The testing also involved linux/windows communications...<br><br>From the webpage:<br>"Note: All cards were tested at 1500, 3000, 4000, and
6000 values for the TCP MTU size. The drivers for the cards were not
modified. Cards based upon the dp83820 chipset were limited to
6000MTU due to driver defaults. All other cards were tested through
9000MTU."<br><br>NOTE: <a href="http://www.cs.uni.edu/~gray/gig-over-copper/gig-over-copper.html">http://www.cs.uni.edu/~gray/gig-over-copper/gig-over-copper.html</a> describes that the benchmarks were gathered using Netpipe
2.4:<br><a href="http://www.scl.ameslab.gov/netpipe">http://www.scl.ameslab.gov/netpipe</a> <br><br>From the above netpipe webpage ... "Gigabit Ethernet delivers around 900 Mbps with latencies of 25-62 microseconds
from 64-bit 66 MHz PCI buses."<br><br><br> .. I haven't substantiated this but you may wish to consider the 100K+ packet size as the
TcpWindowSize as described in <br><br><a href="http://support.microsoft.com/kb/314053/en-us">http://support.microsoft.com/kb/314053/en-us</a><br><br>"
Description: This parameter determines the
maximum TCP receive window size of
the computer. The receive window specifies the
number of bytes a sender can transmit without
receiving an acknowledgment. Generally,
larger receive windows improve
performance over high (delay * bandwidth)
networks. For highest efficiency, the receive
window must be an even multiple of the TCP
Maximum Segment Size (MSS)."<br><br><br>The complete benchmark write-up is located at:<br><br><font size="-1"><font color="#660099"><a href="http://www.cns.uni.edu/%7Egray/gig-over-copper/hsln-lcn.ps" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.cns.uni.edu/~gray/gig-over-copper/hsln-lcn.ps</a><br><br></font>
</font><br>These were the OS environments:( Note: Even the win2k tests were over 600mbs....)<br><div><br>"Each system was configured to boot</div>
<div>into Debian 3.0 GNU/Linux and Microsoft Windows 2000</div>
<div>Professional. Versions 2.4.7, 2.4.19-pre9-ac2, and 2.5.7 of</div>
<div>the Linux kernel were used for testing on the GNU/Linux</div>
<div>operating system. Service pack 2 was applied to the Win-</div>
<div>dows 2000 installation."<br><br></div><br>Tim<br><br><div><span class="gmail_quote">On 8/21/06, <b class="gmail_sendername">Dan Hunter</b> <<a href="mailto:dmhunter@charter.net">dmhunter@charter.net</a>> wrote:
</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div>
<div bgcolor="#ffffff">
<div><font face="Arial" size="2">I have already set the packet size as large as the
windows driver allow - MTU = 7 Kb</font></div>
<div><font face="Arial" size="2"></font> </div>
<div><font face="Arial" size="2">But coping from a very high speed disk, or RAM
drive is certainly a good question.</font></div>
<div><font face="Arial" size="2">Is the limitation the ethernet, PCI bus, hard
driver controller, or the hard drive itself????</font></div>
<div><font face="Arial" size="2"></font> </div>
<div><font face="Arial" size="2">Hmmm, me thinks an experiment or two is in the
wind!</font></div>
<blockquote style="border-left: 2px solid rgb(0, 0, 0); padding-right: 0px; padding-left: 5px; margin-left: 5px; margin-right: 0px;"></blockquote></div><div><span class="e" id="q_10d3373bb159a871_1">
<div style="font-family: arial; font-style: normal; font-variant: normal; font-weight: normal; font-size: 10pt; line-height: normal; font-size-adjust: none; font-stretch: normal;">----- Original Message ----- </div>
<div style="background: rgb(228, 228, 228) none repeat scroll 0% 50%; -moz-background-clip: -moz-initial; -moz-background-origin: -moz-initial; -moz-background-inline-policy: -moz-initial; font-family: arial; font-style: normal; font-variant: normal; font-weight: normal; font-size: 10pt; line-height: normal; font-size-adjust: none; font-stretch: normal;">
<b>From:</b>
<a title="tim@timschoenfelder.com" href="mailto:tim@timschoenfelder.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">Tim
Schoenfelder</a> </div>
<div style="font-family: arial; font-style: normal; font-variant: normal; font-weight: normal; font-size: 10pt; line-height: normal; font-size-adjust: none; font-stretch: normal;"><b>To:</b> <a title="rhn-users@redhat.com" href="mailto:rhn-users@redhat.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
Red Hat Network Users List</a> </div>
<div style="font-family: arial; font-style: normal; font-variant: normal; font-weight: normal; font-size: 10pt; line-height: normal; font-size-adjust: none; font-stretch: normal;"><b>Sent:</b> Monday, August 21, 2006 1:37
PM</div>
<div style="font-family: arial; font-style: normal; font-variant: normal; font-weight: normal; font-size: 10pt; line-height: normal; font-size-adjust: none; font-stretch: normal;"><b>Subject:</b> Re: [rhn-users] Gigabit transfer
speeds</div>
<div><br></div>Increase your packet size...<br><br>One of our networking class
projects a couple years ago was to benchmark gigibit ethernet transfer
speeds. <br><br>I benchmarked 700mb/s+ between two Linux PCs which was
reasonable to our instructor. <br><br>He emphasized that the brand/chipset of
cards made a big difference ( it wasn't the most expensive that performed best
... ) as well as the drivers.<br><br>Some of his published tests as described
below show that you can achieve at least 700mb/s even with 32bit Netgear
GA622T cards as illustrated in a ps link from his website: <br><br><a href="http://www.cs.uni.edu/%7Egray/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">http://www.cs.uni.edu/~gray/</a><br><br>
Benchmark
project links:<br><a href="http://www.cs.uni.edu/%7Egray/gig-over-copper/" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">http://www.cs.uni.edu/~gray/gig-over-copper/</a><br><a href="http://www.cs.uni.edu/%7Egray/gig-over-copper/gig-over-copper.html" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://www.cs.uni.edu/~gray/gig-over-copper/gig-over-copper.html</a><br><br>PS
file showing performance benchmarks ( note 700mb/s+ requires packet size of
100k+ bytes ): <br><br><a href="http://www.cs.uni.edu/%7Egray/gig-over-copper/hsln-lcn.ps" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">http://www.cs.uni.edu/~gray/gig-over-copper/hsln-lcn.ps</a>
<br><br>Tim<br><br>
<div><span class="gmail_quote">On 8/21/06, <b class="gmail_sendername">Corne
Beerse</b> <<a href="mailto:cbeerse@gmail.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">cbeerse@gmail.com</a>>
wrote:</span>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Dan
Hunter wrote:<br><br>> I regularly move large files from a windows XP
machine to a Fedora<br>> Core 5 machine.<br>> To avoid bothering other
network users, I purchased two Linksys EG1032<br>> gigabit ether net
cards. <br>> The cards have been installed. a direct PC to PC
gigabit cross over<br>> cable connects them.<br>> A separate set of IP
address have been set up.<br>> I can verify that data is being transfered
over the gigabit cable with <br>> windows task manager.<br>><br>>
Transfers over the 10/100 link show up at 82 or 83 percent - about
80<br>> Mbps.<br>> Transfers over the gigabit link show up at 20 to 26
percent - about<br>> 230 Mbps. <br>><br>> What do I have set up
wrong?<br>><br>><br>Your expectations are to high. If you have about
20% utilisation with a<br>Gbit nic, I'd say you are doing verry
nice.<br><br>With the move to gigabit networking, the nic is no longer the
<br>bottleneck. If you move data from disk on one machine or to disk on
an<br>other, the disk-controllers, disk interfaces or even the actual
disks<br>are most likely the bottlenecks. Then, the disk does not need to be
the <br>bottleneck. Think about the speed of internal busses and for which
they<br>are used while you are dumping
data...<br><br><br>CBee<br><br>_______________________________________________<br>rhn-users
mailing list<br><a href="mailto:rhn-users@redhat.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">rhn-users@redhat.com</a><br><a href="https://www.redhat.com/mailman/listinfo/rhn-users" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
https://www.redhat.com/mailman/listinfo/rhn-users</a><br></blockquote></div><br><br clear="all"><br>-- <br>Tim Schoenfelder<br><a href="http://timschoenfelder.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
http://timschoenfelder.com</a>
</span></div><div><p>
</p><hr></div><div><span class="q">
<p></p>_______________________________________________<br>rhn-users mailing
list<br><a href="mailto:rhn-users@redhat.com" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">rhn-users@redhat.com</a><br><a href="https://www.redhat.com/mailman/listinfo/rhn-users" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
https://www.redhat.com/mailman/listinfo/rhn-users</a></span></div><div><p></p></div>
</div><br>_______________________________________________<br>rhn-users mailing list<br><a onclick="return top.js.OpenExtLink(window,event,this)" href="mailto:rhn-users@redhat.com">rhn-users@redhat.com</a><br><a onclick="return top.js.OpenExtLink(window,event,this)" href="https://www.redhat.com/mailman/listinfo/rhn-users" target="_blank">
https://www.redhat.com/mailman/listinfo/rhn-users</a><br><br></blockquote></div><br><br clear="all"><br>-- <br>Tim Schoenfelder<br><a href="http://timschoenfelder.com">http://timschoenfelder.com</a>