[Click] Nsclick optimalization in todevice
Wim Vandenberghe
wim.vandenberghe at intec.ugent.be
Fri Jan 12 07:35:46 EST 2007
Hi,
because my nsclick simulations took a long time to complete, I analysed
them using a profiler, and i discovered that the main problem was that
the run_task of the tosimdevice whas called a lot, and that it tried to
pull upstream, even if there were no elements available in the queues
before it.
So i added a notifier to the tosimdevice (like in a unqueue), so that
the task will only be scheduled if there is a packet available upstream.
First results where very promising, simulations were performed five
times faster then before. I have included the changed tosimdevice
sources in attachment, maybe some more people could test this code and
verify if this speeds up their simulations, and if the simulation
results are still the same.
Kind regards,
Wim
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tosimdevice.cc
Type: text/x-c++src
Size: 5153 bytes
Desc: not available
Url : https://amsterdam.lcs.mit.edu/pipermail/click/attachments/20070112/2a0a6dca/tosimdevice.cc
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: tosimdevice.hh
Url: https://amsterdam.lcs.mit.edu/pipermail/click/attachments/20070112/2a0a6dca/tosimdevice.diff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: wim.vandenberghe.vcf
Type: text/x-vcard
Size: 303 bytes
Desc: not available
Url : https://amsterdam.lcs.mit.edu/pipermail/click/attachments/20070112/2a0a6dca/wim.vandenberghe.vcf
More information about the click
mailing list