Hi,
I have some whopper tables for machine learning. One table has a handful
of 200-500 double precision arrays (representing feature vectors). It's
a 9.5 on a SSD (over USB3). Each table has 5-15M rows in them.
For each array I've added, and populated, any dealings with the table
has
Hi,
Are there highly efficient C extensions out there for math operations on
arrays? Dot product and whatnot.
Example usecase: sort an item by euclid distance.
Kind regards,
Marcus
--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
On 24/12/15 07:13, Pavel Stehule wrote:
Hi
2015-12-24 8:05 GMT+01:00 Marcus Engene <meng...@engene.se
<mailto:meng...@engene.se>>:
Hi,
Are there highly efficient C extensions out there for math
operations on arrays? Dot product and whatnot.
what you mean "
Hi,
I'm working with a retail site with tens of millions of products in
several languages.
For the detail pages, we try to cache in memcached. We also have quite a
bit of keyword translation lookups (for international queries to solr).
We're thinking of adding a nosql layer that takes the
Hi,
I have a local db behind a firewall etc. Basically, I'd like to do what
I'd locally would...
create table abc
as
select
*
from
local_expensive_view;
abc - on RDS
local_expensive_view - on local machine
How would you go about doing this?
Thanks,
Marcus
--
Sent via
On 06/05/14 16:58, bricklen wrote:
A very quick search shows that rds supports dblink, so perhaps that
would work.
http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/CHAP_PostgreSQL.html
Then I'd need to open our servers to external visits. It would be lovely
if dblink_exec could push a
On 06/05/14 17:15, bricklen wrote:
On Tue, May 6, 2014 at 8:07 AM, Marcus Engene meng...@engene.se
mailto:meng...@engene.se wrote:
On 06/05/14 16:58, bricklen wrote:
A very quick search shows that rds supports dblink, so perhaps
that would work.
http
Hi,
I have a production database with some heavy views.
I'd like to make a hourly cron (or smth) that bulk inserts new entries
into a table on RDS.
It's easy to get data down to the prod side, but is it possible to do a
bulk insert the other way around? All examples I've seen adds the data
Hi list,
In Oracle I can...
create table a
(
b varchar2(10 chars)
);
...and then, regardless of character encoding and how much space an
ascii character vs a รถ takes, 10 characters will fit there.
If I do say a web-thing in php I have to do horrors like...
if (10 mb_strlen ($b, '8bit'))
{
On 9/21/10 1:29 , Terry Lee Tucker wrote:
On Tuesday, September 21, 2010 07:23:45 Massa, Harald Armin wrote:
I recommend to use TEXT as type for that kind of columns.
99 out of 100 theories about this value will never be longer then xx
characters fail in the long run.
And text, limited
Hi List,
With automated queries where I have COLUMN IN (), I get a different plan
from COLUMN = (). That would make sense if there are several arguments,
but in the case with just one argument the penalty can be seveare. One
query went from 5s execution time to a few houndreds of mS when I
On 7/11/10 12:42 , Alban Hertroys wrote:
On 11 Jul 2010, at 11:38, Marcus Engene wrote:
Hi List,
With automated queries where I have COLUMN IN (), I get a different plan from
COLUMN = (). That would make sense if there are several arguments, but in the
case with just one argument
Hi,
-- a
select
...
from
base_table bt
left outer join sub_table_1 st1 on (st1.base_table = bt.id)
left outer join sub_table_2 st1 on (st2.base_table = bt.id)
-- b
select
...
from
base_table bt
left outer join sub_table_1 st1 on (bt.objecttype = 1 AND
st1.base_table =
Hi list,
I have a table, not that many rows (400k) but with about 55 columns.
There are some maintenance selects I want to do (every now and then)
that I don't want to add indexes for.
select
...
from
quite_big_table qbt
where
qbt.owner = 123 AND
exists (
select null
Tom Lane wrote:
Marcus Engene meng...@engene.se writes:
... In a case with some 5000 rows belonging to owner 123, this select really
takes a long time. Way longer than without the subselect and order by
filelength. I agree that with the subselect it would take some extra
juice, but in my
Richard Huxton wrote:
I imagine it's cheaper disk dump wise to do the array thing in this and
some other similar usages, and therefore it would be nice to have a
non-ugly usage pattern.
Don't imagine, test. And then factor in the cost of fiddling around with
arrays when you need to
Hi List,
I have the might_like table that contains products a user might like if
he likes the present one (item).
CREATE TABLE might_like
(
item INTEGER NOT NULL
,created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP NOT NULL
,child INTEGER
Richard Huxton wrote:
Marcus Engene wrote:
Hi List,
I have the might_like table that contains products a user might like if
he likes the present one (item).
CREATE TABLE might_like
(
item INTEGER NOT NULL
,created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
Dianne Yumul wrote:
Hello,
I have some html forms that I save the settings into the database,
things like which item was selected in the menu and if a checkbox was
checked. The table looks like this:
user_id | report_id | info
Hi!
I'm using tsearch2 and sometimes there are blocks of text that shouldn't
be indexed.
There is a trigger function that gathers data from the usual suspects
and updates the index. in this trigger, I'd like to exclude thing in a
[noindex] tag:
select
regexp_replace
Tom Lane wrote:
Marcus Engene [EMAIL PROTECTED] writes:
I would like to have a function like the above that returns innan klas
for this data. I would have expected it to as I use the non greedy version.
regression=# select
regexp_replace
('innan[noindex]apa[/noindex]klas[noindex
Phoenix Kiula wrote:
If you don't want to store IPs for registered users, I'd use:
user_id INTEGER,
ip cidr,
CONSTRAINT must_have_userstamp
CHECK ( user_id IS NOT NULL OR ip IS NOT NULL)
... and yes, I'd use a functional index to look it up, or even a
trigger-maintained cache of the
Pierre Thibaudeau wrote:
I have a PERSONS table.
I also have a PROGENY table, which is a many-to-many association table
with two foreign keys to the PERSONS table to itself.
(In this day and age, not only can an individual have any number of
children, but also a person can have any number of
Hi,
If I have a table like...
CREATE TABLE apa
(
objectid SERIAL PRIMARY KEY NOT NULL
,created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP NOT NULL
,somedata text
)
WITHOUT OIDS;
...where if rowX har higher objectid than rowY, it is implied that
Oleg Bartunov wrote:
On Sat, 30 Jun 2007, Marcus Engene wrote:
Hi!
bond= SELECT to_tsvector('default','animation animal');
to_tsvector
-
'anim':1,2
(1 row)
bond=
Sorry for a silly question, I wonder, how do I override this? I would
want different stemmed words
Hi!
bond= SELECT to_tsvector('default','animation animal');
to_tsvector
-
'anim':1,2
(1 row)
bond=
Sorry for a silly question, I wonder, how do I override this? I would
want different stemmed words for these.
Best regards,
Marcus
---(end of
Tom Lane skrev:
Marcus Engene [EMAIL PROTECTED] writes:
- Index Scan using apa_item_common_x1 on apa_item_common aic
(cost=0.00..4956.68 rows=1174 width=8) (actual time=19.854..9557.606
rows=1226 loops=1)
If the table only has 12000 rows then it should never have used
Michael Fuhr skrev:
On Sat, Apr 07, 2007 at 09:28:58AM +0200, Marcus Engene wrote:
INFO: apa_item_common: removed 9028 row versions in 3651 pages
DETAIL: CPU 0.24s/0.36u sec elapsed 30.69 sec.
INFO: apa_item_common: found 9028 removable, 12863 nonremovable row
versions in 14489 pages
Hi,
There are some performance issues I am still confused over.
I've got a Linux box, raid1, 1GB memory
CPU-wise the server is mostly idle
PG 8.0.10,
shared_buffers = 1
work_mem = 16348
maintenance_work_mem = 65536
Parameters are tweaked without much difference.
The following query is
Hi,
I'm on 8.0.10 and there is a query I cannot quite get adequately fast.
Should it take 2.5s to sort these 442 rows? Are my settings bad? Is
my query stupid?
Would appreciate any tips.
Best regards,
Marcus
apa= explain analyze
apa- select
apa- ai.objectid as ai_objectid
apa- from
Hi again,
I was thinking, in my slow query it seems the sorting is the villain.
Doing a simple qsort test I notice that:
[EMAIL PROTECTED] /cygdrive/c/pond/dev/tt
$ time ./a.exe 430
real0m0.051s
user0m0.030s
sys 0m0.000s
[EMAIL PROTECTED] /cygdrive/c/pond/dev/tt
$ time ./a.exe
Tom Lane skrev:
Marcus Engene [EMAIL PROTECTED] writes:
Should it take 2.5s to sort these 442 rows?
Limit (cost=54.40..54.43 rows=12 width=8) (actual
time=2650.254..2651.093 rows=442 loops=1)
- Sort (cost=54.40..54.43 rows=12 width=8) (actual
time=2650.251..2650.515 rows
Hi list,
I'd like to generate the latest year dynamically with generate_series.
This select works day wise:
select date_trunc ('month', now())::date + s.a
from generate_series(0, 11) as s(a)
I tried this but it didn't work:
select date_trunc ('month', now())::date + interval s.a || ' months'
Tom Lane skrev:
Marcus Engene [EMAIL PROTECTED] writes:
I tried this but it didn't work:
select date_trunc ('month', now())::date + interval s.a || ' months'
from generate_series(0, 11) as s(a)
People keep trying that :-(. The typename 'foo' syntax is for a
*literal constant* only
semi-ambivalent skrev:
All,
I have a simple web page that inserts data into a table in 8.1.5 using
PHP4. It's pretty amateurish but gets the job done. A few days ago I
upgraded to 8.2 (this is on a FreeBSD system, and I used the port) but
when I tried to use the web page this morning I got
[EMAIL PROTECTED] skrev:
I really like TOAD for building Oracle queries. Is there a TOAD-like,
FOSS query builder for PostgreSQL?
Hi,
I use Toad for Oracle too.
pgAdminIII is included Postgres, its good but I prefer pg lightning
admin which I've been using daily for almost a year now. It's
John DeSoi skrev:
There are highly productive IDEs for the Mac with all the goodies you
mention. But few are cross-platform.
Your statement about Windows desktop market share is correct, but it is
not the relevant point. Many people are interested in cross-platform
tools because they want
I see we have a C J Date fan on the list! ;-)
There is one other case where I personally find nullable
columns a good thing: process_me ish flags. When a row
is not supposed to be processed that field is null and
when a field is null it wont be in the index [at least
on Oracle].
Best regards,
Ilja Golshtein skrev:
Hello!
How could I find out if a temporary table
(or index on a temporary table) was created
by current session?
The problem is something like
SELECT COUNT(*) FROM PG_INDEXES WHERE INDEXNAME='tmpind1'
does not work since temporary indexes from other sessions
are
Rick Schumeyer skrev:
I will have to try the WIN1252 encoding.
On the client side, my application is a web browser. On the server
side, it is php scripts on a linux box. The data comes from copying
data from a browser window (pointing to another web site) and pasting it
into an html
Hi,
I have a table with items.
One field is site_user, which fk:s the user row that owns the item.
There are a few fields i throw to tsearch2() via an index right now,
but one user wanted his name to be in the search as well. So, what I
would want to do is:
CREATE FUNCTION euits(int)
RETURNS
Teodor Sigaev skrev:
CREATE FUNCTION euits(int)
RETURNS text AS 'select username || \' \' || firstname || \' \' ||
lastname from site_user where objectid = $1;' LANGUAGE SQL;
CREATE TRIGGER site_item_fts
BEFORE UPDATE OR INSERT ON site_item
FOR EACH ROW EXECUTE PROCEDURE
tsearch2(idxfti,
Greg Stark wrote:
Alexander Scholz [EMAIL PROTECTED] writes:
Hi, thank you for your answer.
Regarding the performance flow when trying to find out how many records are
currently being stored in the table, I don't see how an index should help...
Nevertheless we've created an unique index on ID
That a simple case, what about languages as norwegian or german? They
has compound words and ispell dictionary can split them to lexemes.
But, usialy there is more than one variant of separation:
forbruksvaremerkelov
forbruk vare merke lov
forbruk vare merkelov
forbruk varemerke lov
Hi,
I use the tsearch full text search with pg 8.0.3. It works great, but I
wonder if it's possible to search for compound words?
Ie if I search for New York i want to get a match on
New York has traffic problems.
but not on
New axe murderer incident in brittish York.
Is this possible?
I
Oleg Bartunov wrote:
On Mon, 5 Dec 2005, Marcus Engene wrote:
Hi,
I use the tsearch full text search with pg 8.0.3. It works great, but
I wonder if it's possible to search for compound words?
Ie if I search for New York i want to get a match on
New York has traffic problems
46 matches
Mail list logo
