17 Jul
2014
17 Jul
'14
2:03 a.m.
On Jul 16, 2014 4:18 PM, "Peter Corlett" <abuse at cabal.org.uk> wrote:
The NVAX is 80-90MHz which is pleasingly similar to
the maximum speed of
that
68000 core. Did you fancy a cluster of four NVAXen
that runs on a chip
that
costs something like $15 in bulk and can be powered by
a lemon with a
couple of
nails banged into it?
Yes. But 2014 FPGA technology will not provide it at that price or at that
level of power consumption. Maybe in another ten years.
Even the off-brand FPGAs that are marketed as ultra low power aren't low
enough power to run from a lemon, and those aren't anywhere near the
density or speed you're talking about. From a lemon, about all you're going
to run is a single chip micro (MSP430, PIC, AVR, ARM Cortex-M0) at only a
couple MHz max.
An NVAX is *much* more complex than a 68000. I doubt that even the new
(read: unobtanium) Xilinx "Ultrascale" FPGAs would be able to run an
NVAX-like core at 80-90 MHz. Certainly the Virtex-7 won't, and the price of
a large enough Virtex-7 for four NVAX-like cores is closer "in bulk" to
$15000. (Last fall I did some consulting at a company that was prototyping
a new processor architecture in the largest available Virtex-6, and they
had to strip it down some to get it to fit. It was less complex than the
NVAX, but had more parallelism.)
A simpler VAX design, with much less agressive cycle count reduction, is
quite feasible in a low cost FPGA (around $50), but won't have nearly the
performance of an NVAX, and still won't run from a lemon.
In terms of price/performance, simulating a VAX on an x86 is going to be
far better than an FPGA. The point to doing it in an FPGA is as an exercise
to sharpen one's hardware design skills, and for fun.