From owner-freebsd-questions Fri Jun 29 11:37:53 2001 Delivered-To: freebsd-questions@freebsd.org Received: from karon.dynas.se (karon.dynas.se [192.71.43.4]) by hub.freebsd.org (Postfix) with SMTP id CBE4E37B405 for ; Fri, 29 Jun 2001 11:37:47 -0700 (PDT) (envelope-from mikko@dynas.se) Received: (qmail 24686 invoked from network); 29 Jun 2001 18:37:45 -0000 Received: from spirit.sto.dynas.se (HELO spirit.dynas.se) (172.16.1.10) by 172.16.1.1 with SMTP; 29 Jun 2001 18:37:45 -0000 Received: (qmail 29601 invoked from network); 29 Jun 2001 18:37:44 -0000 Received: from explorer.rsa.com (10.81.217.59) by spirit.dynas.se with SMTP; 29 Jun 2001 18:37:44 -0000 Received: (from mikko@localhost) by explorer.rsa.com (8.11.3/8.11.3) id f5TIbf714026; Fri, 29 Jun 2001 11:37:41 -0700 (PDT) (envelope-from mikko) Date: Fri, 29 Jun 2001 11:37:41 -0700 (PDT) From: Mikko Tyolajarvi Message-Id: <200106291837.f5TIbf714026@explorer.rsa.com> To: dleimbac@MPI-Softtech.Com Cc: questions@freebsd.org Subject: Re: Kernel options for faster loopback? Newsgroups: local.freebsd.questions References: <20010629114724.A1074@mpi-plusplus.mpi-softtech.com> X-Newsreader: NN version 6.5.6 (NOV) Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG In local.freebsd.questions you write: >Anyone know of anyway for me to speed up loopback performance on FreeBSD? >My times are awful for ping 127.0.0.1 in comparison to basically every other >OS I have tried. Are you sure you're not falling into the "meaningless microbenchmark" trap? Unless you are actually planing to write an application that sends ICMP echo requests through the loopback interface, may I suggest that you test what you are going to do instead. Ping times are not necessarily the same as loopback "performance," if for example you are really interested in throughput, or roundtrip times for some client-server app (c.f. X server). I get times around 35 microseconds, much of which may well be system call overhead (a couple of gettimeofday(), select(), sendmsg(), recvmsg()). >Is there some IPV6 overhead I can reduce? If you build your kernel without "options INET6" all IPv6 code should be #ifdeffed out. $.02, /Mikko P.S. I tried ping on AIX, HP-UX and Solaris, all of which excluded any decimals on the number of milliseconds, thus ending up as "0". Very fast indeed :-) -- Mikko Työläjärvi_______________________________________mikko@rsasecurity.com RSA Security To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message