[GraphHopper] GraphHopper init slowed down with 6c8f0549

[Thomas Ferris Nicolaisen]

Wow, that sounds great. I'll test this today. Sorry that I took so
long to reply here, was out of town for a couple of days.

On Wed, Dec 4, 2013 at 9:53 PM, Peter K <peathal at yahoo.de> wrote:
> Ok, finally this should be fixed, although I will play a bit around to
> tune it.
>
> Because I could apply a trick that I've tried some months before
> without effect. But now this trick gives us an amazing speed up of the
> preparation - it is only 50% of the original time :) ! E.g. world wide
> routing for bikes can be prepared in only 2h and full Germany is ready
> in 9 minutes (plus approx. 3min for import).
>
> Regards,
> Peter.
>
>
>> Uhm, no. For now just use the proposed workaround via excluding
>> tracktype stuff.
>>
>> I need to completely fix this here ...
>>
>> Regards,
>> Peter.
>>
>>> Fixed via:
>>>
>>> https://github.com/graphhopper/graphhopper/commit/d07ac2463bc1b49bc4b7b264eb8621f3872384ca
>>>
>>>> Hi Thomas,
>>>>
>>>> I can confirm that the change regarding the tracktype today introduced
>>>> this behaviour. But can you still confirm this problem for older
>>>> versions? Don't forget to run a
>>>> ./graphhopper.sh clean
>>>> before you measure for new git checkouts.
>>>>
>>>> BTW: one can measure differences in speed automagically via:
>>>>
>>>> ./graphhopper.sh measurement <your-osm> N
>>>>
>>>> where N is optional and the number of git checkouts to go back in
>>>> history. For every checkout a new <git-message>.properties file is
>>>> created where you can see some perf numbers.
>>>>
>>>> Regards,
>>>> Peter.
>>>>
>>>>> Hi Thomas,
>>>>>
>>>>> on which OS are you on and how much RAM do you give the import process?
>>>>>
>>>>> I've integration and performance tests running and at least until 29.11
>>>>> everything should be fine. I'll investigate.
>>>>>
>>>>> Regards,
>>>>> Peter.
>>>>>
>>>>>
>>>>>>
>>>>>> On Tue, Dec 3, 2013 at 4:04 PM, Thomas Ferris Nicolaisen
>>>>>> <tfnico at gmail.com> wrote:
>>>>>>>
>>>>>>> I'm not sure why this is, but after updating to the latest master, I
>>>>>>> have to wait ~18 minutes for a pretty small pbf
>>>>>>> (koeln-regbez-latest.osm.pbf):
>>>>>>>
>>>>>>
>>>>>>
>>>>>> Upon testing a bit more with older revisions (0.2 tag), this still
>>>>>> occurs randomly, so some other factor is playing in. Please disregard
>>>>>> my "accusation" of the commit in question.
>>>>>>
>>>>>> _______________________________________________
>>>>>> GraphHopper mailing list
>>>>>> GraphHopper at openstreetmap.org
>>>>>> https://lists.openstreetmap.org/listinfo/graphhopper
>>>>>>
>>>>>
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> GraphHopper mailing list
>>>>> GraphHopper at openstreetmap.org
>>>>> https://lists.openstreetmap.org/listinfo/graphhopper
>>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> GraphHopper mailing list
>>>> GraphHopper at openstreetmap.org
>>>> https://lists.openstreetmap.org/listinfo/graphhopper
>>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> GraphHopper mailing list
>>> GraphHopper at openstreetmap.org
>>> https://lists.openstreetmap.org/listinfo/graphhopper
>>>
>>
>>
>>
>> _______________________________________________
>> GraphHopper mailing list
>> GraphHopper at openstreetmap.org
>> https://lists.openstreetmap.org/listinfo/graphhopper
>>
>
>
>
> _______________________________________________
> GraphHopper mailing list
> GraphHopper at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/graphhopper