Re: Possible regression in PG18 beta1 - Mailing list pgsql-hackers

From Sadeq Dousti
Subject Re: Possible regression in PG18 beta1
Date
Msg-id CADE6LvgV5nro2SCe4OP2vBZirsP2cpT1oNG3Rg+os38=7ZS0ow@mail.gmail.com
Whole thread Raw
In response to Possible regression in PG18 beta1  (Sadeq Dousti <msdousti@gmail.com>)
List pgsql-hackers
Yes, this is intentional as of 95dbd827f2edc
Thanks!

 
>> > * Changing the table from TEMP to UNLOGGED makes both versions use
>> > "Parallel Seq Scan", with PG18 being ~25% faster. (190ms vs. 150ms).

I'll be curious about tests with a normal table as well with a
sufficiently large shared_buffers.

Here are results for a normal table with default shared_buffers (128 MB) and large shared_buffers (4GB):

==== PG17.5, shared_buffers=128MB ======
                      QUERY PLAN
------------------------------------------------------
 Gather (actual rows=1 loops=1)
   Workers Planned: 2
   Workers Launched: 2
   Buffers: shared hit=3584 read=50496
   ->  Parallel Seq Scan on t (actual rows=0 loops=3)
         Filter: (k = 1)
         Rows Removed by Filter: 3333333
         Buffers: shared hit=3584 read=50496
 Settings: random_page_cost = '1.1'
 Planning Time: 0.119 ms
 Execution Time: 141.374 ms
(11 rows)

==== PG17.5, shared_buffers=4GB ======
select * from t where k = 1;
                      QUERY PLAN
------------------------------------------------------
 Gather (actual rows=1 loops=1)
   Workers Planned: 2
   Workers Launched: 2
   Buffers: shared hit=3872 read=50208
   ->  Parallel Seq Scan on t (actual rows=0 loops=3)
         Filter: (k = 1)
         Rows Removed by Filter: 3333333
         Buffers: shared hit=3872 read=50208
 Settings: random_page_cost = '1.1'
 Planning Time: 0.118 ms
 Execution Time: 141.846 ms
(11 rows)


==== PG18 beta1, shared_buffers=128MB ======
                       QUERY PLAN
---------------------------------------------------------
 Gather (actual rows=1.00 loops=1)
   Workers Planned: 2
   Workers Launched: 2
   Buffers: shared hit=15986 read=38094
   ->  Parallel Seq Scan on t (actual rows=0.33 loops=3)
         Filter: (k = 1)
         Rows Removed by Filter: 3333333
         Buffers: shared hit=15986 read=38094
 Settings: random_page_cost = '1.1'
 Planning Time: 0.123 ms
 Execution Time: 110.650 ms
(11 rows)


==== PG18 beta1, shared_buffers=4GB ======
                        QUERY PLAN
---------------------------------------------------------
 Gather (actual rows=1.00 loops=1)
   Workers Planned: 2
   Workers Launched: 2
   Buffers: shared hit=15991 read=38089
   ->  Parallel Seq Scan on t (actual rows=0.33 loops=3)
         Filter: (k = 1)
         Rows Removed by Filter: 3333333
         Buffers: shared hit=15991 read=38089
 Settings: random_page_cost = '1.1'
 Planning Time: 0.095 ms
 Execution Time: 109.073 ms
(11 rows)


Best Regards,
Sadeq Dousti

pgsql-hackers by date:

Previous
From: Sadeq Dousti
Date:
Subject: Re: Possible regression in PG18 beta1
Next
From: "Regina Obe"
Date:
Subject: RE: pg_upgrade ability to create extension from scripts