On Fri, 2 Dec 2022 at 20:35, Peter Eisentraut
<peter.eisentraut@enterprisedb.com> wrote:
> If we are happy with this patch, then it's okay with me if you push this
> first. I'll probably need to do another pass over my patch anyway, so a
> bit more work isn't a problem.
Thanks. I'll start looking at the patch again now. If I don't find any
problems then I'll push it.
I just did some performance tests by COPYing in 40 million INTs
ranging from -/+ 20 million.
create table ab(a int, b int);
insert into ab select x,x from generate_series(-20000000,20000000)x;
copy ab to '/tmp/ab.csv';
-- master
truncate ab; copy ab from '/tmp/ab.csv';
Time: 10219.386 ms (00:10.219)
Time: 10252.572 ms (00:10.253)
Time: 10202.940 ms (00:10.203)
-- patched
truncate ab; copy ab from '/tmp/ab.csv';
Time: 9522.020 ms (00:09.522)
Time: 9441.294 ms (00:09.441)
Time: 9432.834 ms (00:09.433)
About ~8% faster
David