Re: [PERFORM] Yet another abort-early plan disaster on 9.3

[Evgeniy Shishkin]

Sorry for disrupting the thread,

i am wondering will it be possible to use BRIN indexes to better estimate 
distribution?

I mean create btree index and brin index,
probe brin during planning and estimate if abort early plan with btree will be 
better. 

-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance

[PERFORM] HASH

[Artem Tomyuk]

Hi all.

Is the speed of hash operations stands on the performance of CPU?
Below you can see part from output of explain analyze command

*Intel(R) Xeon(R) CPU   E7520  @ 1.87GHz*
"  ->  Hash  (cost=337389.43..337389.43 rows=3224443 width=34)
(actual time=15046.382..15046.382 rows=3225191 loops=1)"
"Buckets: 524288  Batches: 1  Memory Usage: 207874kB"

*Intel(R) Xeon(R) CPU E5-2670 v2 @ 2.50GHz*

"  ->  Hash  (cost=340758.94..340758.94 rows=3191894 width=34)
(actual time=2692.878..2692.878 rows=3192103 loops=1)"
"Buckets: 524288  Batches: 1  Memory Usage: 205742kB"

*Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz*
"  ->  Hash  (cost=337389.43..337389.43 rows=3224443 width=34)
(actual time=8559.849..8559.849 rows=3225293 loops=1)"
"Buckets: 524288  Batches: 1  Memory Usage: 207881kB"

*Intel(R) Xeon(R) CPU E5-2630 v3 @ 2.40GHz*
"  ->  Hash  (cost=356613.23..356613.23 rows=3224623 width=40)
(actual time=3635.931..3635.931 rows=3224623 loops=1)"
"Buckets: 524288  Batches: 1  Memory Usage: 207838kB"

Thanks.

Re: [PERFORM] HASH

[Jeff Janes]

On Thu, Nov 5, 2015 at 1:11 AM, Artem Tomyuk  wrote:
> Hi all.
>
> Is the speed of hash operations stands on the performance of CPU?

Yes, but the variation is probably not as much as the raw timing in
your example indicates.

> Below you can see part from output of explain analyze command
>
> Intel(R) Xeon(R) CPU   E7520  @ 1.87GHz
>
> "  ->  Hash  (cost=337389.43..337389.43 rows=3224443 width=34)
> (actual time=15046.382..15046.382 rows=3225191 loops=1)"
> "Buckets: 524288  Batches: 1  Memory Usage: 207874kB"

A lot of that time was probably spent reading the data off of disk so
that it could hash it.

You should turn track_io_timing on, run "explain (analyze, buffers)
..." and then show the entire explain output, or at least also show
the entries downstream of the Hash node.

Cheers,

Jeff


-- 
Sent via pgsql-performance mailing list (pgsql-performance@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance