Date   : Sun, 24 Jul 2005 21:07:52 +0100
From   : Sprow <info@...>
Subject: Re: ARM copros, speech cartridges, real time clocks, etc

In article <fcbb928f4d.philpem@...>,
   Philip Pemberton <philpem@...> wrote:
> In message <1122206439.10259.37.camel@...>
>           Jules Richardson <julesrichardsonuk@...> wrote:

> > I seem to recall the need for the 6502 was partly due to speed
> > limitations at the ISA bus, and partly due to the overhead in processing
> > on the PC itself. 
> > 
> > The second bit would unlikely be a problem these days. Not sure about
> > the former as I don't know what USB throughput is like compared to the
> > ISA bus.
>
> About twice as fast IIRC.
> PCI is much faster - 33MHz main clock (IIRC - might be 66MHz)

Conventional PCI is 33MHz, with the option to run at 66MHz. There are also
32 and 64 bit versions. During a burst therefore you get approx half a
gigabyte per second.

In practice the address cycles, turnaround time, and other bus users slow
you down.

> Simple way: FPGA + PIC + serial EEPROM. Blow the microcode (fuse map) for 
> the
> FPGA into the SEEPROM, then rig the PIC to upload said microcode at boot
> time. By the time the BIOS does a PCI scan, the card should be up and
> running.

I'm lost, why not just wire the EEPROM to the FPGA? I've never used an FPGA
which couldn't load its own config,
Sprow.