[mkgmap-dev] FW: Splitter defaulting on Japan

Sat May 31 07:33:20 BST 2014

Hi Felix,

it seems that the default --search-limit is too low for japan with sea data.
With --search-limit=5000000 I see a very good split after ~30 seconds:
Solution is nice. Can't find a better solution: 86 tile(s). The smallest node count is 1111027 (92 %), the worst aspect ratio is near 3.94

For many other input files, this will just increase run time a lot.

I am not very happy with this option. I think I will replace it
with something more useful like
--wanted-fill-ratio=90
which means try to find a solution where the least populated
tile has 90% of max-nodes.

The problem is that some areas do not allow such a good result,
so I am still looking for good criteria to stop the search, maybe
I'll use the overall search time in seconds:
--max-search-time=60
would use the best result found after 60 seconds.

Any other ideas?

Gerd

From: gpetermann_muenchen at hotmail.com
To: mkgmap-dev at lists.mkgmap.org.uk
Date: Sat, 31 May 2014 08:21:43 +0200
Subject: [mkgmap-dev] FW:  Splitter defaulting on Japan

From: gpetermann_muenchen at hotmail.com
To: extremecarver at gmail.com
Subject: RE: [mkgmap-dev] Splitter defaulting on Japan
Date: Sat, 31 May 2014 08:17:31 +0200

Hi Felix,

thanks for reporting. I can reproduce the problem. It disappears when
you remove the --precomp-sea option.
I'll try to find out why.

> Where is the log saved? Do I need to provide it? Splitter not fully up 
This message is printed to stderr for the case that you pipe stdout to a file like this:
java -jar splitter ....   > splitter.log

Gerd

> Date: Sat, 31 May 2014 05:33:06 +0200
> From: extremecarver at gmail.com
> To: mkgmap-dev at lists.mkgmap.org.uk
> Subject: [mkgmap-dev] Splitter defaulting on Japan
> 
> Failed to calculate areas. See log for details.
> 
> Where is the log saved? Do I need to provide it? Splitter not fully up 
> to date - see compiled date (svn update just before)...
> I'll retry with up to date splitter still...
> 
> c:\OpenMTBMap\maps>java -Xmx9600m -jar c:\openmtbmap\splitter.jar 
> --precomp-sea=c:\openmtbmap\maps\sea.zip --max-nodes=1200000 
> --output=pbf "--keep-complete" --max-areas=255 
> --geonames-file=cities5000 --description=japan --mapid=65590000 c:\OpenMTBMa
> p\osmpbf_geofabrik\japan.o5m
> Splitter version unknown compiled 2014-05-28T18:16:29+0200
> boundary-tags=use-exclude-list
> cache=
> description=japan
> geonames-file=cities5000
> keep-complete=true
> mapid=65590000
> max-areas=255
> max-nodes=1200000
> max-threads=8 (auto)
> mixed=false
> no-trim=false
> num-tiles=
> output=pbf
> output-dir=
> overlap=auto
> polygon-desc-file=
> polygon-file=
> precomp-sea=c:\openmtbmap\maps\sea.zip
> problem-file=
> problem-report=
> resolution=13
> search-limit=1000000
> split-file=
> status-freq=120
> stop-after=dist
> write-kml=
> Elapsed time: 0s   Memory: Current 184MB (2MB used, 182MB free) Max 8533MB
> Time started: Sat May 31 02:40:37 CEST 2014
> Map is being split for resolution 13:
>   - area boundaries are aligned to 0x800 map units (0.0439453125 degrees)
>   - areas are multiples of 0x800 map units wide and high
> Processing c:\OpenMTBMap\osmpbf_geofabrik\japan.o5m
> Bounding box 122.56070000000001 20.08228 154.4709 45.80245
> 10'000'000 nodes processed... id=752184948
> 20'000'000 nodes processed... id=922482748
> 30'000'000 nodes processed... id=1233696234
> 40'000'000 nodes processed... id=1315824764
> 50'000'000 nodes processed... id=1421002430
> 60'000'000 nodes processed... id=1497506798
> 70'000'000 nodes processed... id=1751094845
> 80'000'000 nodes processed... id=2040104150
> 90'000'000 nodes processed... id=2350328457
> in 1 file
> Time: Sat May 31 02:41:01 CEST 2014
> Counting nodes of precompiled sea data ...
> Bounding box 144.84375 19.6875 145.546875 20.390625
> Bounding box 135.703124999 20.390625 136.406249999 21.09375
> Bounding box 144.84375 20.390625 145.546875 21.09375
> Bounding box 122.34375000000001 23.90625 123.04687499900001 24.609375
> 
> 
> ........
> cut away here...
> .......
> Bounding box 142.734375 45.703125 143.4375 46.40625
> Bounding box 143.4375 45.703125 144.140625 46.40625
> Bounding box 149.0625 45.703125 149.765625 46.40625
> Bounding box 149.765625 45.703125 150.46875 46.40625
> Bounding box 150.46875 45.703125 151.171875 46.40625
> Added 835594 nodes from precompiled sea data.
> Precompiled sea data pass took 1797 ms
> Exact map coverage is (20.08227825164795,122.56068706512451) to 
> (45.80243110656738,154.47088479995728)
> Rounded map coverage is (20.0390625,122.51953125) to 
> (45.8349609375,154.51171875)
> Splitting nodes into areas containing a maximum of 1'200'000 nodes each...
> Highest node count in a single grid element is 205'139
> Trying to find nice split for (20.0390625,122.51953125) to 
> (45.8349609375,154.51171875) with 98'482'055 nodes
> searching for split with min-nodes 12000, learned 0 good partial solutions
> Split was not yet succesfull. Trying to remove large empty areas...
> Trying again with 1 trimmed partition(s), also allowing big empty parts.
> Solving partition (23.818359375,122.6953125) to 
> (45.8349609375,153.0615234375) with 98'482'055 nodes
> Trying to find nice split for (23.818359375,122.6953125) to 
> (45.8349609375,153.0615234375) with 98'482'055 nodes
> searching for split with min-nodes 12000, learned 0 good partial solutions
> Warning: No solution found for partition (23.818359375,122.6953125) to 
> (45.8349609375,153.0615234375) with 98'482'055 nodes
> Final solution has 0 tile(s). The smallest node count is 
> 9223372036854775807 (0 %), the worst aspect ratio is near -1.0
> Failed to calculate areas. See log for details.
> Failed to calculate areas.
> Sorry. Cannot split the file without creating huge, almost empty, tiles.
> Please specify a bounding polygon with the --polygon-file parameter.
> Time finished: Sat May 31 02:41:04 CEST 2014
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

_______________________________________________
mkgmap-dev mailing list
mkgmap-dev at lists.mkgmap.org.uk
http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.mkgmap.org.uk/pipermail/mkgmap-dev/attachments/20140531/1adc1123/attachment-0001.html>