Thank you for these data points. If you can somehow provide inputs for
reproducing the problem (a simulation that does not finish) then I can fix
this (though maybe not in time for your deadline).

regards,
Jakob

Am Di., 24. Aug. 2021 um 04:52 Uhr schrieb Haotian Li <
haotian...@ucdconnect.ie>:

> To whom it may concern:
>
> The simulation could complete without the SSM device; The simulation could
> also finish with edgedumps; The simulation could finish using SSM device in
> the scale of 1.8(nearly 2) with 25% memory usage. However, it would shut
> down using SSM device in the scale of 3 with only 40% memory usage by that
> time. I think the hardware requirement is under mine. As a result, I would
> try to change the critiria and methodology with this specfic scenario if
> the simulation could finish and the conclusion is obvious, logic and
> coherent. Otherwise I may forgive it. Have a nice day!
>
> Best wishes,
> Haotian Li
>
>
> On Mon, 23 Aug 2021 at 17:57, Mirko Barthauer <m.bartha...@t-online.de>
> wrote:
>
>> Hi,
>>
>> I had similar issues as described by Haotian Li with the SSM device when
>> calculating SSM measures for many vehicles. I had the impression the number
>> of simultaneous SSM vehicles was more important in this case but did not do
>> more formal checks. In my case, I solved it by refining the research
>> question to only enable the device for certain vehicles (e.g. by type, by
>> route,...)
>>
>> Best regards
>> Mirko
>> Am 23.08.2021 um 11:47 schrieb Jakob Erdmann:
>>
>> Hello,
>> the SSM device is not very memory efficient and has only been used on
>> medium sized scenarios thus far.
>> However, a shutdown of the simulation might also be due to a software bug
>> (i.e. due to a combination of untested options or an untested traffic
>> situation).
>>
>> Does the simulation complete successfully without SSM  device?
>> Does the simulation complete with SSM device and default scale value? How
>> much memory does it consume in this case?
>> How big is the simulation (total number of vehicles, approximate number
>> of simultaneous vehicles)?
>>
>> regards,
>> Jakob
>>
>> Am Mo., 23. Aug. 2021 um 09:58 Uhr schrieb Haotian Li <
>> haotian...@ucdconnect.ie>:
>>
>>> To whom it may concern:
>>>
>>> I am Haotian Li, a senior student from UCD and BJUT. I'm here to ask
>>> something about the hardware requirement for sumo-gui. I am working on a
>>> summer internship using this software as a tool. When I tried to simulate
>>> the urban networks using the scale of 3(try to provide traffic jams), I
>>> used SSM devices to detect TTC and PET and record conflict loading in
>>> conflict.xml and do statistics research after simulation ended. However, I
>>> found that 16G RAM can't afford the stress of caching index files and it
>>> would break down while simulating. Then, I tried the 32G one and it would
>>> also shut down even if the memory usage is not high. Would you please give
>>> me some suggestions to fix it? Or I may forgive this scenario because the
>>> DDL is approaching. Looking forward to your reply. Have a nice day!
>>>
>>> Yours sincerely,
>>> Haotian Li
>>> _______________________________________________
>>> sumo-dev mailing list
>>> sumo-dev@eclipse.org
>>> To unsubscribe from this list, visit
>>> https://www.eclipse.org/mailman/listinfo/sumo-dev
>>>
>>
>> _______________________________________________
>> sumo-dev mailing listsumo-...@eclipse.org
>> To unsubscribe from this list, visit 
>> https://www.eclipse.org/mailman/listinfo/sumo-dev
>>
>> _______________________________________________
>> sumo-dev mailing list
>> sumo-dev@eclipse.org
>> To unsubscribe from this list, visit
>> https://www.eclipse.org/mailman/listinfo/sumo-dev
>>
> _______________________________________________
> sumo-dev mailing list
> sumo-dev@eclipse.org
> To unsubscribe from this list, visit
> https://www.eclipse.org/mailman/listinfo/sumo-dev
>
_______________________________________________
sumo-dev mailing list
sumo-dev@eclipse.org
To unsubscribe from this list, visit 
https://www.eclipse.org/mailman/listinfo/sumo-dev

Reply via email to