possible, but very invasive. I'd want to get a good sense
of the pay-off before undertaking such a project.
--
Peter Geoghegan