Date: Mon, 20 Apr 1998 23:46:36 +0900 From: Kenjiro Cho <kjc@csl.sony.co.jp> To: Scott Michel <scottm@cs.ucla.edu> Cc: current@FreeBSD.ORG Subject: Re: Networking strategy for -current Message-ID: <199804201446.XAA01605@hotaka.csl.sony.co.jp> In-Reply-To: Your message of "Sun, 19 Apr 1998 16:37:42 MST." <199804192337.QAA05685@mordred.cs.ucla.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Scott Michel said: >> Kenjiro's work is spectacular for low speed and medium speed >> links (<= 10Mbs). For 100Mbs links, well, ... let's just say >> it appears to have a lot of overhead. >> I'm not sure whether it's Matt Thomas' de driver which is the >> issue (certainly, the driver code needs some simplification) >> or if ALTQ requires the send interrupt to be triggered so that >> he knows a packet has drained. It's not the overhead of CBQ. o CBQ needs a fine-grained kernel timer (say 1KHz) for 100baseT. It's the design of the original CBQ. (a simple math proves it, MTU of 100baseT is too small for 100Mbps) o There is a problem in the de driver patch of altq-0.4.3. (de driver is too clever and too complex :-) Some improvement is added in altq-1.0.1. If you still have problems with altq-1.0.1 with a fine-grained timer, let me know. --Kenjiro To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199804201446.XAA01605>