On 2018-03-02 09:42:22 +0000, Yuqi Gu wrote:
> > Could you show whether it actually improves performance? Usually bulk
> > loading data with parallel COPYs is a good way to hit this code path.
> The mini benchmark code:
I'd be more interested in a benchmark using postgres itself...
> > What's the availability of these headers and functions on non-linux
> > platforms?
> This Arm64 optimization code only supports linux os so far.
That's not ok for postgres, unfortunately...
> What does it mean "stock autoconf" ?
> Should the configure script be made by specific version autoconf ?
Yes, it's the version from the autoconf project, rather than with
modifications by distributions. Don't worry about it, the committer can
take care of that.