Re: [sumo-user] About the state of the simulation when multiple parking

[Jakob Erdmann <namdre.sumo@xxxxxxxxx>]

- if for some reason the number of traci calls is increased, this can have an impact on running speed. You could test this be using libsumo which removes this communication overhead (https://sumo.dlr.de/docs/Libsumo.html)

- it would be helpful if you could figure out whether the additional running time is spent in sumo or in your own script. Currently, the only way to do this is to log the duration of the traci.simulationStep calls from within your script

regards,

Jakob

Am Mo., 12. Apr. 2021 um 11:40 Uhr schrieb g2121044 <g2121044@xxxxxxxxx>:

Thank you for your reply.

- Are you using TraCI to control the vehicle?

Yes, I am controlling the vehicle with TraCI.

- Are you using a cab device? If so, how are you doing it?

No, I am not using a cab.

- Are you using Rerouter with parkingAreaReroute?

No, I am not using rerouter.

The simple flow of the simulation is as follows.

1. Move each vehicle to the parking area as soon as the simulation starts. 2. (This process is only in rou.xml)

2. When a customer request is received, the dispatching algorithm and the duarouter determine the route and vehicle. 

3. The route is changed to the target vehicle using traci.vehicle.setRoute.

4. When the vehicle arrives at the boarding and alighting point, if all the assigned requests have been fulfilled, the vehicle will change target to the parking area.

From the state 1 of this flow, the process is slowing down.

And when I compared the speed of the simulation by changing only the number of parking areas, the difference in speed was noticeable. (All other conditions were the same except for the parking area.)

Therefore, I thought that the parking area was the cause.

However, when I found out that it was not the cause, the problem became even more difficult.

regard,

Kazuma Nishikawa

----- Original Message -----
From: Jakob Erdmann <namdre.sumo@xxxxxxxxx>
To: Sumo project User discussions <sumo-user@xxxxxxxxxxx>
Date: 2021-04-12 17:59:32
Subject: Re: [sumo-user] About the state of the simulation when multiple parking areas are created

Hello,

ParkingAreas by themselves should have no impact on simulation speed.

- are you controlling the vehicles via TraCI?

- are you using the taxi device? If so, how?

- are you using rerouters with parkingAreaReroute?

regards,

Jakob

Am Mo., 12. Apr. 2021 um 10:51 Uhr schrieb g2121044 <g2121044@xxxxxxxxx>:

Hello Everyone

I have a question about the state of the simulation when multiple parking areas are created.

I am a graduate student and I am using SUMO for my research on DARP (dial a ride problem).
I set up four parking areas on the four sides of a square road network, and simulated DAR on the road network.
The number of vehicles is set to 8. Each of the four parking areas has two vehicles waiting for a customer's request.

As a result of the simulation, the processing speed of the simulation was clearly reduced compared to the case with one parking area. The speed of the simulation was probably reduced by more than 90%.

Does the simulation process become slower when multiple parking areas are created?

I have tried to use multi-threading (--device.router.thread), but it does not work.

Best regards,
Kazuma Nishikawa
_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user

_______________________________________________
sumo-user mailing list
sumo-user@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/sumo-user