PF scheduling scheme takes a very long time during simulation

[noorsification]

Hi, i'm still running 'mmwave-epc-tdma' Why the PF scheduling scheme takes a very long time during simulation ? even though I made the simtime 1s and PC with 12 GB RAM. if I run 20-50 ue it will take a whole day. The other scheduler doesn't take that long.

Is this an algorithm problem?

Thankyou Best, ajinoors

[mychele]

Hi, we are looking into this, we'll let you know. Michele

Il giorno mar 30 ott 2018 alle ore 07:44 ajinoors [email protected] ha scritto:

Hi, i'm still running 'mmwave-epc-tdma' Why the PF scheduling scheme takes a very long time during simulation ? even though I made the simtime 1s and PC with 12 GB RAM. if I run 20-50 ue it will take a whole day. The other scheduler doesn't take that long.

Is this an algorithm problem?

Thankyou Best, ajinoors

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/nyuwireless-unipd/ns3-mmwave/issues/68, or mute the thread https://github.com/notifications/unsubscribe-auth/AG1b5TZv2-UfQ-ojJWI9DrULPcxNvyXXks5up_VagaJpZM4YBR9N .

-- Michele Polese

Ph.D. Student Dept. of Information Engineering (DEI) - SIGNET Research Group University of Padova Via Gradenigo 6/B, 35131 Padova (Italy)

email: [email protected] web: polese.io - mmwave.dei.unipd.it

[noorsification]

hi, Thank you. I'll wait for this. Best, ajinoors

[mychele]

Hi, we fixed a bug in the PF scheduler in 56849c8d578888dafeee3e419bea9ddb8331cb45 Can you check if the issue still exists? Thanks, Michele

[noorsification]

Hi, I have to add your fix script on "mmwave-flex-tti-pf-mac-scheduler.cc" and run "mmwave-epc-tdma" with PF scheduler. But the simulation still takes long time like before. if I may know, what is the function of the added script?

Thanks, Aji Sanyoto

[mychele]

Hi, we are looking into the issue related to the difference in execution time, we will let you know. The fix we introduced is related to the fact that the MAC header was not accounted for in the scheduling decision (see commit https://github.com/nyuwireless-unipd/ns3-mmwave/commit/56849c8d578888dafeee3e419bea9ddb8331cb45).

[noorsification]

Hi, is there still no improvement in this issue?

Thanks, Aji sanyoto

[tommasozugno]

Hi,

I'm sorry for the delay. I did some tests and I've seen that lot of time is lost to sort the vector m_ueStatHeap https://github.com/nyuwireless-unipd/ns3-mmwave/blob/7c25c1b91d005374746c1529a3bed5beb849a455/src/mmwave/model/mmwave-flex-tti-pf-mac-scheduler.cc#L1135 I noticed that its size is constantly increasing because it is never emptied and this may be the cause of this issue. Unfortunately I did not have time to try to fix it, hopefully I will in the next days.

Best, Tommaso

[noorsification]

Does this bug will affect the results such as the delay on DlPdcpStats?

[noorsification]

I also run the MR scheduler with the same PC and it takes a very long time too

[tommasozugno]

Does this bug will affect the results such as the delay on DlPdcpStats?

No, it should affect the performance only.

[tommasozugno]

I also run the MR scheduler with the same PC and it takes a very long time too

We will check also this one, thanks :)

[lamboy24]

Hi @tommasozugno and mas @ajinoors sorry in advance, are the issue found already?

we were trying to run, and the executing PF and MR took very long time.

thanks in advance.

Victor Sinaga

[tommasozugno]

Hi Victor, No, the issue is still there.