Ethernet Cards

Thomas Hoberg tmh at bigfoot.FOKUS.GMD.DBP.DE
Tue Dec 4 12:51:51 AEST 1990


In article <707 at denwa.uucp>, jimmy at denwa.info.com (Jim Gottlieb) writes:
|> Usually, we use the 8-bit Ethercard Plus in our systems running
|> Interactive 2.2 + TCP/IP 1.2.  Is there any reason not to move up to
|> the 16-bit card (I know we have to get the latest drivers from WD)?
|> It's not that much more expensive.
|> 
|> 
|> A recent post mentioned problems with the Ethercard Plus in 33MHz
|> systems.  Likewise, we had problems with it (the 8-bit version) in our
|> Olivetti M380/XP-9 machines (basically a 33MHz version of the AT&T
|> 6386E WGS).  We switched to the 3Com 3C503 and all was fine.  Our video
|> card (Olivetti supplied) doesn't have the option to switch it into
|> 8-bit mode.  I wonder if the 16-bit Ethercard Plus would have worked.
I am all confused! I remember to have read that with the AT bus design a 16-bit
peripheral will automatically use up 128k of memory space (something to do with
the signals available on the connector). This is probably not true for
peripherals in the I/O space (disk controllers and the like) and I can't really
believe it's true for LIM 3.2 type memory boards (such as the first Intel
Above Board), that use a single 64k window. However I recently experienced
trouble with a VGA board that was jumpered for 16-bit ROM access, thus over-
laying our Adaptec's ROM area. Rejumpering for 8-bit access solved the problem
(and we use shadow RAM anyway for DOS; Unix could not care less about ROM).
I really wonder whether there is any advantage to using a 16-bit Ethernet card
then, other than having a couple more IRQ lines available. The WD Ethercards
use only a 16k window and I guess nobody could afford to waste 128k of address
space for that. In any case the bandwidth provided by an 8-bit slot should be
quite sufficient for a 10Mb Ethernet. Somebody in this newsgroup said, that the
performance difference between Mylex's 32-bit EISA Ethernet card and a WD8003
was in the order of 5%, I'd really like a confirmation on that, though.
----
Thomas M. Hoberg   | UUCP: tmh at prosun.first.gmd.de  or  tmh%gmdtub at tub.UUCP
c/o GMD Berlin     |       ...!unido!tub!gmdtub!tmh (Europe) or
D-1000 Berlin 12   |       ...!unido!tub!tmh
Hardenbergplatz 2  |       ...!pyramid!tub!tmh (World)
Germany            | BITNET: tmh%DB0TUI6.BITNET at DB0TUI11 or
+49-30-254 99 160  |         tmh at tub.BITNET



More information about the Comp.unix.sysv386 mailing list