On Sat, 2007-02-24 at 15:13, Sasha Khapyorsky wrote: > There are various performance improvements for up/down routing engine: > - updn_node object which is referenced by switch's priv pointer > - ranking for switches only > - replace time consuming cl_list by cl_qlist > - reuse already collected up/down related information (in updn_node > structure) instead of rediscovering > - eliminate many inner loops > - mask time consuming logging > - elminate using two lists with BFS > - minor cleaups > > Now up/down looks 5-6 times faster.
Nice work! > Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]> Thanks. Applied (to master only at least for right now; will get to ofed_1_2 in a bit). -- Hal _______________________________________________ openib-general mailing list openib-general@openib.org http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general