On Thu, 2014-09-25 at 11:13 +0100, David Woodhouse wrote:
I suggested a couple of specific optimisations which the query planner
might be able to make, which should hopefully have benefits wider than
just my own use case. Are those not viable?
I'm preparing to commit a workaround to
On 10/09/2014 04:38 PM, David Woodhouse wrote:
On Thu, 2014-09-25 at 11:13 +0100, David Woodhouse wrote:
I suggested a couple of specific optimisations which the query planner
might be able to make, which should hopefully have benefits wider than
just my own use case. Are those not viable?
I'm
Simon Slavin wrote...
On 24 Sep 2014, at 2:13pm, jose isaias cabrera jic...@cinops.xerox.com
wrote:
This would be a nice set of options. On my case, I would set all
connections to our project to be max_performance, as it is what we
need. Just thinking out loud.
How much max is max ?
In the vein of configurability, and in a day dream I just had, it would be
nice (But probably not possible as there could be compiler directives you
can't use at the same time) that we could have a single output
DLL/SO/whatever dumped from the compiler that had everything available,
then, via
On Wed, 2014-09-24 at 19:36 -0600, Keith Medcalf wrote:
Interesting. From that code you might want to try something like this:
SELECT uid, vcard, bdata
FROM folder_id
WHERE uid in ( select uid FROM email_list where value like 'p%'
union
select uid from
On Tue, 2014-09-23 at 17:48 +0100, David Woodhouse wrote:
That looks really promising; thanks for all this work.
Tristan, you have a comprehensive set of benchmarks for Evolution's
addressbook; is it possible for someone else to run those or would it
take more of your time to babysit than it
On Fri, 2014-09-19 at 21:14 -0400, Richard Hipp wrote:
The 50% faster number above is not about better query plans.
Speaking of better query plans, though... here's a query which takes
about 1700ms on my data set, followed by a couple of optimisations which
seem like they might be generically
Would it not be more efficient to skip the join altogether since all you want
is the list of uid's, and assuming that you have maintained the referential
integrity of your database mail_list(list_uid) references main(uid)?
SELECT list_uid
FROM mail_list
WHERE email LIKE 'al%'
UNION
SELECT
Roger Binns wrote...
On 22/09/14 10:48, Richard Hipp wrote:
But if you have any new ideas on how we can further reduce the I/O, we'd
love to hear from you.
The single biggest problem for me is defaults. SQLite supports memory
mapped i/o which has many advantages. The stat4 analyze does a
On 24 Sep 2014, at 2:13pm, jose isaias cabrera jic...@cinops.xerox.com wrote:
This would be a nice set of options. On my case, I would set all connections
to our project to be max_performance, as it is what we need. Just thinking
out loud.
How much max is max ? Are you willing to give up
On Wed, 2014-09-24 at 06:13 -0600, Keith Medcalf wrote:
Would it not be more efficient to skip the join altogether since all
you want is the list of uid's, and assuming that you have maintained
the referential integrity of your database mail_list(list_uid)
references main(uid)?
SELECT
On 24/09/14 06:19, Simon Slavin wrote:
How much max is max ?
Not giving up ACID. But for example stat4 is better than the default stat1.
Memory mapping (especially on 64 bit) is great. So is WAL. All are off by
default.
If you want to give up ACID then you should really be on your own to
]
Sent: Wednesday, 24 September, 2014 07:25
To: Keith Medcalf
Cc: General Discussion of SQLite Database
Subject: Re: [sqlite] 50% faster than 3.7.17
On Wed, 2014-09-24 at 06:13 -0600, Keith Medcalf wrote:
Would it not be more efficient to skip the join altogether since all
you want is the list
Are any of these improvements specifically in the area of the online backup
API, or are they more in the general running of SQLite?
On 20 September 2014 11:14, Richard Hipp d...@sqlite.org wrote:
The latest SQLite 3.8.7 alpha version (available on the download page
On Tue, Sep 23, 2014 at 2:33 AM, Donald Shepherd donald.sheph...@gmail.com
wrote:
Are any of these improvements specifically in the area of the online backup
API, or are they more in the general running of SQLite?
The speedtest1 benchmark does not use the backup API. However, many of
the
On Fri, 2014-09-19 at 21:14 -0400, Richard Hipp wrote:
The latest SQLite 3.8.7 alpha version (available on the download page
http://www.sqlite.org/download.html) is 50% faster than the 3.7.17 release
from 16 months ago. That is to say, it does 50% more work using the same
number of CPU
On 22/09/14 10:48, Richard Hipp wrote:
But if you have any new ideas on how we can further reduce the I/O, we'd love
to hear from you.
The single biggest problem for me is defaults. SQLite supports memory
mapped i/o which has many advantages. The stat4 analyze does a really good
job. WAL
On Fri, Sep 19, 2014 at 09:14:17PM -0400, Richard Hipp wrote:
The latest SQLite 3.8.7 alpha version (available on the download page
http://www.sqlite.org/download.html) is 50% faster than the 3.7.17 release
from 16 months ago. That is to say, it does 50% more work using the same
number of
Hi,
This 3.8.7alpha release seems to bring about 5% win from 3.8.6 , on my
particular SQL test case.
Question : PRAGMA threads=2 didn't bring any speed-up on my 2 true
cores machine.
Did I miss a compilation option ?
___
sqlite-users mailing list
On Mon, Sep 22, 2014 at 1:43 PM, big stone stonebi...@gmail.com wrote:
Hi,
This 3.8.7alpha release seems to bring about 5% win from 3.8.6 , on my
particular SQL test case.
Question : PRAGMA threads=2 didn't bring any speed-up on my 2 true
cores machine.
Did I miss a compilation option ?
On Mon, Sep 22, 2014 at 8:29 AM, Valentin Davydov sqlite-u...@soi.spb.ru
wrote:
On Fri, Sep 19, 2014 at 09:14:17PM -0400, Richard Hipp wrote:
The latest SQLite 3.8.7 alpha version (available on the download page
http://www.sqlite.org/download.html) is 50% faster than the 3.7.17
release
ok,
Nearly all the time is spent in a big 'CTE' select.
So maybe this sort of ugly CTE query is not threadable.
with
f0_k as
(SELECT f.rowid as nof2, f.*, p.Dn, p.g1, p.g2, p.w, p.other FROM F0 AS
f, Pt AS p WHERE f.Io =p.Io)
,F0_mcalc as
(SELECT f.*, p.*, (case when Priority=999 then Cg
In trying to see if the new version breaks any of my queries, I ran several
of them repeatedly, and they all appear to have produced the expected
output.
The only thing I noticed which maybe of interest in relation to speed
performance was (with .timer on) that although the first two run time
On Sat, Sep 20, 2014 at 1:34 PM, to...@acm.org wrote:
In trying to see if the new version breaks any of my queries, I ran
several of them repeatedly, and they all appear to have produced the
expected output.
The only thing I noticed which maybe of interest in relation to speed
performance
The latest SQLite 3.8.7 alpha version (available on the download page
http://www.sqlite.org/download.html) is 50% faster than the 3.7.17 release
from 16 months ago. That is to say, it does 50% more work using the same
number of CPU cycles.
This performance gain is over and above the query
Richard Hipp wrote...
The latest SQLite 3.8.7 alpha version (available on the download page
http://www.sqlite.org/download.html) is 50% faster than the 3.7.17 release
from 16 months ago. That is to say, it does 50% more work using the same
number of CPU cycles.
This performance gain is over
I, as well, wish to thank you for this tool Dr. Hipp. I've never published
a public application using this engine, but, at my place of employment,
where my primary responsibility is to just monitor servers world wide, I've
coded a few tidbit web and desktop applications that have made my job SO
27 matches
Mail list logo