Re: EPROM speed question

[Georg Hager <ghager@theorie3.physik.uni-erlangen.de>]

On Tue, 24 Feb 1998, Ken Yap wrote:

> >Hi all,
> >
> >I have managed to get a 16k boot image run from a floopy -- everything
> >works fine. After burning the image.rom into a 27128 EPROM and plugging it
> >into my NE2000 clone (yes, it allows 16k), the client didn't recognize the
> >presence of the PROM and booted from the local disk. With DEBUG I found
> >out that the PROM did not appear at the memory location I had specified in
> >the NE2000 setup (D8000); all it reads there is 0xFF. If I remove the PROM
> >from the card, DEBUG reads 0x80 from those locations.
> 
> It's more likely you are decoding the wrong part of the EPROM or there
> is a clash with existing memory or you have not told the EMS driver to
> exclude those locations. Normally blank areas read as 0xFF.

No, this is impossible: 1) I'm sure that I've decoded the right part of
the EPROM, because in the meantime I have managed to adapt my prommer to
150ns EPROMs and everything works fine. 2) There can't be any clashes with
existing memory, because there are no other adapters (except VGA) around,
and shadowing is turned off in the BIOS. 3) The EMS driver is not running
-- I have run DEBUG from a plain DOS disk with no drivers at all.

> 
> I am using 250ns EPROMS on my 3c509 NICs on a 386 and a 486. They work
> perfectly. I have this 450ns EPROM which still reads fine on a 16MHz
> 286. I suspect NICs insert wait states for access to onboard EPROMs so
> that they are less dependent on motherboards. However as a rule of thumb
> it should be safe to use EPROMs at least as fast at the BIOS (E)EPROMs.
> 

Well, maybe I'm out of luck and my NE2000 clone provides some very
unpleasant timing for the EPROM /OE and /CE lines.

BTW, there's some good news for people with very old (original) NE2000
cards: Although they are frequently reported to support only 8k EPROMs, I
noticed that address line A13 is actually connected to the ISA bus on an
old card I could lay my hands on (original from Novell, 1990; under the
socket on the card it says ``2764'' as well). Plugging in a 27128 worked
out of the box, using the minimal kernel (about 13k of the EPROM used). 

Bye,
Georg.

=========================================================================
Georg Hager             | Email:   ghager@theorie3.physik.uni-erlangen.de
Theoretische Physik III |                   Tel. (Uni): (+49)9131/85-8815
FAU Erlangen-N"urnberg  |                   Tel. (Pri):   (+49)911/735142
Staudtstr. 7            | Home URL:
D-91058 Erlangen        | http://theorie3.physik.uni-erlangen.de/~ghager/
=========================================================================