Commit: r2159: Improving performance of remove-short-arcs option

Version 2159 was commited by wanmil on 2012-01-02 21:17:38 +0000 (Mon, 02 Jan 2012) Improving performance of remove-short-arcs option The information about the arcs is now stored directly in the Coord object. --GerdP

Version 2159 was commited by wanmil on 2012-01-02 21:17:38 +0000 (Mon, 02 Jan 2012)
Improving performance of remove-short-arcs option
The information about the arcs is now stored directly in the Coord object. --GerdP
Hi all, all available performance patches are now commited. A small comparison with my small test map (15 european tiles, one thread, index with bounds) - not representative like all performance comparisons: r2153: 320s mean r2159: 250s mean improvement: 70s => ~20% A big thank to Gerd, who posted most of the improvements! WanMil

Hello Gerd, hello WanMil, thanks for the improvement of the LocationHook. On my Mac, using 2 threads, I just get 10% improvement, but this doesn't matter. Is it somehow possible, to create more than one entry in the index for one street, with the same name on the same tile?! Sounds confusing, I know. But let me give you an example: In Berlin more than one Friedrichstraße exist (having different zip-codes). But currently I can only find one Friedrichstraße. So maybe we can use the zip-code to distinguish the streets. Or you have a another idea. It would be also nice to get the zip-code search working, but it's just a dream ;) Cheers and thank you again for your great work! Martin Am 2012-01-02 22:37, schrieb WanMil:
Version 2159 was commited by wanmil on 2012-01-02 21:17:38 +0000 (Mon, 02 Jan 2012)
Improving performance of remove-short-arcs option
The information about the arcs is now stored directly in the Coord object. --GerdP Hi all,
all available performance patches are now commited. A small comparison with my small test map (15 european tiles, one thread, index with bounds) - not representative like all performance comparisons:
r2153: 320s mean r2159: 250s mean
improvement: 70s => ~20%
A big thank to Gerd, who posted most of the improvements!
WanMil
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Hello Martin, sorry, I am still far away from understanding how the indexes work. Gerd
Date: Tue, 3 Jan 2012 11:08:54 +0100 From: mkgmap@snailrun.de To: mkgmap-dev@lists.mkgmap.org.uk Subject: Re: [mkgmap-dev] Commit: r2159: Improving performance of remove-short-arcs option
Hello Gerd, hello WanMil,
thanks for the improvement of the LocationHook. On my Mac, using 2 threads, I just get 10% improvement, but this doesn't matter. Is it somehow possible, to create more than one entry in the index for one street, with the same name on the same tile?! Sounds confusing, I know. But let me give you an example: In Berlin more than one Friedrichstraße exist (having different zip-codes). But currently I can only find one Friedrichstraße. So maybe we can use the zip-code to distinguish the streets. Or you have a another idea. It would be also nice to get the zip-code search working, but it's just a dream ;)
Cheers and thank you again for your great work! Martin
Am 2012-01-02 22:37, schrieb WanMil:
Version 2159 was commited by wanmil on 2012-01-02 21:17:38 +0000 (Mon, 02 Jan 2012)
Improving performance of remove-short-arcs option
The information about the arcs is now stored directly in the Coord object. --GerdP Hi all,
all available performance patches are now commited. A small comparison with my small test map (15 european tiles, one thread, index with bounds) - not representative like all performance comparisons:
r2153: 320s mean r2159: 250s mean
improvement: 70s => ~20%
A big thank to Gerd, who posted most of the improvements!
WanMil
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Martin, sounds more like a problem of the index creation and not like a problem of the LocationHook. Can you please post a new thread with an exact description how to reproduce the problem, what you expect and what you see? Thanks! WanMil
Hello Gerd, hello WanMil,
thanks for the improvement of the LocationHook. On my Mac, using 2 threads, I just get 10% improvement, but this doesn't matter. Is it somehow possible, to create more than one entry in the index for one street, with the same name on the same tile?! Sounds confusing, I know. But let me give you an example: In Berlin more than one Friedrichstraße exist (having different zip-codes). But currently I can only find one Friedrichstraße. So maybe we can use the zip-code to distinguish the streets. Or you have a another idea. It would be also nice to get the zip-code search working, but it's just a dream ;)
Cheers and thank you again for your great work! Martin
Am 2012-01-02 22:37, schrieb WanMil:
Version 2159 was commited by wanmil on 2012-01-02 21:17:38 +0000 (Mon, 02 Jan 2012)
Improving performance of remove-short-arcs option
The information about the arcs is now stored directly in the Coord object. --GerdP Hi all,
all available performance patches are now commited. A small comparison with my small test map (15 european tiles, one thread, index with bounds) - not representative like all performance comparisons:
r2153: 320s mean r2159: 250s mean
improvement: 70s => ~20%
A big thank to Gerd, who posted most of the improvements!
WanMil
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Hello, I already sent this to the list, but maybe it was somehow lost. http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2011q4/013071.html With Version 2097 I was able to find more than one Friedrichstraße (Lichterfelde 12205, Spandau 13585 and Mitte 10117) in Berlin. With Version 2098 I only could found one Friedrichstraße (in Spandau) (but I can find them if I have just one tile on my Garmin and do not enter the city (http://youtu.be/8WETq30IFJ4)). My idea was, to create for every Friedrichstraße, which has a different zip-code, an entry in the index. I do not really know how the index work (is there any documentation?!), but as I said, it's just an idea. Cheers, Martin

Hi, On Tue, Jan 03, Martin wrote:
Hello,
I already sent this to the list, but maybe it was somehow lost. http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2011q4/013071.html With Version 2097 I was able to find more than one Friedrichstraße (Lichterfelde 12205, Spandau 13585 and Mitte 10117) in Berlin. With Version 2098 I only could found one Friedrichstraße (in Spandau) (but I can find them if I have just one tile on my Garmin and do not enter the city (http://youtu.be/8WETq30IFJ4)). My idea was, to create for every Friedrichstraße, which has a different zip-code, an entry in the index. I do not really know how the index work (is there any documentation?!), but as I said, it's just an idea.
I tested it with my maps, but I'm not able to reproduce. Works fine for me with MapSource and the 62s. http://osm.thkukuk.de/ contains my map, maybe you could try to test that maps to make sure I did really test your case? Thorsten -- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

Hello Thorsten, in Mapsource it also works for me, but not on my Oregon. I will try it with your maps. I also tried it with the default style, so my question is, which settings you use? Which boundary files? Cheers, Martin Am 03.01.2012 um 21:42 schrieb Thorsten Kukuk:
Hi,
On Tue, Jan 03, Martin wrote:
Hello,
I already sent this to the list, but maybe it was somehow lost. http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2011q4/013071.html With Version 2097 I was able to find more than one Friedrichstraße (Lichterfelde 12205, Spandau 13585 and Mitte 10117) in Berlin. With Version 2098 I only could found one Friedrichstraße (in Spandau) (but I can find them if I have just one tile on my Garmin and do not enter the city (http://youtu.be/8WETq30IFJ4)). My idea was, to create for every Friedrichstraße, which has a different zip-code, an entry in the index. I do not really know how the index work (is there any documentation?!), but as I said, it's just an idea.
I tested it with my maps, but I'm not able to reproduce. Works fine for me with MapSource and the 62s. http://osm.thkukuk.de/ contains my map, maybe you could try to test that maps to make sure I did really test your case?
Thorsten
-- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg) _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev

Hi, On Tue, Jan 03, Martin wrote:
Hello Thorsten,
in Mapsource it also works for me, but not on my Oregon. I will try it with your maps. I also tried it with the default style, so my question is, which settings you use? Which boundary files?
Don't have an Oregon for testing, but there shouldn't be real firmware differences to the 62s. The boundarys I'm using are the ones WanMil posted here some time ago. All scripts and options are documented in the style archive available from the webpage, too. Thorsten
Cheers, Martin
Am 03.01.2012 um 21:42 schrieb Thorsten Kukuk:
Hi,
On Tue, Jan 03, Martin wrote:
Hello,
I already sent this to the list, but maybe it was somehow lost. http://www.mkgmap.org.uk/pipermail/mkgmap-dev/2011q4/013071.html With Version 2097 I was able to find more than one Friedrichstraße (Lichterfelde 12205, Spandau 13585 and Mitte 10117) in Berlin. With Version 2098 I only could found one Friedrichstraße (in Spandau) (but I can find them if I have just one tile on my Garmin and do not enter the city (http://youtu.be/8WETq30IFJ4)). My idea was, to create for every Friedrichstraße, which has a different zip-code, an entry in the index. I do not really know how the index work (is there any documentation?!), but as I said, it's just an idea.
I tested it with my maps, but I'm not able to reproduce. Works fine for me with MapSource and the 62s. http://osm.thkukuk.de/ contains my map, maybe you could try to test that maps to make sure I did really test your case?
Thorsten
-- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg) _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
-- Thorsten Kukuk, Project Manager/Release Manager SLES SUSE LINUX Products GmbH, Maxfeldstr. 5, D-90409 Nuernberg GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg)

Hi Martin
In Berlin more than one Friedrichstraße exist (having different zip-codes). But currently I can only find one Friedrichstraße. So maybe we can use the zip-code to distinguish the streets. Or you have a another idea. It would be also nice to get the zip-code search working, but it's just a dream ;)
I haven't forgotten this problem, but I simply have no idea of what the problem is. It is not a simple bug of them just being missing, because they are all shown fine in mapsource, so to find it will require luck a flash of insight, or maybe it only works if there are zip codes or house numbers. I have an idea of something that I will try tomorrow, it may work or it may not... ..Steve

With both my maps I've no difference between Mapsource and the search results in my Nüvi 205. However, my Central Europe map shows both the Friedrichstrasse in Spandau close to the roundabout and the Friedrichstrasse close to the Unter den Linden. My North-West Europe map (build with the same parameters) only presents hits for the Friedrichstrasse in Berlin-Zehlendorf, close to the railway station. Probably the Central Europe map was lucky in the split of the tiles. However, in essence it has to be a boundary problem: all three Berlins Friedrichstrassen are located in three different neighbourhoods of Berlin... Cheers, Johan On 03.01.2012 23:30, Steve Ratcliffe wrote:
Hi Martin
In Berlin more than one Friedrichstraße exist (having different zip-codes). But currently I can only find one Friedrichstraße. So maybe we can use the zip-code to distinguish the streets. Or you have a another idea. It would be also nice to get the zip-code search working, but it's just a dream ;)
I haven't forgotten this problem, but I simply have no idea of what the problem is. It is not a simple bug of them just being missing, because they are all shown fine in mapsource, so to find it will require luck a flash of insight, or maybe it only works if there are zip codes or house numbers.
I have an idea of something that I will try tomorrow, it may work or it may not...
..Steve _______________________________________________ mkgmap-dev mailing list mkgmap-dev@lists.mkgmap.org.uk http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
participants (8)
-
Gerd Petermann
-
Martin
-
Martin
-
navmaps
-
Steve Ratcliffe
-
svn commit
-
Thorsten Kukuk
-
WanMil