From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 26054 invoked by alias); 10 Jul 2008 18:44:44 -0000 Received: (qmail 26036 invoked by uid 22791); 10 Jul 2008 18:44:43 -0000 X-Spam-Check-By: sourceware.org Received: from aussmtpmrkpc120.us.dell.com (HELO aussmtpmrkpc120.us.dell.com) (143.166.82.159) by sourceware.org (qpsmtpd/0.31) with ESMTP; Thu, 10 Jul 2008 18:44:23 +0000 X-IronPort-AV: E=Sophos;i="4.30,338,1212382800"; d="scan'208";a="352700173" Received: from unknown (HELO M31.equallogic.com) ([12.110.134.31]) by aussmtpmrkpc120.us.dell.com with SMTP; 10 Jul 2008 13:44:22 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Message-ID: <18550.22660.364998.466763@gargle.gargle.HOWL> Date: Thu, 10 Jul 2008 18:44:00 -0000 From: Paul Koning To: pedro@codesourcery.com Cc: gdb@sourceware.org, gdb-patches@sourceware.org Subject: Re: [remote protocol] support for disabling packet acknowledgement References: <200807101823.13728.pedro@codesourcery.com> X-Mailer: VM 7.17 under 21.4 (patch 19) "Constant Variable" XEmacs Lucid Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2008-07/txt/msg00154.txt.bz2 >>>>> "Pedro" == Pedro Alves writes: Pedro> Hi, This patch adds support to the remote protocol to disable Pedro> packet acknowlegment. This is useful when the transport being Pedro> used is itself reliable, e.g., TCP/IP. In these cases, by Pedro> removing the acking we reduce the amount of roundtrips, and Pedro> decrease the communication latency. I'm not sure this is a good idea. For one thing, if you want to work on performance, there are much more dramatic changes to the protocol that could be done that would help much more. I can't believe that the cost of acks is significant compared to all the other bottlenecks. Also, TCP is reliable delivery at the transport layer. It doesn't do reliable delivery at the application layer -- that's what the gdb remote protocol ACKs do. The fact that TCP delivered a packet to the stub doesn't mean the stub acted on it. paul