[geos-devel] Performance 3.10 -> 3.11

Paul Ramsey pramsey at cleverelephant.ca
Mon Jul 4 11:45:00 PDT 2022


Forgot to note the source of the test framework, for those interesting
in running themselves:

https://github.com/pramsey/geos-performance

On Mon, Jul 4, 2022 at 11:43 AM Paul Ramsey <pramsey at cleverelephant.ca> wrote:
>
> I thought I'd run the "standard" performance comparison set over the
> latest release. We haven't really done any broad-based performance
> work this cycle, so anything we picked up is incidental, and in
> general the changes were quite small.
>
> The only performance regression was in basic STRtree queries. (Which
> should probably be renamed for clarity, since it's a query on the
> default CAPI tree, which is now the TemplateRtree). The measurements
> were all pretty stable, which was gratifying, so for the differences
> of 4% or so I feel pretty confident of them. The < 1% differences
> should probably be read as "samesame".
>
> Not sure what/where a next best place to attach performance
> systematically is. For overlay, I'd say the spare copy of the data we
> make in the NodedSegmentString could be a target. Would require a
> difference lifecycle contract for the Noder, however, but could yield
> some widely effective gains, since it's such heavily used code.
>
> P.


More information about the geos-devel mailing list