> You can also try this for even better results > > > mpiexec -n 992 Ray -test-network-only \ > -route-messages -connection-type kautz \ > -o Kautz-992-degree-31-diameter-2 > > > Details: > > 992.routing:[GraphImplementationKautz::makeConnections] degree= 31 > diameter= 2 base= 32 vertices= 992 > > > You will get a lower latency and a very small standard deviation. > > There will be 32 unused cores, but no unused node because ranks are > mapped to cores in a round-robin fashion.
It appears when k >= 10 the test fails for kautz at any value of d. Reproduce: Diameter 2 graph, sebhtml-ray-82b4a0e/unit-tests$ ./graph_main 90 kautz 1 1 [GraphImplementationKautz::makeConnections] degree= 9 diameter= 2 base= 10 vertices= 90 Done computing connections 0/90 90/90 sebhtml-ray-82b4a0e/unit-tests$ ./graph_main 110 kautz 1 1 Error: cannot create a Kautz graph with 110 vertices Warning: using a complete graph because type kautz can not used with 110 vertices 0/110 100/110 110/110 Diameter 3 graph, sebhtml-ray-82b4a0e/unit-tests$ ./graph_main 810 kautz 1 1 [GraphImplementationKautz::makeConnections] degree= 9 diameter= 3 base= 10 vertices= 810 Done computing connections 0/810 100/810 200/810 300/810 400/810 500/810 600/810 700/810 800/810 810/810 sebhtml-ray-82b4a0e/unit-tests$ ./graph_main 1100 kautz 1 1 Error: cannot create a Kautz graph with 1100 vertices Warning: using a complete graph because type kautz can not used with 1100 vertices 0/1100 100/1100 200/1100 300/1100 400/1100 500/1100 600/1100 700/1100 800/1100 900/1100 1000/1100 1100/1100 >> With the formula for kautz, n=(k+1)*k^(d-1), what values does one use >> for k and d? > I've found 768 is a good number as it works for kautz and is divisible by 16, i.e. will fit evenly on 48 nodes. Also 1536 works well. Thanks, Ben ------------------------------------------------------------------------------ Virtualization & Cloud Management Using Capacity Planning Cloud computing makes use of virtualization - but cloud computing also focuses on allowing computing to be delivered as a service. http://www.accelacomm.com/jaw/sfnl/114/51521223/ _______________________________________________ Denovoassembler-users mailing list Denovoassembler-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/denovoassembler-users
