Dear Jörg,
there are several points to this:
- how large is the network where re-computation takes 60s? The network from
the log in your mail appears to take less time
- as a comparison I'm looking at a network with 14k nodes and 31k edges
which takes about 3s for re-computation
- are you using the debug-version of netedit by chance?
- if the reported times for the individual steps do not add up to the total
compute-time there might be another step that's eating all the time and I'd
like to figure out which one it is. (maybe you can just send me the network
that is taking so long)
- some tools require an updated network computation (connect and tls mode)
- this computation is only triggered if the network was changed by
deleting or adding edges
- there is room for optimization here since only the area that was
changed would actually have to be recomputed
- possibly 0.28.0 has added conditions for triggering a full recomputed.
This could be a bug.
- between witch tools are you switching
- what changes did you do in between?
best regards,
Jakob
2016-11-11 16:10 GMT+01:00 Joerg Schweizer <[email protected]>:
> Dear all,
> I've been experiencing long delays (up to a minute) when changing tools
> in netedit v0.28. Pressing further buttons during delay results often in
> a crash. Apparently junctions are being recalculated when switching
> tools. I receive long lists of warning messages, here is the end of it:
>
> ************************************
> :
> :
> :
> Warning: Shape for junction '710705007' has distance 22.35 to its given
> position
> Warning: Shape for junction '820929012' has distance 23.51 to its given
> position
> Warning: Shape for junction '86673123' has distance 25.68 to its given
> position
> Warning: Shape for junction '938385352' has distance 25.08 to its given
> position
> Warning: Shape for junction 'cluster_257082791_257082936_257083205' has
> distance 34.57 to its given position
> done (1986ms).
> Computing edge shapes... done (606ms).
> Computing node types... done (1ms).
> Computing priorities... done (48ms).
> Computing approached edges... done (4ms).
> Computing approaching lanes... done (4ms).
> Dividing of lanes on approached lanes... done (108ms).
> Processing turnarounds... done (0ms).
> Rechecking of lane endings... done (15ms).
> Assigning nodes to traffic lights... done (0ms).
> Computing traffic light control information... done (13ms).
> Computing node logics... done (421ms).
> Computing traffic light logics... done (3ms).
> 506 traffic light(s) computed.
> Building inner edges...
> Warning: Connection '156710467#52_0->156710474#4_0' is only 0.07 short.
> Warning: Connection '23275345#13_1->23210836#0_1' is only 0.02 short.
> Warning: Connection '81447274_0->-81447326_0' is only 0.00 short.
> Warning: Connection '40504395#6_2->-82789377_1' is only 0.02 short.
> Warning: Connection '134140082#0_0->134140082#1_0' is only 0.00 short.
> Warning: Connection '36948586_0->323715234_0' is only 0.05 short.
> done (805ms).
> -----------------------------------------------------
> Summary:
> Node type statistics:
> Unregulated junctions : 845
> Dead-end junctions : 1983
> Priority junctions : 10419
> Right-before-left junctions : 1793
> Network boundaries:
> Original boundary : 0.00,-1.58,14107.50,11394.12
> Applied offset : 0.00,0.00
> Converted boundary : 0.00,-1.58,14107.50,11394.12
> -----------------------------------------------------
>
> Finished computing junctions.
> ************************************
>
> Actually the new netedit version is much more stable, but these delays
> are very time consuming when editing larger networks. Is there an option
> to avoid this?
>
> Thanks,
> Joerg
>
>
> --
> Joerg Schweizer
> University of Bologna
> DICAM - Transportation Group
> Viale Risorgimento, 2, I-40136 Bologna, Italy
> Tel: +39 051 209 3338, Fax: +39 051 209 3337
>
> ------------------------------------------------------------
> ------------------
> Developer Access Program for Intel Xeon Phi Processors
> Access to Intel Xeon Phi processor-based developer platforms.
> With one year of Intel Parallel Studio XE.
> Training and support from Colfax.
> Order your platform today. http://sdm.link/xeonphi
> _______________________________________________
> sumo-user mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/sumo-user
>
------------------------------------------------------------------------------
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today. http://sdm.link/xeonphi
_______________________________________________
sumo-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/sumo-user
