From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-2.3 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_1 autolearn=no autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id E64B7CA9EA0 for ; Tue, 22 Oct 2019 12:08:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id C148C21872 for ; Tue, 22 Oct 2019 12:08:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S2388954AbfJVMId (ORCPT ); Tue, 22 Oct 2019 08:08:33 -0400 Received: from mga11.intel.com ([192.55.52.93]:16238 "EHLO mga11.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S2388106AbfJVMId (ORCPT ); Tue, 22 Oct 2019 08:08:33 -0400 X-Amp-Result: UNKNOWN X-Amp-Original-Verdict: FILE UNKNOWN X-Amp-File-Uploaded: False Received: from fmsmga006.fm.intel.com ([10.253.24.20]) by fmsmga102.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 22 Oct 2019 05:08:32 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.67,327,1566889200"; d="scan'208";a="398994993" Received: from jerryopenix.sh.intel.com (HELO jerryopenix) ([10.239.158.171]) by fmsmga006.fm.intel.com with ESMTP; 22 Oct 2019 05:08:31 -0700 Date: Tue, 22 Oct 2019 20:05:56 +0800 From: "Liu, Changcheng" To: Doug Ledford Cc: linux-rdma@vger.kernel.org Subject: Re: rdma performance verification Message-ID: <20191022120556.GA830170@jerryopenix> References: <20190916094234.GA11772@jerryopenix> <4f01d52616a4a8f767b95bda7fd68e62d8c1f8ae.camel@redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <4f01d52616a4a8f767b95bda7fd68e62d8c1f8ae.camel@redhat.com> User-Agent: Mutt/1.9.4 (2018-02-28) Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org On 16:30 Mon 21 Oct, Doug Ledford wrote: > On Mon, 2019-09-16 at 17:42 +0800, Liu, Changcheng wrote: > > Hi all, > > I'm working on using rdma to improve message transaction > qperf is nice because it will do both the tcp and rdma testing, so the > same set of options will make it behave the same way under both tests. @Doug Ledford: I'll check how to use it to compare RDMA & TCP. > > I think you are mis-reading the instructions on ib_send_bw. First of > all, IB RC queue pairs are, when used in send/recv mode, message passing > devices, not a stream device. When you specified the -s parameter of @Doug Ledford: What's the difference between "message passing device" and "stream device"? > 1GB, you were telling it to use messages of 1GB in size, not to pass a > total of 1GB of messages. And the default number of messages to pass is > 1,000 iterations (the -n or --iters options), so you were actually @Doug Ledford: Thanks for your information. It helps me a lot. > testing a 1,000GB transfer. You would be better off to use a smaller > message size and then set the iters to the proper value. This is what I > got with 1000 iters and 1GB message size: > > #bytes #iterations BW peak[MB/sec] BW average[MB/sec] MsgRate[Mpps] > 1073741824 1000 6159.64 6159.46 0.000006 > --------------------------------------------------------------------------------------- > > real 3m3.101s > user 3m2.430s > sys 0m0.450s > > I tried dropping it to 1 iteration to make a comparison, but that's not > even allowed by ib_send_bw, it wants a minimum of 5 iterations. So I > did 8 iterations at 1/8th GB in size and this is what I got: > > #bytes #iterations BW peak[MB/sec] BW average[MB/sec] MsgRate[Mpps] > 134217728 8 6157.54 6157.54 0.000048 > --------------------------------------------------------------------------------------- > > real 0m2.506s > user 0m2.411s > sys 0m0.059s > > When I adjust that down to 1MB and 1024 iters, I get: > > #bytes #iterations BW peak[MB/sec] BW average[MB/sec] MsgRate[Mpps] > 1048576 1024 6157.74 6157.74 0.006158 > --------------------------------------------------------------------------------------- > > real 0m0.427s > user 0m0.408s > sys 0m0.002s > > The large difference in time between these last two tests, given that > the measured bandwidth is so close to identical, explains the problem > you are seeing below. > > The ib_send_bw test is a simple test. It sets up a buffer by > registering its memory, then just slams that buffer over the wire. With > a 128MB buffer, you pay a heavily memory registration penalty. That's > factored into the 2s time difference between the two runs. When you use > a 1GB buffer, the delay is noticeable to the human eye. There is a very > visible pause as the server and client start their memory registrations. @Doug Ledford: Do you mean that every RDMA-SGE(Scatter/Gather element) will use seperate MR(Memory Region)? If all the RDMA-SGE use only one pre-allocated MR-1GB, the two tests shouldn't have so much time consuming difference. > > > In Ceph, the result shows that rdma performance (RC transaction > > type, > > SEDN operation) is worse or not much better than TCP implemented > > performance. > > Test A: > > 1 client thread send 20GB data to 1 server thread (marked as > > 1C:1S) > > Result: > > 1) implementation based on RDMA > > Take 171.921294s to finish send 20GB data. > > 2) implementation based on TCP > > Take 62.444163s to finish send 20GB data. > > > > Test B: > > 16 client threads send 16x20GB data to 1 server thread (marked > > as 16C:1S) > > Result: > > 1) implementation base on RDMA > > Take 261.285612s to finish send 16x20GB data. > > 2) implementation based on TCP > > Take 318.949126 to finish send 16x20GB data. > > I suspect your performance problems here are memory registrations. As > noted by Chuck Lever in some of his recent postings, memory > registrations can end up killing performance for small messages, and the > tests I've shown here indicate, they're also a killer for huge memory > blocks if they are repeatedly registered/deregistered. TCP has no I think we could pre-registered 1GB MR and then all the SGE share with the same MR, then it could mitigate the penalty in register/deregister. > memory registration overhead, so in the single client case, it is > outperforming the RDMA case. But in the parallel case with lots of > clients, the memory registration overhead is spread out among many > clients, so we are able to perform better overall. In Ceph implementation, all the threads in the same process share with the same pre-registered 1GB MR. The MR is divided into lots of chunks to be used as SGE. In this way, how to explain the test result between Test-A & Test-B? > > In a nutshell, it sounds like the Ceph transfer engine over RDMA is not > optimized at all, and is hitting problems with memory registration > overhead. Ceph/RDMA seems not widely used and some implementation need to be optimized. I'm going to work on it in future. > > -- > Doug Ledford > GPG KeyID: B826A3330E572FDD > Fingerprint = AE6B 1BDA 122B 23B4 265B 1274 B826 A333 0E57 2FDD