diff --git a/doc/TODO.detail/qsort b/doc/TODO.detail/qsort index 321babd27f..4284c73941 100644 --- a/doc/TODO.detail/qsort +++ b/doc/TODO.detail/qsort @@ -1,111 +1,584 @@ -Index: doc/src/FAQ/FAQ_DEV.html -=================================================================== -RCS file: /projects/cvsroot/pgsql/doc/src/FAQ/FAQ_DEV.html,v -retrieving revision 1.107 -diff -c -r1.107 FAQ_DEV.html -*** doc/src/FAQ/FAQ_DEV.html 24 Dec 2005 19:29:38 -0000 1.107 ---- doc/src/FAQ/FAQ_DEV.html 16 Feb 2006 20:08:51 -0000 -*************** -*** 156,180 **** - -

1.5) I've developed a patch, what next?

- -!

Generate the patch in contextual diff format. If you are -! unfamiliar with this, you might find the script -! src/tools/makediff/difforig useful. Unified diffs are -! only preferrable if the file changes are single-line changes and -! do not rely on the surrounding lines.

-! -!

Ensure that your patch is generated against the most recent -! version of the code. If it is a patch adding new functionality, the -! most recent version is CVS HEAD; if it is a bug fix, this will be -! the most recently version of the branch which suffers from the bug -! (for more on branches in PostgreSQL, see 1.15).

-! -!

Finally, submit the patch to pgsql-patches@postgresql.org. It - will be reviewed by other contributors to the project and will be -! either accepted or sent back for further work. Also, please try to -! include documentation changes as part of the patch. If you can't do -! that, let us know and we will manually update the documentation when -! the patch is applied.

- -

1.6) Where can I learn more about the - code?

---- 156,231 ---- - -

1.5) I've developed a patch, what next?

- -!

You will need to submit the patch to pgsql-patches@postgresql.org. It - will be reviewed by other contributors to the project and will be -! either accepted or sent back for further work. To help ensure your patch -! is reviewed and committed in a timely fasion, please try to make sure your -! submission conforms to the following guidelines: -!

    -!
  1. Has the patch been discussed previously? If it has, give a direct link -! to the message and/or bug# from the mail archives -! (http://archives.postgresql.org/). -! If it has not and the patch is of any complexity it is strongly -! recommended you post a message to the appropriate list or you risk -! getting your patch rejected. Refer back to 1.4 for -! email guidelines.
    2. -! -!
  2. Ensure that your patch is generated against the most recent version -! of the code, which for developers is CVS HEAD. For more on branches in -! PostgreSQL, see 1.15.
    3. -! -!
  3. Try to make your patch as readable as possible. Try to follow the -! project's code-layout conventions; again, this makes it easier for the -! reviewer, and there's no point in trying to do it -! differently than pgindent. Also avoid unnecessary whitespace -! changes, they just distract the reviewer, and your formatting changes -! will probably not survive the next pgindent run anyway.
    4. -! -!
  4. The patch should be generated in contextual diff format and should -! be applicable from the root directory. If you are unfamiliar with -! this, you might find the script src/tools/makediff/difforig -! useful.
    5. -! -!
  5. PostgreSQL is licensed under a BSD license, so any submissions must -! conform to the BSD license to be included. If you use code that is -! available under some other license that is BSD compatible (eg. public -! domain) please note that code in your email submission
    6. -! -!
  6. Confirm that your changes can pass the regression tests and list the -! platforms you have tested this on. If your changes are port specific, -! list the ports that it applies to.
    7. -! -!
  7. Provide an implementation overview, preferably in code comments.
    8. -! -!
  8. If it is a performance patch, provide confirming test results to -! show the benefits of your patch. It is OK to post patches without -! this information, though the patch will not be applied until *somebody* -! has tested the patches and found a valuable performance effect directly -! attributable to the patch. Given that writing performance tests is not -! terribly exciting, it is recommended you take this task upon yourself.
    9. -! -!
  9. If it is a new feature patch, confirm that it has been tested for -! all desired scenarios. If it has not, this should be clearly stated as -! a request for a particular kind of test to be performed. Note that the -! patch will go no further until that test has been performed.
    10. -! -!
  10. New feature patches should also be accompanied by doc patches, and -! pointers to any relevant sections of the SQL standard are recommended -! as well. See 1.16 for more information on the -! SQL standards.
    11. -! -!
  11. If your patch changes any existing defaults, you will need to -! explain why this is *required* or the patch will likely be rejected.
    12. -!
-! -!

Even if you pass all of the above, the patch may still be rejected -! for other technical reasons. You should be prepared to listen to -! comments received and perform any agreed rework. Even if you have -! received positive comments from some community members, others may spot -! problems with your approach, coding style or many other issues.

-! -!

Successful patches will be notified to you by email and you will be -! credited for that work in the next set of release notes.

- -

1.6) Where can I learn more about the - code?

+From pgsql-performance-owner+M17204@postgresql.org Wed Feb 15 16:28:34 2006 +Return-path: +Received: from ams.hub.org (ams.hub.org [200.46.204.13]) + by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id k1FLSV527014 + for ; Wed, 15 Feb 2006 16:28:31 -0500 (EST) +Received: from postgresql.org (postgresql.org [200.46.204.71]) + by ams.hub.org (Postfix) with ESMTP id 168C967B584; + Wed, 15 Feb 2006 17:28:29 -0400 (AST) +X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org +Received: from localhost (av.hub.org [200.46.204.144]) + by postgresql.org (Postfix) with ESMTP id BB0AB9DCB9E + for ; Wed, 15 Feb 2006 17:27:56 -0400 (AST) +Received: from postgresql.org ([200.46.204.71]) + by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) + with ESMTP id 22055-07 + for ; + Wed, 15 Feb 2006 17:27:57 -0400 (AST) +X-Greylist: from auto-whitelisted by SQLgrey- +Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) + by postgresql.org (Postfix) with ESMTP id F385E9DCB98 + for ; Wed, 15 Feb 2006 17:27:53 -0400 (AST) +Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) + by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id k1FLRsqd019780; + Wed, 15 Feb 2006 16:27:54 -0500 (EST) +To: Gary Doades +cc: pgsql-performance@postgresql.org +Subject: Re: [PERFORM] Strange Create Index behaviour +In-Reply-To: <19510.1140036968@sss.pgh.pa.us> +References: <43F38867.6010701@gpdnet.co.uk> <19510.1140036968@sss.pgh.pa.us> +Comments: In-reply-to Tom Lane + message dated "Wed, 15 Feb 2006 15:56:08 -0500" +Date: Wed, 15 Feb 2006 16:27:54 -0500 +Message-ID: <19779.1140038874@sss.pgh.pa.us> +From: Tom Lane +X-Virus-Scanned: by amavisd-new at hub.org +X-Spam-Status: No, score=0.11 required=5 tests=[AWL=0.110] +X-Spam-Score: 0.11 +X-Mailing-List: pgsql-performance +List-Archive: +List-Help: +List-Id: +List-Owner: +List-Post: +List-Subscribe: +List-Unsubscribe: +Precedence: bulk +Sender: pgsql-performance-owner@postgresql.org +Status: ORr + +I wrote: +> Interesting. I tried your test script and got fairly close times +> for all the cases on two different machines: +> old HPUX machine: shortest 5800 msec, longest 7960 msec +> new Fedora 4 machine: shortest 461 msec, longest 608 msec + +> So what this looks like to me is a corner case that FreeBSD's qsort +> fails to handle well. + +I tried forcing PG to use src/port/qsort.c on the Fedora machine, +and lo and behold: + new Fedora 4 machine: shortest 434 msec, longest 8530 msec + +So it sure looks like this script does expose a problem on BSD-derived +qsorts. Curiously, the case that's much the worst for me is the third +in the script, while the shortest time is the first case, which was slow +for Gary. So I'd venture that the *BSD code has been tweaked somewhere +along the way, in a manner that moves the problem around without really +fixing it. (Anyone want to compare the actual FreeBSD source to what +we have?) + +This is pretty relevant stuff, because there was a thread recently +advocating that we stop using the platform qsort on all platforms: +http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php + +It's really interesting to see a case where port/qsort is radically +worse than other qsorts ... unless we figure that out and fix it, +I think the idea of using port/qsort everywhere has just taken a +major hit. + + regards, tom lane + +---------------------------(end of broadcast)--------------------------- +TIP 5: don't forget to increase your free space map settings + +From pgsql-performance-owner+M17212@postgresql.org Wed Feb 15 18:29:07 2006 +Return-path: +Received: from ams.hub.org (ams.hub.org [200.46.204.13]) + by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id k1FNT6509074 + for ; Wed, 15 Feb 2006 18:29:06 -0500 (EST) +Received: from postgresql.org (postgresql.org [200.46.204.71]) + by ams.hub.org (Postfix) with ESMTP id 2BE6267B58B; + Wed, 15 Feb 2006 19:29:04 -0400 (AST) +X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org +Received: from localhost (av.hub.org [200.46.204.144]) + by postgresql.org (Postfix) with ESMTP id 7C3D49DC803; + Wed, 15 Feb 2006 19:28:30 -0400 (AST) +Received: from postgresql.org ([200.46.204.71]) + by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) + with ESMTP id 47149-10; Wed, 15 Feb 2006 19:28:32 -0400 (AST) +X-Greylist: from auto-whitelisted by SQLgrey- +X-Greylist: from auto-whitelisted by SQLgrey- +Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) + by postgresql.org (Postfix) with ESMTP id C56AD9DC843; + Wed, 15 Feb 2006 19:28:27 -0400 (AST) +Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) + by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id k1FNSTkm020782; + Wed, 15 Feb 2006 18:28:29 -0500 (EST) +To: Gary Doades +cc: pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org +Subject: qsort again (was Re: [PERFORM] Strange Create Index behaviour) +In-Reply-To: <43F39E53.1020009@gpdnet.co.uk> +References: <43F38867.6010701@gpdnet.co.uk> <19510.1140036968@sss.pgh.pa.us> <19779.1140038874@sss.pgh.pa.us> <43F39E53.1020009@gpdnet.co.uk> +Comments: In-reply-to Gary Doades + message dated "Wed, 15 Feb 2006 21:34:11 +0000" +Date: Wed, 15 Feb 2006 18:28:29 -0500 +Message-ID: <20781.1140046109@sss.pgh.pa.us> +From: Tom Lane +X-Virus-Scanned: by amavisd-new at hub.org +X-Spam-Status: No, score=0.11 required=5 tests=[AWL=0.110] +X-Spam-Score: 0.11 +X-Mailing-List: pgsql-performance +List-Archive: +List-Help: +List-Id: +List-Owner: +List-Post: +List-Subscribe: +List-Unsubscribe: +Precedence: bulk +Sender: pgsql-performance-owner@postgresql.org +Status: OR + +Gary Doades writes: +> If I run the script again, it is not always the first case that is slow, +> it varies from run to run, which is why I repeated it quite a few times +> for the test. + +For some reason I hadn't immediately twigged to the fact that your test +script is just N repetitions of the exact same structure with random data. +So it's not so surprising that you get random variations in behavior +with different test data sets. + +I did some experimentation comparing the qsort from Fedora Core 4 +(glibc-2.3.5-10.3) with our src/port/qsort.c. For those who weren't +following the pgsql-performance thread, the test case is just this +repeated a lot of times: + +create table atest(i int4, r int4); +insert into atest (i,r) select generate_series(1,100000), 0; +insert into atest (i,r) select generate_series(1,100000), random()*100000; +\timing +create index idx on atest(r); +\timing +drop table atest; + +I did this 100 times and sorted the reported runtimes. (Investigation +with trace_sort = on confirms that the runtime is almost entirely spent +in qsort() called from our performsort --- the Postgres overhead is +about 100msec on this machine.) Results are below. + +It seems clear that our qsort.c is doing a pretty awful job of picking +qsort pivots, while glibc is mostly managing not to make that mistake. +I haven't looked at the glibc code yet to see what they are doing +differently. + +I'd say this puts a considerable damper on my enthusiasm for using our +qsort all the time, as was recently debated in this thread: +http://archives.postgresql.org/pgsql-hackers/2005-12/msg00610.php +We need to fix our qsort.c before pushing ahead with that idea. + + regards, tom lane + + +100 runtimes for glibc qsort, sorted ascending: + +Time: 459.860 ms +Time: 460.209 ms +Time: 460.704 ms +Time: 461.317 ms +Time: 461.538 ms +Time: 461.652 ms +Time: 461.988 ms +Time: 462.573 ms +Time: 462.638 ms +Time: 462.716 ms +Time: 462.917 ms +Time: 463.219 ms +Time: 463.455 ms +Time: 463.650 ms +Time: 463.723 ms +Time: 463.737 ms +Time: 463.750 ms +Time: 463.852 ms +Time: 463.964 ms +Time: 463.988 ms +Time: 464.003 ms +Time: 464.135 ms +Time: 464.372 ms +Time: 464.458 ms +Time: 464.496 ms +Time: 464.551 ms +Time: 464.599 ms +Time: 464.655 ms +Time: 464.656 ms +Time: 464.722 ms +Time: 464.814 ms +Time: 464.827 ms +Time: 464.878 ms +Time: 464.899 ms +Time: 464.905 ms +Time: 464.987 ms +Time: 465.055 ms +Time: 465.138 ms +Time: 465.159 ms +Time: 465.194 ms +Time: 465.310 ms +Time: 465.316 ms +Time: 465.375 ms +Time: 465.450 ms +Time: 465.535 ms +Time: 465.595 ms +Time: 465.680 ms +Time: 465.769 ms +Time: 465.865 ms +Time: 465.892 ms +Time: 465.903 ms +Time: 466.003 ms +Time: 466.154 ms +Time: 466.164 ms +Time: 466.203 ms +Time: 466.305 ms +Time: 466.344 ms +Time: 466.364 ms +Time: 466.388 ms +Time: 466.502 ms +Time: 466.593 ms +Time: 466.725 ms +Time: 466.794 ms +Time: 466.798 ms +Time: 466.904 ms +Time: 466.971 ms +Time: 466.997 ms +Time: 467.122 ms +Time: 467.146 ms +Time: 467.221 ms +Time: 467.224 ms +Time: 467.244 ms +Time: 467.277 ms +Time: 467.587 ms +Time: 468.142 ms +Time: 468.207 ms +Time: 468.237 ms +Time: 468.471 ms +Time: 468.663 ms +Time: 468.700 ms +Time: 469.235 ms +Time: 469.840 ms +Time: 470.472 ms +Time: 471.140 ms +Time: 472.811 ms +Time: 472.959 ms +Time: 474.858 ms +Time: 477.210 ms +Time: 479.571 ms +Time: 479.671 ms +Time: 482.797 ms +Time: 488.852 ms +Time: 514.639 ms +Time: 529.287 ms +Time: 612.185 ms +Time: 660.748 ms +Time: 742.227 ms +Time: 866.814 ms +Time: 1234.848 ms +Time: 1267.398 ms + + +100 runtimes for port/qsort.c, sorted ascending: + +Time: 418.905 ms +Time: 420.611 ms +Time: 420.764 ms +Time: 420.904 ms +Time: 421.706 ms +Time: 422.466 ms +Time: 422.627 ms +Time: 423.189 ms +Time: 423.302 ms +Time: 425.096 ms +Time: 425.731 ms +Time: 425.851 ms +Time: 427.253 ms +Time: 430.113 ms +Time: 432.756 ms +Time: 432.963 ms +Time: 440.502 ms +Time: 440.640 ms +Time: 450.452 ms +Time: 458.143 ms +Time: 459.212 ms +Time: 467.706 ms +Time: 468.006 ms +Time: 468.574 ms +Time: 470.003 ms +Time: 472.313 ms +Time: 483.622 ms +Time: 492.395 ms +Time: 509.564 ms +Time: 531.037 ms +Time: 533.366 ms +Time: 535.610 ms +Time: 575.523 ms +Time: 582.688 ms +Time: 593.545 ms +Time: 647.364 ms +Time: 660.612 ms +Time: 677.312 ms +Time: 680.288 ms +Time: 697.626 ms +Time: 833.066 ms +Time: 834.511 ms +Time: 851.819 ms +Time: 920.443 ms +Time: 926.731 ms +Time: 954.289 ms +Time: 1045.214 ms +Time: 1059.200 ms +Time: 1062.328 ms +Time: 1136.018 ms +Time: 1260.091 ms +Time: 1276.883 ms +Time: 1319.351 ms +Time: 1438.854 ms +Time: 1475.457 ms +Time: 1538.211 ms +Time: 1549.004 ms +Time: 1744.642 ms +Time: 1771.258 ms +Time: 1959.530 ms +Time: 2300.140 ms +Time: 2589.641 ms +Time: 2612.780 ms +Time: 3100.024 ms +Time: 3284.125 ms +Time: 3379.792 ms +Time: 3750.278 ms +Time: 4302.278 ms +Time: 4780.624 ms +Time: 5000.056 ms +Time: 5092.604 ms +Time: 5168.722 ms +Time: 5292.941 ms +Time: 5895.964 ms +Time: 7003.164 ms +Time: 7099.449 ms +Time: 7115.083 ms +Time: 7384.940 ms +Time: 8214.010 ms +Time: 8700.771 ms +Time: 9331.225 ms +Time: 10503.360 ms +Time: 12496.026 ms +Time: 12982.474 ms +Time: 15192.390 ms +Time: 15392.161 ms +Time: 15958.295 ms +Time: 18375.693 ms +Time: 18617.706 ms +Time: 18927.515 ms +Time: 19898.018 ms +Time: 20865.979 ms +Time: 21000.907 ms +Time: 21297.585 ms +Time: 21714.518 ms +Time: 25423.235 ms +Time: 27543.052 ms +Time: 28314.182 ms +Time: 29400.278 ms +Time: 34142.534 ms + +---------------------------(end of broadcast)--------------------------- +TIP 1: if posting/reading through Usenet, please send an appropriate + subscribe-nomail command to majordomo@postgresql.org so that your + message can get through to the mailing list cleanly + +From pgsql-hackers-owner+M79733@postgresql.org Wed Feb 15 20:22:07 2006 +Return-path: +Received: from ams.hub.org (ams.hub.org [200.46.204.13]) + by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id k1G1M6529533 + for ; Wed, 15 Feb 2006 20:22:06 -0500 (EST) +Received: from postgresql.org (postgresql.org [200.46.204.71]) + by ams.hub.org (Postfix) with ESMTP id E5C5467B58F; + Wed, 15 Feb 2006 21:22:03 -0400 (AST) +X-Original-To: pgsql-hackers-postgresql.org@localhost.postgresql.org +Received: from localhost (av.hub.org [200.46.204.144]) + by postgresql.org (Postfix) with ESMTP id 3DAA69DCACE; + Wed, 15 Feb 2006 21:21:34 -0400 (AST) +Received: from postgresql.org ([200.46.204.71]) + by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) + with ESMTP id 76351-01; Wed, 15 Feb 2006 21:21:36 -0400 (AST) +X-Greylist: from auto-whitelisted by SQLgrey- +X-Greylist: from auto-whitelisted by SQLgrey- +Received: from sss.pgh.pa.us (sss.pgh.pa.us [66.207.139.130]) + by postgresql.org (Postfix) with ESMTP id 2FBB59DCA3F; + Wed, 15 Feb 2006 21:21:31 -0400 (AST) +Received: from sss2.sss.pgh.pa.us (tgl@localhost [127.0.0.1]) + by sss.pgh.pa.us (8.13.1/8.13.1) with ESMTP id k1G1LXXi021616; + Wed, 15 Feb 2006 20:21:33 -0500 (EST) +To: Ron +cc: pgsql-performance@postgresql.org, pgsql-hackers@postgresql.org +Subject: Re: [HACKERS] qsort again (was Re: [PERFORM] Strange Create Index behaviour) +In-Reply-To: <7.0.1.0.2.20060215194635.03b55da0@earthlink.net> +References: <43F38867.6010701@gpdnet.co.uk> <19510.1140036968@sss.pgh.pa.us> <19779.1140038874@sss.pgh.pa.us> <43F39E53.1020009@gpdnet.co.uk> <20781.1140046109@sss.pgh.pa.us> <7.0.1.0.2.20060215194635.03b55da0@earthlink.net> +Comments: In-reply-to Ron + message dated "Wed, 15 Feb 2006 19:57:51 -0500" +Date: Wed, 15 Feb 2006 20:21:33 -0500 +Message-ID: <21615.1140052893@sss.pgh.pa.us> +From: Tom Lane +X-Virus-Scanned: by amavisd-new at hub.org +X-Spam-Status: No, score=0.11 required=5 tests=[AWL=0.110] +X-Spam-Score: 0.11 +X-Mailing-List: pgsql-hackers +List-Archive: +List-Help: +List-Id: +List-Owner: +List-Post: +List-Subscribe: +List-Unsubscribe: +Precedence: bulk +Sender: pgsql-hackers-owner@postgresql.org +Status: OR + +Ron writes: +> How are we choosing our pivots? + +See qsort.c: it looks like median of nine equally spaced inputs (ie, +the 1/8th points of the initial input array, plus the end points), +implemented as two rounds of median-of-three choices. With half of the +data inputs zero, it's not too improbable for two out of the three +samples to be zeroes in which case I think the med3 result will be zero +--- so choosing a pivot of zero is much more probable than one would +like, and doing so in many levels of recursion causes the problem. + +I think. I'm not too sure if the code isn't just being sloppy about the +case where many data values are equal to the pivot --- there's a special +case there to switch to insertion sort, and maybe that's getting invoked +too soon. It'd be useful to get a line-level profile of the behavior of +this code in the slow cases... + + regards, tom lane + +---------------------------(end of broadcast)--------------------------- +TIP 3: Have you checked our extensive FAQ? + + http://www.postgresql.org/docs/faq + +From pgsql-performance-owner+M17282@postgresql.org Fri Feb 17 23:11:11 2006 +Return-path: +Received: from ams.hub.org (ams.hub.org [200.46.204.13]) + by candle.pha.pa.us (8.11.6/8.11.6) with ESMTP id k1I4BA515503 + for ; Fri, 17 Feb 2006 23:11:10 -0500 (EST) +Received: from postgresql.org (postgresql.org [200.46.204.71]) + by ams.hub.org (Postfix) with ESMTP id 2825F67B5F5; + Sat, 18 Feb 2006 00:11:07 -0400 (AST) +X-Original-To: pgsql-performance-postgresql.org@localhost.postgresql.org +Received: from localhost (av.hub.org [200.46.204.144]) + by postgresql.org (Postfix) with ESMTP id 7BB8A9DCC4F; + Wed, 15 Feb 2006 21:37:57 -0400 (AST) +Received: from postgresql.org ([200.46.204.71]) + by localhost (av.hub.org [200.46.204.144]) (amavisd-new, port 10024) + with ESMTP id 79365-02; Wed, 15 Feb 2006 21:38:00 -0400 (AST) +X-Greylist: from auto-whitelisted by SQLgrey- +X-Greylist: from auto-whitelisted by SQLgrey- +Received: from postal.corporate.connx.com (postal.corporate.connx.com [65.212.159.187]) + by postgresql.org (Postfix) with ESMTP id 33BEA9DCACE; + Wed, 15 Feb 2006 21:37:54 -0400 (AST) +X-MimeOLE: Produced By Microsoft Exchange V6.5 +Content-class: urn:content-classes:message +MIME-Version: 1.0 +Content-Type: text/plain; + charset="us-ascii" +Subject: Re: [HACKERS] qsort again (was Re: [PERFORM] Strange Create Index behaviour) +Date: Wed, 15 Feb 2006 17:37:58 -0800 +Message-ID: +Thread-Topic: [HACKERS] qsort again (was Re: [PERFORM] Strange Create Index behaviour) +Thread-Index: AcYyl2fPgxfNXHIRRyOEN4ZGeHtA3wAAEaNQ +From: "Dann Corbit" +To: "Tom Lane" , "Ron" +cc: , +X-Virus-Scanned: by amavisd-new at hub.org +X-Spam-Status: No, score=0.075 required=5 tests=[AWL=0.075] +X-Spam-Score: 0.075 +X-Mailing-List: pgsql-performance +List-Archive: +List-Help: +List-Id: +List-Owner: +List-Post: +List-Subscribe: +List-Unsubscribe: +Precedence: bulk +Sender: pgsql-performance-owner@postgresql.org +Content-Transfer-Encoding: 8bit +X-MIME-Autoconverted: from quoted-printable to 8bit by candle.pha.pa.us id k1I4BA515503 +Status: ORr + + + +> -----Original Message----- +> From: pgsql-hackers-owner@postgresql.org [mailto:pgsql-hackers- +> owner@postgresql.org] On Behalf Of Tom Lane +> Sent: Wednesday, February 15, 2006 5:22 PM +> To: Ron +> Cc: pgsql-performance@postgresql.org; pgsql-hackers@postgresql.org +> Subject: Re: [HACKERS] qsort again (was Re: [PERFORM] Strange Create +Index +> behaviour) +> +> Ron writes: +> > How are we choosing our pivots? +> +> See qsort.c: it looks like median of nine equally spaced inputs (ie, +> the 1/8th points of the initial input array, plus the end points), +> implemented as two rounds of median-of-three choices. With half of +the +> data inputs zero, it's not too improbable for two out of the three +> samples to be zeroes in which case I think the med3 result will be +zero +> --- so choosing a pivot of zero is much more probable than one would +> like, and doing so in many levels of recursion causes the problem. + +Adding some randomness to the selection of the pivot is a known +technique to fix the oddball partitions problem. However, Bentley and +Sedgewick proved that every quick sort algorithm has some input set that +makes it go quadratic (hence the recent popularity of introspective +sort, which switches to heapsort if quadratic behavior is detected. The +C++ template I submitted was an example of introspective sort, but +PostgreSQL does not use C++ so it was not helpful). + +> I think. I'm not too sure if the code isn't just being sloppy about +the +> case where many data values are equal to the pivot --- there's a +special +> case there to switch to insertion sort, and maybe that's getting +invoked +> too soon. + +Here are some cases known to make qsort go quadratic: +1. Data already sorted +2. Data reverse sorted +3. Data organ-pipe sorted or ramp +4. Almost all data of the same value + +There are probably other cases. Randomizing the pivot helps some, as +does check for in-order or reverse order partitions. + +Imagine if 1/3 of the partitions fall into a category that causes +quadratic behavior (have one of the above formats and have more than +CUTOFF elements in them). + +It is doubtful that the switch to insertion sort is causing any sort of +problems. It is only going to be invoked on tiny sets, for which it has +a fixed cost that is probably less that qsort() function calls on sets +of the same size. + +>It'd be useful to get a line-level profile of the behavior of +> this code in the slow cases... + +I guess that my in-order or presorted tests [which often arise when +there are very few distinct values] may solve the bad partition +problems. Don't forget that the algorithm is called recursively. + +> regards, tom lane +> +> ---------------------------(end of +broadcast)--------------------------- +> TIP 3: Have you checked our extensive FAQ? +> +> http://www.postgresql.org/docs/faq + +---------------------------(end of broadcast)--------------------------- +TIP 2: Don't 'kill -9' the postmaster +