[Networkit] current benchmark results (4.0 Dev)

Christian Staudt christian.staudt at kit.edu
Mon Oct 5 15:32:51 CEST 2015


Hi Matteo,
you might be interested in the full benchmark, see the attached file.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: brandNewBench2.zip
Type: application/zip
Size: 431806 bytes
Desc: not available
URL: <https://lists.ira.uni-karlsruhe.de/mailman/private/networkit/attachments/20151005/33477f2c/attachment-0001.zip>
-------------- next part --------------



Edges/second is simply an intuitive way to measure algorithm speed by relating network size and running time. It does not literally mean that the processor scans edges at that speed, though for most algorithms this is roughly what is happening.

Network structure definitely affects running time in nontrivial ways.

I agree that the betweenness approximation does not really belong in there, because the amount of work done depends on the number of samples, so you can produce all the speeds between a BFS and exact betweenness.

Best,
Chris





On 05 Oct 2015, at 14:57, Matteo Riondato <matteo at cs.brown.edu> wrote:

> 
>> On Oct 5, 2015, at 3:09 AM, Christian Staudt <christian.staudt at kit.edu> wrote:
>> 
>> <epsSummary.pdf>
> 
> Sorry for the ``theoretician'? comment =), but what do these result mean?
> 
> I?m not sure I get the concept of edges per second, especially in terms of an approximation algorithm that may not touch all edges in the network, and in general any algorithm runtime may depend on other properties than the number of edges.
> Can you please elaborate on why this is the right measure, and how is it computed?
> Are you running the algorithms on networks of different sizes (no. of edges) and just computing the running time? How do other structural parameters of the networks affect the runtime?
> 
> Thanks!
> Matteo
> _______________________________________________
> NetworKit mailing list
> NetworKit at ira.uni-karlsruhe.de
> https://lists.ira.uni-karlsruhe.de/mailman/listinfo/networkit

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://lists.ira.uni-karlsruhe.de/mailman/private/networkit/attachments/20151005/33477f2c/attachment-0001.sig>


More information about the NetworKit mailing list