From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [PATCH 2/2] rdma/cm: allow user to specify IP to DGID mapping Date: Wed, 7 Oct 2009 15:45:59 -0600 Message-ID: <20091007214559.GU5191@obsidianresearch.com> References: <4ACAF913.3050909@voltaire.com> <20091006200739.GP5191@obsidianresearch.com> <20091006231720.GR5191@obsidianresearch.com> <3F7D26D4BA1C46F18F2F87BDD7EB7F36@amr.corp.intel.com> <20091007052639.GB18578@obsidianresearch.com> <20ADF14BE2B24B459DC3921F69449E61@amr.corp.intel.com> <20091007203257.GT5191@obsidianresearch.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: "Hefty, Sean" Cc: 'Or Gerlitz' , linux-rdma , Roland Dreier List-Id: linux-rdma@vger.kernel.org On Wed, Oct 07, 2009 at 02:20:29PM -0700, Hefty, Sean wrote: > >You are trying to make the smallest change possible to work around a > >performance problem caused by ineffecient abstractions by completely > >breaking the abstraction. > > The performance problem being addressed here is not caused by an > inefficient abstraction. It's caused by having a single, > centralized SM incapable of scaling to the cluster sizes that > customers want to use. The SM scaling is a side effect of the IP RDMA CM design. Since the IP RDMA CM design and abstraction require the ARP and PR queries - it is the root inefficiency - the SM scaling is the consequence. IB GID CM does not have this mandatory behavior, you can get your PR data from wherever you want. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html