Date: Mon, 19 Oct 2009 23:38:58 -0700 (PDT) From: Paul Pathiakis <pathiaki2@yahoo.com> To: Adrian Chadd <adrian@freebsd.org>, =?iso-8859-1?Q?Istv=E1n?= <leccine@gmail.com> Cc: Hongtao Yin <htyin@huawei.com>, freebsd-performance@freebsd.org, Brent Jones <brent@servuhome.net> Subject: Re: Comparison of FreeBSD/Linux TCP Throughput performance Message-ID: <687096.90163.qm@web110506.mail.gq1.yahoo.com> In-Reply-To: <d763ac660910191939y2416535ck6e2c351a3f329677@mail.gmail.com> References: <78DB4AE8EF5F4A1EBD3992D7404B2725@china.huawei.com> <d763ac660910180755i7f6fd3c7q8578bfed11978b9d@mail.gmail.com> <ee9f3b480910181305x5c8661a7mb7bfdd2ddd0a267d@mail.gmail.com> <d763ac660910181836p45aedc37v1c77f5e96b6df13b@mail.gmail.com> <b8592ed80910191716v11b978c1i8bf82170e4ed6a37@mail.gmail.com> <d763ac660910191939y2416535ck6e2c351a3f329677@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,=0A=0Agoing to chime in on this one....just trying to help.=0A=0AThere's= some simple things to get Gb, jumbo frames (MTU > 1500 on both the switch = port and the card) is a simple way.=0A=0AHowever, I'd have to read back on = this thread as I haven't had time of late. Basically, and I've seen this o= n many, many Gb cards, chipsets and Drives make the world of difference. = =0A=0AI tried for a few days to try and get an HP DL360 with it's dual on-b= oard Broadcom bge NIC to get to 1 Gb.... just plain no way. If anyone has = settings for that, I'd like to know them. Also, this is the same chip set = that a lot of vendors use and it is cheap and inexpensive. When I couldn't= get the thing to go beyond 720Mb, I tried something simple. I ordered an = Intel dual Gb port card and put that in. WITHOUT tuning, this thing starte= d at almost 800 Mb throughput and I almost got it to 850 Mb within a few ho= urs.=0A=0AI wish I could send those settings to this list but it was well o= ver a year ago that I did this.=0A=0ASadly, most large vendors start with B= roadcom chipsets and don't want to spent the extra $10 for the Intel chipse= t. (No, I am not a fan boy of Intel, more of AMD if anything, but their NI= Cs rock.)=0A=0AP.=0A=0A=0A=0A=0A________________________________=0AFrom: Ad= rian Chadd <adrian@freebsd.org>=0ATo: Istv=E1n <leccine@gmail.com>=0ACc: Ho= ngtao Yin <htyin@huawei.com>; freebsd-performance@freebsd.org; Brent Jones = <brent@servuhome.net>=0ASent: Mon, October 19, 2009 10:39:53 PM=0ASubject: = Re: Comparison of FreeBSD/Linux TCP Throughput performance=0A=0A2009/10/20 = Istv=E1n <leccine@gmail.com>:=0A=0A=0A> therefore i like netpipe runs you c= an see the performance and the latency as=0A> well using the packet size as= your "x" axis, i think it makes more sense=0A> then just 1 number=0A=0AMy = point was to demonstrate that saturating gigabit ethernet is very=0Adoable = with FreeBSD, and his limitation is more likely somewhere other=0Athan "TCP= ".=0A=0AI've told him privately to check CPU utilisation. I'll do the same = on=0Amy boxes when I get some time; I'd like to know why I'm only seeing ~= =0A800mbit with large buffers.=0A__________________________________________= _____=0Afreebsd-performance@freebsd.org mailing list=0Ahttp://lists.freebsd= .org/mailman/listinfo/freebsd-performance=0ATo unsubscribe, send any mail t= o "freebsd-performance-unsubscribe@freebsd.org"=0A=0A=0A=0A
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?687096.90163.qm>