5 May
2001
5 May
'01
11:50 a.m.
From: Chuck McManis <cmcmanis(a)mcmanis.com>
tools are *free*. You do have to take the time to learn
VHDL but so far
the
payoff has been worth it for me. I'm having a blast
with this thing. My
VHDL.... that was the cost I refer to. Most of me design experience
is pre VHDL availbility. I already have the Lattice Synario and that
bends my mind greatly. I havent used the small 2064 and 3030 FPGAs
I have a good handful of yet. Time is costly for me these days.
"final" is a PDP-8 w/ Serial terminal (think
DECMate in a single chip)
with
full lights and switches. This chip can do that easily.
I don't think it
Major cool. My personal "I'd like to do" is a 32 bit wide '8 by
grafting
another 20 bits to the right side of the word and running it fast. Same
instrcution set and the left 5 bits would remain the same. I'd use all
32 bits for OPR instructions to eliminate decoding (one per bit) and
simplify the IOT interface some. Obviously the page (formerly 128
words) would be much bigger but direct addressing of 256MW
out of a 4gb address space wouldn't be a significant shortcomming.
I doubt I'd need to do the EMA. ;)
Allison
6 May
6 May
4:28 a.m.
Sadly, the Xilinx 3000-series, my own stock of which was purchased in '90 or so,
when the smaller ones cost $300 or so each, have been orphaned in favor of a
later generation, which XILINX also would like to have go away, in favor of
their current stuff. The current software doesn't support the plain
3000-series, though it does still support the architecture in the form of the
3000A and 3100 series. My old XACT tools unfortunately don't support it either,
however, as the tools I bought only went through the 2K series. I've been
unable to find the dongle for that, however, though I don't miss it.
The OS under which the XILINX tools seem to run best, at least according to
their tech support folks, is Win98, though they claim it runs just as well under
NT4. I see essentially no difference, myself. There are new tools for LINUX,
by the way, though I don't know how well they work or are supported at XILINX.
They are VERY generous with support, however, once you've bought into their
tools, even though I only bought the cheapest set.
I'd not be afraid to give the SPARTAN series a try for building an experimental
version of a '70's-'80's generation CPU. They are relatively (choke)
inexpensive and available in versions with sufficient on-board RAM that you
might not need the higher pin counts required to interface external memory.
Unfortunately, the high-pin-count parts are hard to prototype.
The solution is to go to a gun show and buy a pistol, then stop at the local
liquor store to make a withdrawal. Then you'll be able to afford the
prototyping sockets, normally costing $400-500 or so when built onto a
wire-wrap-compatible adapter. Life is easier when one simply uses the 84-pin
PLCC versions.
Dick
----- Original Message -----
From: "ajp166" <ajp166(a)bellatlantic.net>
To: <classiccmp(a)classiccmp.org>
Sent: Saturday, May 05, 2001 5:50 PM
Subject: Re: Allison: 2910c version of z80
From: Chuck McManis <cmcmanis(a)mcmanis.com>
tools are *free*. You do have to take the time to
learn VHDL but so far
the
payoff has been worth it for me. I'm having a
blast with this thing. My
VHDL.... that was the cost I refer to. Most of me design experience
is pre VHDL availbility. I already have the Lattice Synario and that
bends my mind greatly. I havent used the small 2064 and 3030 FPGAs
I have a good handful of yet. Time is costly for me these days.
"final" is a PDP-8 w/ Serial terminal
(think DECMate in a single chip)
with
full lights and switches. This chip can do that
easily. I don't think it
Major cool. My personal "I'd like to do" is a 32 bit wide '8 by
grafting
another 20 bits to the right side of the word and running it fast. Same
instrcution set and the left 5 bits would remain the same. I'd use all
32 bits for OPR instructions to eliminate decoding (one per bit) and
simplify the IOT interface some. Obviously the page (formerly 128
words) would be much bigger but direct addressing of 256MW
out of a 4gb address space wouldn't be a significant shortcomming.
I doubt I'd need to do the EMA. ;)
Allison
5:19 a.m.
Sorry Richard, I have to jump in here ;-)
Richard Erlacher wrote:
I'd not be afraid to give the SPARTAN series a try for building an experimental
version of a '70's-'80's generation CPU. They are relatively (choke)
inexpensive
$20 for the xcs10, plcc84 package.
What did you pay for your 3k gates 3030 before ?
and available in versions with sufficient on-board RAM
that you
might not need the higher pin counts required to interface external memory.
Unfortunately, the high-pin-count parts are hard to prototype.
There really many prototype boards out there. And cheap ones ...
The solution is to go to a gun show and buy a pistol,
then stop at the local
liquor store to make a withdrawal.
Student license is $100 & the small prototype board around $150.
Still much cheaper than a good gun ;-)
Then you'll be able to afford the
prototyping sockets, normally costing $400-500 or so when built onto a
wire-wrap-compatible adapter. Life is easier when one simply uses the 84-pin
PLCC versions.
Cheers
6:50 a.m.
In 1990, when the 3000-series parts were new, the 3030's cost around $300, each,
in thousands, the 3042's, a bit more, and the 3090'w which were the biggest at
the time, cost about $650 in the <2k units per year we were buying.
There are still plenty of XILINX parts that, say, AVNET wants >$1k for in small
quantities. I agree, however, that the SPARTAN series, which seems to be the
most accessible series of XILINX parts, are quite reasonable, and the PLCC-84 is
far and away the most reasonable package for prototyping.
However, if you want to use the >300-pin packages that some devices force you to
use, the adapters for applying them to a prototype board, cost >$500. True,
this is for a socketed part, and the bulk of the cost is for the socket.
However, I have a tough time justifying a prototyping adapter for each of
several devices in the same package, particularly since they're so difficult to
solder by hand. (I'm gradually going blind ... ) Moreover, the BGA prototyping
sockets required for the really high-pin-count parts cost some real money.
Making an adapter PCB for these is not a trivial process either, since there's
no way to rework the BGA's if their contacts don't all "make" and
I'm not
willing to gamble with the high cost of these devices.
It's just as difficult accomodating the smaller parts from other vendors, with
their odd-fraction-of-a-millimeter pin pitch, etc. Eventually I'll figure out
how to manage the problem. For now, however, I'm going to complain every time I
get a chance.
XILINX, as I said before, is VERY generous with support of all sorts,
maintaining an active presence on the comp.arch.fpga newsgroup to help with all
sorts of questions. Once you've bought even the cheapest support software,
they'll help you all you want. Moreover, their free WEBPACK software is
adequate for many tasks. The higher-end tools required to do a proper job on a
BIG task cost some dough, though.
Dick
----- Original Message -----
From: "emanuel stiebler" <emu(a)ecubics.com>
To: <classiccmp(a)classiccmp.org>
Sent: Sunday, May 06, 2001 11:19 AM
Subject: Re: Allison: 2910c version of z80
84-pin
Sorry Richard, I have to jump in here ;-)
Richard Erlacher wrote:
>
> I'd not be afraid to give the SPARTAN series a try for building an
experimental
version of a
'70's-'80's generation CPU. They are relatively (choke)
inexpensive
$20 for the xcs10, plcc84 package.
What did you pay for your 3k gates 3030 before ?
and available in versions with sufficient
on-board RAM that you
might not need the higher pin counts required to interface external memory.
Unfortunately, the high-pin-count parts are hard to prototype.
There really many prototype boards out there. And cheap ones ...
The solution is to go to a gun show and buy a
pistol, then stop at the local
liquor store to make a withdrawal.
Student license is $100 & the small prototype board around $150.
Still much cheaper than a good gun ;-)
> Then you'll be able to afford the
> prototyping sockets, normally costing $400-500 or so when built onto a
> wire-wrap-compatible adapter. Life is easier when one simply uses the PLCC versions.
Cheers
5:36 a.m.
New subject: XiLinx XC40xx-PC84
Sadly, the Xilinx 3000-series, my own stock of which
was purchased in '90
or so,
when the smaller ones cost $300 or so each, have been orphaned in favor of a
I mentioned this to the list last Summer, but events in the family caught
me up and I didn't follow up, so here it is again. ;)
I found a dozen or so Xilinx protoboards, size of a punch card half of it
open, 2 digit display, couple buttons, some kind of serial looking
connection, runs off a 9v battery, big socket in the middle
with a square chip in it of the same name.
XiLinx XC40xx-PC84 the device being a XC4003A -5.
Who wants one for $5 and shipping, on the condition they figure out what it
is and report back?
8635
days inactive
8636
days old
4 comments
4 participants
participants (4)
-
ajp166＠bellatlantic.net -
edick＠idcomm.com -
emu＠ecubics.com -
mikeford＠socal.rr.com