Date: Wed, 18 Feb 2004 09:12:34 -0800 (PST) From: Marian Durkovic <md@bts.sk> To: freebsd-gnats-submit@FreeBSD.org Subject: i386/63027: bad TX performance of xl driver for 3COM 905C cards Message-ID: <200402181712.i1IHCYCf044718@www.freebsd.org> Resent-Message-ID: <200402181720.i1IHKBAD083018@freefall.freebsd.org>
next in thread | raw e-mail | index | archive | help
>Number: 63027
>Category: i386
>Synopsis: bad TX performance of xl driver for 3COM 905C cards
>Confidential: no
>Severity: serious
>Priority: high
>Responsible: freebsd-i386
>State: open
>Quarter:
>Keywords:
>Date-Required:
>Class: sw-bug
>Submitter-Id: current-users
>Arrival-Date: Wed Feb 18 09:20:10 PST 2004
>Closed-Date:
>Last-Modified:
>Originator: Marian Durkovic
>Release: 4.9
>Organization:
Slovak Technical University
>Environment:
>Description:
The xl driver is not able to achieve wire-speed performance for TX direction. The nttcp test shows significant performance degradadion for TX direction as compared to RX direction which is able to run at theoretical limit without problems. Looks like the heavy use of TX path actually blocks the RX direction, and therefore TCP ACKs are being delivered with significant delay (like 24 msec) - verified by tcpdump. Here are the example values from nttcp:
box with Intel E100 PRO -> box with 3c905
Bytes Real s CPU s Real-MBit/s CPU-MBit/s Calls Real-C/s CPU-C/s
l409600000 34.80 2.96 94.1489 1105.1982 100000 2873.20 33728.0
1409600000 34.82 1.18 94.1175 2773.2828 281323 8080.27 238094.6
box with 3c905 -> box with Intel E100 Pro
Bytes Real s CPU s Real-MBit/s CPU-MBit/s Calls Real-C/s CPU-C/s
l409600000 38.86 1.18 84.3323 2766.3223 282788 7277.88 238733.7
1409600000 38.85 1.56 84.3524 2100.2557 100000 2574.23 64094.7
>How-To-Repeat:
Run nttcp
>Fix:
>Release-Note:
>Audit-Trail:
>Unformatted:
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200402181712.i1IHCYCf044718>