Date: Wed, 16 Mar 2011 09:41:49 -0700 From: YongHyeon PYUN <pyunyh@gmail.com> To: Vijay Singh <vijju.singh@gmail.com> Cc: freebsd-net@freebsd.org Subject: Re: Jumbo frame support for BGE_ASICREV_BCM5714 Message-ID: <20110316164149.GA9004@michelle.cdnetworks.com> In-Reply-To: <20110315233844.GE5076@michelle.cdnetworks.com> References: <AANLkTimqca%2B72jvWi89LnrP0F4MNdvK7YaG2pTavNiC3@mail.gmail.com> <20110315024325.GG1577@michelle.cdnetworks.com> <AANLkTi=brPL3Krre%2BvYe-P4jvPu3=rQ%2Be-EHjTB%2Bbb=4@mail.gmail.com> <20110315233844.GE5076@michelle.cdnetworks.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Mar 15, 2011 at 04:38:44PM -0700, YongHyeon PYUN wrote: > On Mon, Mar 14, 2011 at 09:40:36PM -0700, Vijay Singh wrote: > > > As you know, BCM5714, BCM5715 and BCM5780 use unique jumbo frame > > > scheme that is not compatible with other controllers. All other > > > Broadcom controllers have better jumbo frame scheme. These > > > controllers have one send ring, one standard receive producer ring > > > and one receive return ring. In order to receive jumbo frames on > > > these controllers you have to increase Rx buffer size to hold 9k > > > sized jumbo frame. Two Rx modes(standard Rx BDs and extended Rx > > > BDs) are supported for these controllers. Using extended Rx BDs on > > > BCM5714/BCM5715/BCM5780 reduces the number of Rx BDs to 256 entries > > > which shall reduce the performance. I would use standard Rx BDs to > > > hold 512 entries for RX buffers. > > > > > > I think I received jumbo frame support request for these > > > controllers in past. At that time I had no interests on > > > implementing it due to severe implementation differences. What is > > > your main reason to use jumbo frame on this controller? What is > > > your expectation on performance numbers? I guess no other OSes > > > support jumbo frame on this controller. > > > > Hi Pyun, I am stuck with this NIC due to it being the one present in > > the HW platform that I have to support. The performance is expectation > > is mainly for apps that use large payloads (where something TSO would > > have helped). > > > > Here is experimental patch which I tried not to penalize other > controllers. I don't have BCM5714 controllers so I don't know > whether it works or not. The patch was generated against HEAD and > it would be cleanly applied to 8.2R/7.4R. Due to large bge(4) > changes I guess it wouldn't be applied to 7.2R. But I guess you can > install 8.2R/7.4R to one of your box and experiment this patch for > a while then you can backport this to 7.2R. > You can find the patch at the following URL. > http://people.freebsd.org/~yongari/bge/bge.5714.jumbo.diff > There was a bug in the diff. I updated the diff but URL is the same as before. If you have downloaded the file, please try again. > > -vijay
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110316164149.GA9004>