<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.0.6487.1">
<TITLE>Re: [Beowulf] 3com isa and other ethernet adapter : no link</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>On Tue, 7 Sep 2004, Alexandre Abbes wrote:</FONT>
</P>
<P><FONT SIZE=2>> Hello</FONT>
<BR><FONT SIZE=2>> </FONT>
<BR><FONT SIZE=2>> Thanks for your answer.</FONT>
<BR><FONT SIZE=2>></FONT>
<BR><FONT SIZE=2>> But I am sure it is not a 3c59x problem. My pci card wich is a vortex</FONT>
<BR><FONT SIZE=2>> boomerang and uses 3c59x.o module works perfectly. I can connect it (on</FONT>
<BR><FONT SIZE=2>> the same machine) to my Nvidia adapter which uses forcedeth.o module.</FONT>
<BR><FONT SIZE=2>></FONT>
<BR><FONT SIZE=2>> Again, there is a problem with the 3com isa ethernet adapter etherlink</FONT>
<BR><FONT SIZE=2>> III (3c509B) that cannot get the link with other ethernet adapters</FONT>
<BR><FONT SIZE=2>> different from 3c509B.</FONT>
</P>
<P><FONT SIZE=2>This sounds much more likely to be a hardware problem than a software</FONT>
<BR><FONT SIZE=2>problem. For example, the other adapters (or the switch port) on the</FONT>
<BR><FONT SIZE=2>segment with it might have very slightly less voltage or current or</FONT>
<BR><FONT SIZE=2>slightly shifted timing compared to the 509B. Not out of spec,</FONT>
<BR><FONT SIZE=2>necessarily, but shifted enough so that they won't interoperate. This</FONT>
<BR><FONT SIZE=2>problem might even depend on the host systems in some subtle way --</FONT>
<BR><FONT SIZE=2>borderline voltages or browned out power in the hosts might push</FONT>
<BR><FONT SIZE=2>periperals to the edge of their operating regimes, an overclocked</FONT>
<BR><FONT SIZE=2>bus.</FONT>
</P>
<P><FONT SIZE=2>I don't know how one might solve such a problem other than to replace</FONT>
<BR><FONT SIZE=2>the offending NICs with ones that are more interoperable (or fix the</FONT>
<BR><FONT SIZE=2>electrical power, or overclocking, if it turns out to be that instead).</FONT>
<BR><FONT SIZE=2>Fortunately, NICs are cheap, power can be regulated, and overclocking is</FONT>
<BR><FONT SIZE=2>a bad idea anyway, if that is in any way involved.</FONT>
</P>
<P><FONT SIZE=2>Otherwise I'm fairly clueless. If the software drivers work at all for</FONT>
<BR><FONT SIZE=2>any adapter combination, they ought to work for them all unless the</FONT>
<BR><FONT SIZE=2>hardware itself is bent to the edges of the specification.</FONT>
</P>
<P><FONT SIZE=2> rgb</FONT>
</P>
<P><FONT SIZE=2>-- </FONT>
<BR><FONT SIZE=2>Robert G. Brown <A HREF="http://www.phy.duke.edu/~rgb/">http://www.phy.duke.edu/~rgb/</A></FONT>
<BR><FONT SIZE=2>Duke University Dept. of Physics, Box 90305</FONT>
<BR><FONT SIZE=2>Durham, N.C. 27708-0305</FONT>
<BR><FONT SIZE=2>Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu</FONT>
</P>
<BR>
<BR>
<P><FONT SIZE=2>_______________________________________________</FONT>
<BR><FONT SIZE=2>Beowulf mailing list, Beowulf@beowulf.org</FONT>
<BR><FONT SIZE=2>To change your subscription (digest mode or unsubscribe) visit <A HREF="http://www.beowulf.org/mailman/listinfo/beowulf">http://www.beowulf.org/mailman/listinfo/beowulf</A></FONT>
</P>
</BODY>
</HTML>