Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 22 Jul 2011 15:46:06 -0400
From:      Mike Tancsa <mike@sentex.net>
To:        "Michael W. Lucas" <mwlucas@blackhelicopters.org>
Cc:        questions@freebsd.org
Subject:   Re: em0 NIC slow on 8.2-p1 amd64?
Message-ID:  <4E29D37E.8070908@sentex.net>
In-Reply-To: <20110722190821.GA55040@bewilderbeast.blackhelicopters.org>
References:  <20110720160432.GA41775@bewilderbeast.blackhelicopters.org> <4D6348DC01B1E406@> <20110720191640.GA42821@bewilderbeast.blackhelicopters.org> <4D301C9702145855@> <20110721141011.GB47190@bewilderbeast.blackhelicopters.org> <4D7F86D5017EB3D8@> <20110721151210.GA47663@bewilderbeast.blackhelicopters.org> <4D301C9702253683@> <4E29BC1F.6060103@sentex.net> <20110722190821.GA55040@bewilderbeast.blackhelicopters.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On 7/22/2011 3:08 PM, Michael W. Lucas wrote:
> 
> Basically the same.  I don't think it's disk.

Are you able to saturate the ethernet ?  Try something like


/usr/src/tools/tools/netrate/netblast/netblast

on the local ethernet and see if you can generate and receive a full
gigabit on the wire



# ./netblast 10.11.11.243 500 250 10

start:             1311363559.619619931
finish:            1311363569.619418229
send calls:        5210086
send errors:       1249890
approx send rate:  396019
approx error rate: 0


./netblast 10.11.11.241 500 300 10

start:             1311346264.584043978
finish:            1311346274.583848423
send calls:        7417664
send errors:       4214676
approx send rate:  320298
approx error rate: 0


When the sender if an igb nic, I am able to push out 955Mb to the em0
nic acting as a sink. The other way around, 855Mb
The em nic is

em1@pci0:5:0:0: class=0x020000 card=0x34ec8086 chip=0x10d38086 rev=0x00
hdr=0x00
    vendor     = 'Intel Corporation'
    device     = 'Intel 82574L Gigabit Ethernet Controller (82574L)'
    class      = network
    subclass   = ethernet
    bar   [10] = type Memory, range 32, base 0xb2500000, size 131072,
enabled
    bar   [18] = type I/O Port, range 32, base 0x1000, size 32, enabled
    bar   [1c] = type Memory, range 32, base 0xb2520000, size 16384, enabled
    cap 01[c8] = powerspec 2  supports D0 D3  current D0
    cap 05[d0] = MSI supports 1 message, 64 bit
    cap 10[e0] = PCI-Express 1 endpoint max data 128(256) link x1(x1)
    cap 11[a0] = MSI-X supports 5 messages in map 0x1c enabled
ecap 0001[100] = AER 1 0 fatal 0 non-fatal 1 corrected
ecap 0003[140] = Serial 1 001517ffffed36e4

em1: <Intel(R) PRO/1000 Network Connection 7.2.3> port 0x1000-0x101f mem
0xb2500000-0xb251ffff,0xb2520000-0xb2523fff irq 16 at device 0.0 on pci5
em1: Using MSIX interrupts with 3 vectors
em1: [ITHREAD]
em1: [ITHREAD]
em1: [ITHREAD]
em1: Ethernet address: 00:15:17:ed:36:e4



ifstat -b shows

# ifstat -b -i igb0
       igb0
 Kbps in  Kbps out
    0.00      0.00
    0.00      0.00
578794.0      0.00
855298.0      0.00
855365.9      0.00
855316.6      0.00
855335.2      0.00
855346.5      0.00
855358.6      0.00
855368.7      0.00
855356.6      0.00
727163.6      0.00
    0.00      0.00

thats the em equipped machine generating the traffic


	---Mike





> 
> There's kern/152828 claiming a performance regression with em under
> 8.2, but I'm not sure if that is applicable to my system. I'd upgrade
> the kernel to test, but I'm not brave enough to downgrade the kernel
> to 8.0 for comparison. (I've never had good luck running an old kernel
> on a new userland.)
> 
> ==ml
> 


-- 
-------------------
Mike Tancsa, tel +1 519 651 3400
Sentex Communications, mike@sentex.net
Providing Internet services since 1994 www.sentex.net
Cambridge, Ontario Canada   http://www.tancsa.com/



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4E29D37E.8070908>