From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [rdma-core v5 0/8] Broadcom User Space RoCE Driver Date: Tue, 25 Apr 2017 09:28:43 -0600 Message-ID: <20170425152843.GA17680@obsidianresearch.com> References: <1493059995-636-1-git-send-email-devesh.sharma@broadcom.com> <1493067443.3041.63.camel@redhat.com> <20170424215140.GA14907@obsidianresearch.com> <1828884A29C6694DAF28B7E6B8A82373AB1169E4@ORSMSX109.amr.corp.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <1828884A29C6694DAF28B7E6B8A82373AB1169E4-P5GAC/sN6hkd3b2yrw5b5LfspsVTdybXVpNB7YpNyf8@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: "Hefty, Sean" Cc: Doug Ledford , Devesh Sharma , "linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org" List-Id: linux-rdma@vger.kernel.org On Mon, Apr 24, 2017 at 11:47:05PM +0000, Hefty, Sean wrote: > > However, AFAIK, default pthread mutexes well known to be much slower > > than spinlocks for various cases. They have weird performance aspects > > from the fairness policy on unlock we may not really want. > > > > Perahps this is why Sean open-coded locks using atomics in rdmacm and > > acm? > > I went with open-coded locks based on v-tune (or whatever it's > called now -- I should really know this) performance analysis and > benchmarking data. I don't recall the performance gains, but it was > definitely measureable. But libfabric uses pthreads? 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