Re: [HACKERS] 3D Z-curve spatial index
On Mon, Feb 06, 2017 at 11:04:12AM +0700, Boris Muratshin wrote: > The algorithm for 2D is described in articles (in Russian) > https://habrahabr.ru/post/319810/ and > https://habrahabr.ru/post/319096/ . > Goggle-translator generates (IMHO) suitable for understanding text. Welcome to the community! I apologize for not being more clear. Here, it is customary to send a patch rather than a web link when one has code one wishes to share. This is for several reasons: 1. To establish that you are warranting that you have the right to send the patch, i.e. that it is not legally encumbered in some way. 2. To ensure that the patch stays in the archives, as large web sites have gone away in the past, and will in the future. 3. Because this is how the current development process works. When I mentioned documentation, I was referring to the operation of the SQL-callable interface, assuming that there is one. When people apply the patch, they need to have some idea what it is supposed to do and how to make it do those things. Any difference between what it is supposed to do and what it actually does a bug, whether in the implementation, the documentation, or both. These things and many others are in the https://wiki.postgresql.org/wiki/Developer_FAQ Best, David. -- David Fetterhttp://fetter.org/ Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter Skype: davidfetter XMPP: david(dot)fetter(at)gmail(dot)com Remember to vote! Consider donating to Postgres: http://www.postgresql.org/about/donate -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
Re: [HACKERS] 3D Z-curve spatial index
The algorithm for 2D is described in articles (in Russian) https://habrahabr.ru/post/319810/ and https://habrahabr.ru/post/319096/ . Goggle-translator generates (IMHO) suitable for understanding text. 3D version article is not finished yet. The data in figures are obtained in the following way: 1) Test data set is pseudo 3d (x,0,z) array of 100 000 000 random points 2) R-tree for comparison - GiST 2d (x,z) 3) There is a set of experiments by an average number of points in requested random area: 1,10, 100, 1 000, 10 000, 100 000, 1 000 000 4) For each area size I requested a set of random extents (from 100 000 times for 1 point to 100 times for 1 000 000 points). 5) Experiments were done on virtual machine (2 cores, 4Gb) and to exclude noise, all times were got on second (or more) run to warm caches, reads were got on restarted PosgreSQL. 6) For R-tree, times are very unstable and I used the least one in the series. Regards, Boris On Mon, Feb 6, 2017 at 5:08 AM, David Fetterwrote: > > Please send the actual patch and any documentation you write that > comes with it. > > Best, > David. > -- > David Fetter http://fetter.org/ > Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter > Skype: davidfetter XMPP: david(dot)fetter(at)gmail(dot)com > > Remember to vote! > Consider donating to Postgres: http://www.postgresql.org/about/donate >
Re: [HACKERS] 3D Z-curve spatial index
On Sat, Feb 04, 2017 at 03:41:29PM +0700, Boris Muratshin wrote: > Hi hackers, > > The low-level implementation of 3D Z-curve index > (https://github.com/bmuratshin/zcurve/tree/master) > is getting close to GiST R-Tree performance at > significantly lesser number of pages read from disk. > > See attached figures, > times2 - average request execution time VS average points number in result > reads2 - average shared reads number VS average points number in result > > Feel free to connect with me if you have any questions. Please send the actual patch and any documentation you write that comes with it. Best, David. -- David Fetterhttp://fetter.org/ Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter Skype: davidfetter XMPP: david(dot)fetter(at)gmail(dot)com Remember to vote! Consider donating to Postgres: http://www.postgresql.org/about/donate -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
[HACKERS] 3D Z-curve spatial index
Hi hackers, The low-level implementation of 3D Z-curve index (https://github.com/bmuratshin/zcurve/tree/master) is getting close to GiST R-Tree performance at significantly lesser number of pages read from disk. See attached figures, times2 - average request execution time VS average points number in result reads2 - average shared reads number VS average points number in result Feel free to connect with me if you have any questions. -- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers