Re: EPROM speed question

["Peter L. Peres" <plp@actcom.co.il>]

pmfji, I have been lurking till now, but here is my 2 $0.01:

  I have run into a problem with certain BIOS-es that mask out any BIOS
chunks that do not satisfy the BIOS extension signature and SIZE. This was
totally unrelated to netboot, I was using a battery backed SRAM as BIOS
extension on a special card that I developed.

  It turned out that the BIOS shadowed the entire ROM extension area early
in the boot process. However, this meant that my empty SRAM failed the
test and was simply never copied into shadow RAM, no matter what the
settings of the shadow control were. When the SRAM was written on another
computer, and booted on this computer, it was read out correctly. 

  Also, I could not make this computer write to the card at all. The BIOS
simply instructed the chipset to mask out the entire ROM area after
booting, no matter what, and additionally shadowed only those BIOS
extensions that it found to be valid, no matter what I told it to do. Data
written to the location of the ROM was recorded by the RAM, but did not
get written to the SRAM.

  I believe that this would break the netboot flashcard too. 

  Also, re: timing, The standard ISA bus is run at about 8 MHz, which is
125 nsec. Every ROM access pulse I have ever seen took two of these
cycles, at least, and that is 250 nsec. Also, there is a spec somewhere in
the ISA papers that says that a BIOS ROM extension should be 250 nsec, 200
nsec to be really sure (if an extra buffer is used, such as on the netboot
flashcard for example). The only systems that may require faster ones, are
overclocked ISA buses. The spec where I got this is pre-Pentium era, but
the ISA bus has not changed meanwhile. I don't know how this applies to
PCI systems (I've never seen a PCI card with a boot ROM extension socket -
but this does not mean much).

hope this helps someone,
	Peter