<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2900.2802" name=GENERATOR></HEAD>
<BODY>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>Hi
Pantelis,</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>Thanks
for the express reply.</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>I know
that what I say seems incredible. </FONT></SPAN><SPAN
class=609480116-14022006><FONT face=Arial color=#0000ff size=2>But I don't
understand what NAPI does.</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>My
measure is very simple. I display a message </FONT></SPAN><SPAN
class=609480116-14022006><FONT face=Arial color=#0000ff size=2>on the client
when 1000 exchanges are done.</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>I
already check the ifconfig stats on the board after some
seconds.</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>The
measures seems the same with or without NAPI.</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>More
details ?</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff size=2>Thanks
again</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2>Laurent</FONT></SPAN></DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2></FONT></SPAN> </DIV>
<DIV><SPAN class=609480116-14022006><FONT face=Arial color=#0000ff
size=2></FONT></SPAN> </DIV>
<BLOCKQUOTE
style="PADDING-LEFT: 5px; MARGIN-LEFT: 5px; BORDER-LEFT: #0000ff 2px solid">
<DIV class=OutlookMessageHeader dir=ltr align=left><FONT face=Tahoma
size=2>-----Message d'origine-----<BR><B>De :</B> Pantelis Antoniou
[mailto:pantelis.antoniou@gmail.com]<BR><B>Envoyé :</B> mar. 14 février
2006 16:29<BR><B>À :</B> Laurent Lagrange<BR><B>Cc :</B>
linuxppc-embedded@ozlabs.org<BR><B>Objet :</B> Re: Gianfar is slower than
fcc_enet on MPC8541 ???<BR><BR></FONT></DIV>Hi Laurent,<BR><BR>I found that
pretty hard to believe.<BR><BR>What are you measuring exactly?<BR><BR>Speed of
replies? If so it's explainable since the TSECs
use<BR>NAPI.<BR><BR>Regards<BR><BR>Pantelis<BR><BR>
<DIV><SPAN class=gmail_quote>On 2/14/06, <B class=gmail_sendername>Laurent
Lagrange</B> <<A
href="mailto:lagrange@fr.oleane.com">lagrange@fr.oleane.com</A>>
wrote:</SPAN>
<BLOCKQUOTE class=gmail_quote
style="PADDING-LEFT: 1ex; MARGIN: 0pt 0pt 0pt 0.8ex; BORDER-LEFT: rgb(204,204,204) 1px solid"><BR>Hello,<BR><BR>I
work on a cutom MPC8541 board with Linux 2.6.9.<BR>The kernel activates the
L1 cache (instructions and data)<BR>and the L2 cache (entirely used as cache
and not as sram).<BR><BR>I configure<BR>1 FCC (FCC1), <BR>2 TSECs with or
without NAPI (no effect) but without stashing in L2 sram.<BR>All PHYs are
automatically configured in 100MB full
duplex.<BR><BR> eth0: Gianfar
Ethernet Controller Version 1.1,
00:10:cd:48:48:e0<BR> eth0:
Running with NAPI
disabled<BR> eth0: 64/64
RX/TX BD ring size<BR> eth1:
Gianfar Ethernet Controller Version 1.1,
00:10:cd:48:48:e1<BR> eth1:
Running with NAPI
disabled<BR> eth1: 64/64
RX/TX BD ring size <BR> eth2:
FCC ENET Version custom, 00:10:cd:48:48:e2<BR><BR>Then I launch 3 simple TCP
servers, one on each ports.<BR><BR>From remote machines I runs 3 TCP
clients.<BR>The client sends messages of 1000 bytes,<BR>The server receives
and echoes the message <BR>The client receives the echoed message, check the
content<BR>and sends a new message again.<BR><BR>The result is that the 2
TSECs are 2 times slower than the FCC.<BR><BR>If I run a "top" application
on the board, I use less than 10% of the CPU <BR>Each port consumes about
1/3 of the CPU.<BR><BR>Any idea on how to configure the gianfar driver
?<BR><BR>Thanks<BR>Laurent<BR><BR><BR>_______________________________________________<BR>Linuxppc-embedded
mailing list<BR><A
href="mailto:Linuxppc-embedded@ozlabs.org">Linuxppc-embedded@ozlabs.org</A><BR><A
href="https://ozlabs.org/mailman/listinfo/linuxppc-embedded">https://ozlabs.org/mailman/listinfo/linuxppc-embedded</A><BR></BLOCKQUOTE></DIV><BR></BLOCKQUOTE></BODY></HTML>