Re: [HACKERS] path toward faster partition pruning

On 7 April 2018 at 12:35, Amit Langote <amitlangote09@gmail.com> wrote:
> Thank you Alvaro for rest of the cleanup and committing.

+10!

> So this same failure occurs on (noting the architecture):
>
> ppc64:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=quokka&dt=2018-04-06%2020%3A09%3A52
>
> ia64:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=anole&dt=2018-04-06%2022%3A32%3A24
>
> ppc64 (POWER7):
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=tern&dt=2018-04-06%2022%3A58%3A13
>
> ppc64 (POWER7):
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=hornet&dt=2018-04-06%2023%3A02%3A13
>
> powerpc:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=prairiedog&dt=2018-04-06%2023%3A05%3A08
>
> powerpc:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=locust&dt=2018-04-06%2023%3A13%3A23
>
> powerpc 32-bit userspace on ppc64 host:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=termite&dt=2018-04-06%2023%3A40%3A07
>
> Seems to be due to that the hashing function used in partitioning
> gives different answer for a given set of partition key values than
> others.

They all look like bigendian CPUs.

https://en.wikipedia.org/wiki/Comparison_of_instruction_set_architectures#Endianness

-- 
 David Rowley                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services