Re: Progress on fast path sorting, btree index creation time

On Wed, Feb 01, 2012 at 04:12:58PM -0600, Jim Nasby wrote:
> On Jan 26, 2012, at 9:32 PM, Robert Haas wrote:
> > But if we want to put it on a diet, the first thing I'd probably be
> > inclined to lose is the float4 specialization. Some members of the
> > audience will recall that I take dim view of floating point arithmetic
> > generally, but I'll concede that there are valid reasons for using
> > float8. I have a harder time coming up with a good reason to use
> > float4 - ever, for anything you care about. So I would be inclined to
> > think that if we want to trim this back a bit, maybe that's the one to
> > let go. If we want to be even more aggressive, the next thing I'd
> > probably lose is the optimization of multiple sortkey cases, on the
> > theory that single sort keys are probably by far the most common
> > practical case.
>
> I do find float4 to be useful, though it's possible that my understanding is flawed…
>
> We end up using float to represent ratios in our database; things that really, honest to God do NOT need to be exact.
>
> In most cases, 7 digits of precision (which AFAIK is what you're guaranteed with float4) is plenty, so we use float4 rather than bloat the database (though, since we're on 64bit hardware I guess that distinction is somewhat moot…).
>
> Is there something I'm missing that would make float4 useless as compared to float8?
> --
> Jim C. Nasby, Database Architect jim(at)nasby(dot)net
> 512.569.9461 (cell) http://jim.nasby.net
>
If the values stored are float4, it would be nice to have that fast-path
sort available too. The cases where I have used float4 values in the past,
I absolutely did not need any of the float8 baggage and in my case, using
the actual float4 comparison operator resulted in a significant time savings
over the normal float8. This could be processor specific, but it would be
worth testing before throwing it out.

Regards,
Ken