Cost modification

[RomainClem]

Hello mister Kelvin,

I think this small change should be enough to implement what we mentionned in our discussion.

Unfortunately, I can't test loacally. Could you please pull my branch and see if it breaks the code or make the output worse.

God bless you, Wawawa

[notyiming]

I am not sure if the new distance cost parameter helps, it runs slightly slower now. My guess is because of the additional get_distance operation being run for every direction of every exploration iteration. The results seem to be roughly the same, I tested using assessment/run_all twice, and found similar result. Top one is cost-modification branch

[RomainClem]

Intersport

On Wed, Aug 2, 2023, 17:51 notyiming @.***> wrote:

I am not sure if the new distance cost parameter helps, it runs slightly slower now. My guess is because of the additional get_distance operation being run for every direction of every exploration iteration. The results seem to be roughly the same, I tested using assessment/run_all twice, and found similar result. Bottom is cost-modification branch [image: image] https://user-images.githubusercontent.com/55159184/257884729-5eb00e3a-f3af-4d77-aac8-196cf77affdc.png

— Reply to this email directly, view it on GitHub https://github.com/notyiming/global-router/pull/32#issuecomment-1662579927, or unsubscribe https://github.com/notifications/unsubscribe-auth/ANMWKB2HPBIR7V3N2ZRAHZLXTKAQPANCNFSM6AAAAAA3AELPOE . You are receiving this because you were assigned.Message ID: @.***>

[RomainClem]

Test with latest commit:

@notyiming Can you test on your end because it might not be the same.

[RomainClem]

It's shittier than the original version, but we're learning:

[RomainClem]

This run by putting the cost as (congestion, wirelenght, distance) gave a pretty good overflow but the runtime is long.

[RXCL]

Sir @RomainClem this looks like shit.