[vortex] Tyan Thunder K7 S2462 Dual 3c59x

Heflin, Roger A. Roger.A.Heflin@conoco.com
Tue Jan 15 15:09:01 2002


Martin, and others:

Be very sure, on ours the ports appeared to work correctly with the 3com
driver,
under further testing, the 3com driver is doing something wrong and
working much
slower that it should be.   Things were not noticable under normal
network use,
under NFS with the client running sync and the server running async the
write speed
is about 10x slower on a S2562 machine (client) vs. any other machine we
have with Intel
EEPRO ethernet cards.  This was tested using the same server machine for
both
machines.

Copy of an 80 mb file from an nfs server back to an nfs server:
The eepro machine
	file copy		25.58,18.49(probably cached read),19.06
seconds 
	write only test:		9.0MB/second on a 200MB write.
Beckers 99Uc driver		
	file copy		27.98,18.92(probably cached read),17.82
seconds (driver gets errors and stops)
	write only test:		10MB/second on a 200MB write,
machine started getting transmit timed out
				after 930MB, apic or noapic does not
seem to matter, we can duplicate
				this one fairly quickly.
3com driver (3c90x 1.0.0.i)	
	file copy		52,47,44 seconds

	write only test:		1.56MB/second on a 100MB write

On the write test, I tested the eepro machine, then waited a couple of
minutes, and tested
the 99U driver, and then immediately tested the eepro machine again, and
the second
test results were consistant with the first test, this was to verify
that one test was not appearing
to affect another.

					Roger

> -----Original Message-----
> From:	Martin Siegert [SMTP:siegert@sfu.ca]
> Sent:	 1/ 15/ 2002 12:00 PM
> To:	vortex@scyld.com
> Subject:	Re: [vortex] Tyan Thunder K7 S2462 Dual 3c59x
> 
> On Mon, Jan 14, 2002 at 09:24:09PM -0700, Maurice Hilarius wrote:
> > With regards to your message at 09:41 AM 1/14/02, Kirk Bocek. Where
> you stated:
> > >Maurice, no the second IDE is not disabled. It is in use.
> > >
> > >Bogdan, yes I've correctly configured eth1. Here's the relevant
> part of the
> > >"lspci -v" output:
> > >
> > >***********************
> > >00:0f.0 Ethernet controller: 3Com Corporation 3c982 Dual Port
> Server Cyclone
> > >(rev 78)
> > >         Subsystem: Tyan Computer: Unknown device 2462
> > >         Flags: bus master, medium devsel, latency 80, IRQ 12
> > >         I/O ports at 1c00 [size=128]
> > >         Memory at f4004000 (32-bit, non-prefetchable) [size=128]
> > >         Expansion ROM at <unassigned> [disabled] [size=128K]
> > >         Capabilities: [dc] Power Management version 2
> > >
> > >00:10.0 Ethernet controller: 3Com Corporation: Unknown device 9807
> (rev 78)
> > >         Subsystem: Tyan Computer: Unknown device 6301
> > >         Flags: bus master, medium devsel, latency 64, IRQ 11
> > >         I/O ports at 1c80 [size=128]
> > >         Memory at f4004400 (32-bit, non-prefetchable) [size=128]
> > >         Expansion ROM at <unassigned> [disabled] [size=128K]
> > >         Capabilities: [dc] Power Management version 2
> > >***********************
> > >
> > >The first device (ID 9805) installs just fine. The second (ID 9807)
> refuses
> > >to be recognized.
> > >
> > >I've tried the 3Com 3c90x driver (version 1.02). It behaves the
> same way as
> > >the 3c59x.
> > >
> > >Thanks,
> > >Kirk Bocek
> > 
> > 
> > OK, I just realized this is the 982.
> > We also had problems with it.
> > Both in Linux AND in Windows.
> > In Linux what we saw was similar to this, and in Win2K it did not
> work at all.
> > 
> > At the time I wrote it off as a bad card, but I wonder..
> > Maybe a BIOS incompatibility?
> 
> For what it is worth:
> 
> I am using the Tyan Thunder K7 S2462 and its two onboard 3Com NICs
> without
> any problems (using BIOS version 2.09, but 2.02, 2.06, and 2.07a
> worked the
> same way with respect to the NICs).
> 
> Here is the output of lspci -v:
> 
> ======================================================================
> =====
> 00:0f.0 Ethernet controller: 3Com Corporation 3c982 Dual Port Server
> Cyclone (rev 78)
> 	Subsystem: Tyan Computer: Unknown device 2462
> 	Flags: bus master, medium devsel, latency 80, IRQ 3
> 	I/O ports at 2480 [size=128]
> 	Memory at f4008c00 (32-bit, non-prefetchable) [size=128]
> 	Expansion ROM at <unassigned> [disabled] [size=128K]
> 	Capabilities: [dc] Power Management version 2
> 
> 00:10.0 Ethernet controller: 3Com Corporation 3c982 Dual Port Server
> Cyclone (rev 78)
> 	Subsystem: Tyan Computer: Unknown device 2462
> 	Flags: bus master, medium devsel, latency 80, IRQ 11
> 	I/O ports at 2800 [size=128]
> 	Memory at f4009000 (32-bit, non-prefetchable) [size=128]
> 	Expansion ROM at <unassigned> [disabled] [size=128K]
> 	Capabilities: [dc] Power Management version 2
> ======================================================================
> =====
> 
> Secondly: I doubt (but I am not an expert) that this is a 3c982. These
> are the two onboard NICs (which aren't dual port NICs) - the writing
> on the
> chips says "3COM 920-ST03". Thus I always thought, despite the message
> from lspci, that these are 3c920 NICs.
> 
> Anyway, I am using the 3c59x driver from the 2.4.13 kernel (LK1.1.16)
> without
> problems.
> 
> Martin
> 
> ======================================================================
> ==
> Martin Siegert
> Academic Computing Services                        phone: (604)
> 291-4691
> Simon Fraser University                            fax:   (604)
> 291-4242
> Burnaby, British Columbia                          email:
> siegert@sfu.ca
> Canada  V5A 1S6
> ======================================================================
> ==
> _______________________________________________
> vortex mailing list
> vortex@scyld.com
> http://www.scyld.com/mailman/listinfo/vortex