Re: Abbreviated keys for text cost model fix

On 23.2.2015 00:16, Peter Geoghegan wrote:
> On Sun, Feb 22, 2015 at 1:30 PM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
>> You should try it with the data fully sorted like this, but with one
>> tiny difference: The very last tuple is out of order. How does that
>> look?

I'm running that test now, I'll post the results tomorrow.

> Another thing that may be of particular interest to you as a Czech
> person is how various locales perform. I believe that the collation
> rules of Czech and Hungarian are particularly complex, with several
> passes often required for strcoll() (I think maybe more than 4). You
> should either measure that, or control for it. I was prepared to
> attribute the differences in the two systems to differences in compute
> bandwidth between the CPUs involved (which are perhaps more noticeable
> now, since memory latency is less of a bottleneck). However, the
> inconsistent use of collations could actually matter here.

That's a good point, but all the tests were executed with en_US.UTF-8.

--
Tomas Vondra http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services