Date: Tue, 15 Oct 2002 23:26:55 -0700 From: Lars Eggert <larse@ISI.EDU> To: Petri Helenius <pete@he.iki.fi> Cc: Luigi Rizzo <rizzo@icir.org>, freebsd-net@FreeBSD.ORG Subject: Re: ENOBUFS Message-ID: <3DAD06AF.7060701@isi.edu> References: <065901c27495$56a94c40$8c2a40c1@PHE> <3DAC8FAD.30601@isi.edu> <068b01c2749f$32e7cf70$8c2a40c1@PHE> <20021015161055.A27443@carp.icir.org> <06c901c274d8$e5280b80$8c2a40c1@PHE> <3DAD01A7.3020807@isi.edu> <071501c274db$222c3ea0$8c2a40c1@PHE>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --]
Petri Helenius wrote:
>>The 900Mbps are similar to what I see here on similar hardware.
>
> What kind of receive performance do you observe? I havent got that
> far yet.
Less :-) Let me tell you tomorrow, don't have the numbers here right now.
> 600Mbps per interface. Im going to try this out also on -CURRENT
> to see if it changes anything. Interrupts do not seem to pose a big
> problem because Im seeing only a few thousand em interrupts
> a second but since every packet involves a write call there are >100k
> syscalls a second.
So maybe syscalls/second are the bottleneck. On -current, try enabling
zero copy sockets, it seems to help somewhat. Other than that, I've not
found -current to be much different in terms of performance.
If you're just interested in maxing throughput, try sending over TCP
with large write sizes. In that case, syscall overhead is less, since
you amortize it over multiple packets. (But there are different issues
that can limit TCP throughput.)
> Ill try changing the packet sizes to figure out optimum.
I think I remember that 4K packets were fastest with the em hardware in
our case.
Lars
--
Lars Eggert <larse@isi.edu> USC Information Sciences Institute
[-- Attachment #2 --]
0 *H
01 0 + 0 *H
080fErtcvE.0
*H
01 0 UZA10U Western Cape10U Cape Town10U
Thawte Consulting1(0&U Certification Services Division1$0"UThawte Personal Freemail CA1+0) *H
personal-freemail@thawte.com0
000830000000Z
040827235959Z01 0 UZA10U Western Cape10U Cape Town10
U
Thawte1 0U Certificate Services1(0&U Personal Freemail RSA 2000.8.3000
*H
0 32c %E>nx'gڈD)c5*mp<ܮto034qmOe
KaU5u'rװ |CBPQ<9TIf - ki N0L0)U "0 0 10UPrivateLabel1-2970U 0 0 U 0
*H
1KG]qSl]y=&b""I'{9$
*8PUl
LGl X1B li+@]jy.%݊
Z<D&iHΥbb090%A0
*H
01 0 UZA10U Western Cape10U Cape Town10
U
Thawte1 0U Certificate Services1(0&U Personal Freemail RSA 2000.8.300
020824185339Z
030824185339Z0T10
UEggert1
0 U*Lars10U Lars Eggert1 0 *H
larse@isi.edu0"0
*H
0
6Fxΰ7a ED&0+Dj)ֽXCUcnleijmz~S0J jWV~ 1^({IݛLjӖ
ao:bP}WLVܱ욗cD ɖ_Kv.A(W4 9;Z8-uXE
6b
@_0%#d` Rto5 L0R`w@7
r Hcc U 3%7 N_o V0T0*+e!0 00L2uMyffBNUbNJJcdZ2s0U 0
larse@isi.edu0 U 0 0
*H
]Ȕ,fK< cj RZeLa n@Z6,=
fK?yO#8+ Ni*LSfpQg <(aӒ$kTx_AL1>ގ|S090%A0
*H
01 0 UZA10U Western Cape10U Cape Town10
U
Thawte1 0U Certificate Services1(0&U Personal Freemail RSA 2000.8.300
020824185339Z
030824185339Z0T10
UEggert1
0 U*Lars10U Lars Eggert1 0 *H
larse@isi.edu0"0
*H
0
6Fxΰ7a ED&0+Dj)ֽXCUcnleijmz~S0J jWV~ 1^({IݛLjӖ
ao:bP}WLVܱ욗cD ɖ_Kv.A(W4 9;Z8-uXE
6b
@_0%#d` Rto5 L0R`w@7
r Hcc U 3%7 N_o V0T0*+e!0 00L2uMyffBNUbNJJcdZ2s0U 0
larse@isi.edu0 U 0 0
*H
]Ȕ,fK< cj RZeLa n@Z6,=
fK?yO#8+ Ni*LSfpQg <(aӒ$kTx_AL1>ގ|S1'0#001 0 UZA10U Western Cape10U Cape Town10
U
Thawte1 0U Certificate Services1(0&U Personal Freemail RSA 2000.8.30%A0 + a0 *H
1 *H
0 *H
1
021016062656Z0# *H
1=YX m\0R *H
1E0C0
*H
0*H
0
*H
@0+0
*H
(0 *H
101 0 UZA10U Western Cape10U Cape Town10
U
Thawte1 0U Certificate Services1(0&U Personal Freemail RSA 2000.8.30%A0
*H
ۮ!h 6R
_9.iŦd'/1ę|T2(k?s'[?^C$/|{/hG: 8z#LִG~5u!
Pdcy( XԋMcSሧcܚrGt*+ gQ7q5#,oO@iXkwNFIiuMeSi96;mi<.r yy L G3~uwSzl!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3DAD06AF.7060701>