On Friday 07 April 2006 22:39, Pam Greene wrote:
> Is there any way to create a more broadly available, persistent custom
> function, short of building it into sqlite? Or does anyone have
> another suggestion to solve the implicit problem?
Layering. Wrap sqlite3_* into your own set of functions.
> > This may take a while, about 20 hours maybe. The partition has approx
> > 10GB, I can't afford more. Let's hope that this is sufficient.
>
> 20 hours seems rather long. Even if you have to worry about uniqueness
> constraints, there are ways to deal with that that should be much faster
> (deal
> On 3/18/06, Daniel Franke <[EMAIL PROTECTED]> wrote:
> > At Thursday I wrote:
> > > Since there's so much interest in this, I'll submit a couple of timings
> > > as soon as possible =)
On Monday 20 March 2006 15:27, Jay Sprenkle wrote:
> Thanks for
At Thursday I wrote:
> Since there's so much interest in this, I'll submit a couple of timings as
> soon as possible =)
Here they are:
--
System: P4 2.4Ghz, 1GB RAm, 4GB swap
--
> That would be an excellent question to add to the FAQ:
> "How do I estimate the resource requirements for a database?"
I spent some time to create 3GB of sample data (just zeros, about half the
size of the actual data set I have to deal with). I'm currently importing it
into the database. As
> How do you use your data? Do you really need to compare all the information
> about any and all individuals?
> [...]
> You may legitimately need one really large table but most applications
> don't.
The most common scenarios:
- get a single marker from subset of individuals
- get a subset of
Note: this is a combined reply to answers sent by Fred and Jay.
Fred wrote:
> > > If so, are you trying to use a blender to stir the ocean?
> > > You might reevaluate if you're using the right tool for the job.
> >
> > That's my question: IS sqlite the right tool here? =)
>
> And I believe he is
> > > You might reevaluate if you're using the right tool for the job.
> > That's my question: IS sqlite the right tool here? =)
> Then I guess the right question is what are your goals? To make
> maintenance easier?
> Why were the thousands of files a problem?
Short answer: I want to improve
-
> > But now, there's another thing.I figured out how large my database
> > will become and I'm scared of its size: up to 20GB and more! A single
> > table, 4 columns, each holding an integer (32 bit) will have
> > approximately 750 million rows. This mounts up to ~11GB. Adding an
> > unique
Hi all ...
a couple of days back, I had a question about the performance of large
tables. You really got me going there, thanks again!
But now, there's another thing.I figured out how large my database
will become and I'm scared of its size: up to 20GB and more! A single
table, 4 columns, each
> > CREATE TABLE genotypes(
> > markerid integer NOT NULL REFERENCES marker(id),
> > individualid integer NOT NULL REFERENCES individuals(id),
> > genA integer,
> > genB integer,
> > UNIQUE(markerid, individualid));
> >
> > CREATE INDEX genotypeidx ON genotypes(markerid,
Is there any chance to speed this up? Production datasets could easily bring a
billion genotypes ...
Any pointer would be appreciated!
With kind regards
Daniel Franke
--
The database schema:
CREATE TABLE marker (
id integer PRIMARY KEY AUTOINCREMENT,
name varchar UNIQUE);
> kexi: btree.c:1043: sqlite3BtreeOpen: Assertion
> `sizeof(ptr)==sizeof(char*)' failed. KCrash: Application 'kexi' crashing...
See also:
http://www.mail-archive.com/sqlite-users%40sqlite.org/msg04718.html
http://bugs.kde.org/show_bug.cgi?id=92563
Daniel
--
Dipl.-Math. (FH)
y I avoid it?
Regards
Daniel
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie und Statistik
Medizinische Universität zu Luebeck
Ratzeburger Allee 160, Haus 4
23538 Luebeck
Telefon: 0451-500-2786
Telefax: 0451-500-2999
[EMAIL PROTECTED]
/"\
\ / ASCII RIBBON C
at ../../src/main.cpp:163
#13 0x0001cfb4 in main (argc=13, argv=0x7398)
at ../../src/main.cpp:114
--
regards
Daniel
P.S. Is this the correct list to post assertation failures, backtraces and so
on?
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinisc
> --
> Assertion failed: res == 0, file ../src/btree_rb.c, line 1341
> --
Forgot the version:sqlite-2.8.15
with gdb, hopefully I will sort it out soon.
Just a guess: I've seen core dumps of the 32bit binaries before
(NDEBUG=1), it just might be the same problem?
Any hints why this happens?
Regards
Daniel Franke
P.S. the submitted sql statements are the same as described in my last
mail a few
END,
CASE WHEN m1 == 0 OR m2 == 0 THEN NULL WHEN m1 != m2 THEN 0 ELSE -1 END
FROM matched_pairs
ORDER BY id1, id2, locus;
--
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie und Statistik
Medizinische Universität zu Luebeck
Ratzeburger Allee
accumulated 100h of cpu-time already - still
calculating :(
--
PID USERNAME THR PRI NICE SIZE RES STATETIMECPU COMMAND
3819 franke 1 00 1907M 1906M cpu/3 126.0H 1.56% mhs-test-4
28918 franke 1 00 2074M 2074M cpu/480 98.2H 1.56% mhs-test-5
--
Any oth
rigger exists for each of the for columns created by the CASE-WHEN
staments of the INSERT above):
Number of rows to expect: up to 10 million
It takes about 2 hours for smaller datasets (about 600.000 rows) to be
created ... any possible speed-up would be greatly appreciated!
Thanks for your replies
x-trigger */
CREATE TRIGGER catch_deletes AFTER DELETE ON work
BEGIN
UPDATE work SET idx = 0;
UPDATE storeroom SET rowcount = 0;
END;
-- 8< --
Cheers
Daniel
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie und Statistik
Medizinische Universitä
idx) first and then recompute the colors.
Cheers
Daniel
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie und Statistik
Medizinische Universität zu Luebeck
Ratzeburger Allee 160, Haus 4
23538 Luebeck
Telefon: 0451-500-2786
Telefax: 0451-500-2999
[EMAIL PROTECTED]
:
Testing the code above on a newly created table (still 220 rows but less
columns) gives the expected results - maybe an issue with sqlite's memory
management?!
Thanks a lot for any help!
With kind regards
Daniel
--
Dipl.-Math. (FH) Daniel Franke
Institut fuer Medizinische Biometrie
23 matches
Mail list logo