At 05:42 PM 1/9/2014, you wrote: >
>--- In f..., "Younger, Arthur Steven"
> wrote:
> >
> > Along these lines, there is another old book (~ 30 years) on the
> lost art of Bit-Slice development called: "Bit-Slice Microprocessor
> Design" by John Mick and Jim Brick. ISBN 0-07-041781-4.
> >
> > Even though its uses bitslice technology , it has a detailed
> description of the design of a CPU Architecture, similar to the one
> Richard describes below. It would be a guide toward developing a
> FPGA based computer.
> >
> > For those who don't know, BitSlice was used to build many custom
> CPUs before FPGAs, including real-times graphics processors used in
> early Military flight simulations. This book was sort of the
> "bible" of that industry.
> >
> > A lot of the ideas should translate into FPGA implementations.
> >
> > -Steve
> >In fact, I have Mick and Brick on my bookshelf! I took a class in
>bit-slice design back in the early '80s. I still like the microcode
>approach to digital systems. The much more weighty volume
>"Microprogramming - Principles and Practices" by Husson describes
>microprogramming the IBM 360.
>
>Mick and Brick's work involves using a meta-assembler to create the
>very wide control words. At the time I didn't have anything workable
>so hand assembly was the best I could do.
>
>I ran into the 2100 logic series when I was dealing with early hard
>drive interfaces. There was a barn full of digital involved with
>reading hard drives and dealing with a SCSI interface.
>
>Richard That reminds me of the computer that my IO processor board plugged
into. It was an attached array processor, sort of a DSP chip in a
dual rack cabinet performing 100 MFLOPS. The compute head was ECL
gate arrays with ECL control store and was microcoded ala Mick and
Brick style rather than a more conventional assembly language. I
want to say the control word was around 100 bits wide but I can't be
certain since it was some 40 years ago. Although the product didn't
have a long lifetime, it was a very interesting unit to work on.
Rick
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
Mick and Brick was an AWESOME book and taught me a lot about
processor design. At one point I even designed an IO board for a
computer using a sequencer which was in essence the instruction fetch
portion of the designs Mick and Brick talked about.
One thing they disclose in that book that is very relevant to FPGA
designs is the critical path in the clock cycle timing. It is in the
path of fetching the next instruction based on the result of the
current ALU operation. Of course this assumes the design can perform
a branch on every instruction. That is often the critical path in a
CPU design even if the ALU ops and branches are in separate
instructions as both parts involve carry chains and can only be
optimized to a limited extent.
Rick
At 04:51 PM 1/9/2014, you wrote: >Along these lines, there is another old book (~ 30
years) on the
>lost art of Bit-Slice development called: "Bit-Slice Microprocessor
>Design" by John Mick and Jim Brick. ISBN 0-07-041781-4.
>
>Even though its uses bitslice technology , it has a detailed
>description of the design of a CPU Architecture, similar to the one
>Richard describes below. It would be a guide toward developing a
>FPGA based computer.
>
>For those who don't know, BitSlice was used to build many custom
>CPUs before FPGAs, including real-times graphics processors used in
>early Military flight simulations. This book was sort of the "bible"
>of that industry.
>
>A lot of the ideas should translate into FPGA implementations.
>
>-Steve
>
>________________________________________
>From: f... [f...] On Behalf
>Of rtstofer [r...@pacbell.net]
>Sent: Thursday, January 09, 2014 3:35 PM
>To: f...
>Subject: [fpga-cpu] Re: Beginning with FGPA
>
>--- In f..., "reubs@..." wrote:
> >
> > Hello,
> > I am a senior in high school and have been studying ALU and CPU
> design independently. I have just read "Bebop Bytes Back" and have
> just purchased a book titled "HDL Chip Design". Starting with
> FPGA's was recommended to me, but I was overwhelmed after trying to
> choose a board. Would anyone be able to recommend a board for a
> beginner, one with a generous amount of tools (switches, 7-segment
> display etc.) but one that does not take away from the experience
> by having many of the difficult aspects of building a CPU
> previously completed. Thank you very much and I appreciate any input.
> >The CPU design is the fun part. If you want to pursue it in a
>structured manner, consider "Microprocessor Design Using Verilog
>HDL". The author builds a Z80 microprocessor using Excel
>spreadsheets. It is a very structured approach. Nice...
>
>There is another book "Computer Architecture" by Caxton C Foster
>that is available at Alibris.com. The book is 40 years out of date.
>Except... There is a complete description and logic design for a 16
>bit 4k word minicomputer that might have been sold in the late '60s
>or early '70s. It's a trivial machine but the design is complete. A
>little rework in the start/stop logic and the thing could be easily
>ported to a Spartan 3 Starter Board.
>
>So, one of the electronics gurus did just that.
>http://www.youtube.com/watch?v=dt4zezZP8w8
>
>There used to be a web site blue.hotsolder.com that had the complete
>design. It doesn't seem to be available so, if you are interested,
>contact Al Williams at hotsolder.com and see what's happening. I
>have a copy of the distribution.
>
>His design adds a BUNCH of new instructions and registers. In
>addition, and this is a BIG deal, he wrote an assembler for it. In
>the end, these projects are worthless without a way to program them.
>
>That BeMicro project even has a C compiler:
>http://www.altera.com/b/nios-bemicro-evaluation-kit.html
>For $50, there is a lot of education in this project. Unfortunately,
>it all works! There isn't a lot to learn and it's certainly not like
>rolling your own.
>
>Just get real good at designing finite state machines. This whole
>FPGA thing is all about FSAs. And some combinatorial logic...
>
>Richard
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
--- In f..., "Younger, Arthur Steven" wrote: >
> Along these lines, there is another old book (~ 30 years) on the lost art of
Bit-Slice development called: "Bit-Slice Microprocessor Design" by John Mick and
Jim Brick. ISBN 0-07-041781-4.
>
> Even though its uses bitslice technology , it has a detailed description of
the design of a CPU Architecture, similar to the one Richard describes below. It
would be a guide toward developing a FPGA based computer.
>
> For those who don't know, BitSlice was used to build many custom CPUs
before FPGAs, including real-times graphics processors used in early Military
flight simulations. This book was sort of the "bible" of that industry.
>
> A lot of the ideas should translate into FPGA implementations.
>
> -Steve
>
In fact, I have Mick and Brick on my bookshelf! I took a class in bit-slice
design back in the early '80s. I still like the microcode approach to
digital systems. The much more weighty volume "Microprogramming - Principles
and Practices" by Husson describes microprogramming the IBM 360.
Mick and Brick's work involves using a meta-assembler to create the very
wide control words. At the time I didn't have anything workable so hand
assembly was the best I could do.
I ran into the 2100 logic series when I was dealing with early hard drive
interfaces. There was a barn full of digital involved with reading hard drives
and dealing with a SCSI interface.
Richard
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
On 1/9/2014 2:51 PM, Younger, Arthur Steven wrote: > Along these lines, there is another old book (~ 30
years) on the lost
> art of Bit-Slice development called: "Bit-Slice Microprocessor
> Design" by John Mick and Jim Brick. ISBN 0-07-041781-4.
>
> Even though its uses bitslice technology , it has a detailed
> description of the design of a CPU Architecture, similar to the one
> Richard describes below. It would be a guide toward developing a FPGA
> based computer.
>
> For those who don't know, BitSlice was used to build many custom CPUs
> before FPGAs, including real-times graphics processors used in early
> Military flight simulations. This book was sort of the "bible" of
> that industry.
>
> A lot of the ideas should translate into FPGA implementations.
But most machines of that era, used a lot of micro-coding for custom
cpu's. All that knowledge is locked up or thrown out, depending on luck.
Ben.
> -Steve
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
Reply by "Younger, Arthur Steven"●January 9, 20142014-01-09
Along these lines, there is another old book (~ 30 years) on the lost art of
Bit-Slice development called: "Bit-Slice Microprocessor Design" by John Mick and
Jim Brick. ISBN 0-07-041781-4.
Even though its uses bitslice technology , it has a detailed description of the
design of a CPU Architecture, similar to the one Richard describes below. It
would be a guide toward developing a FPGA based computer.
For those who don't know, BitSlice was used to build many custom CPUs
before FPGAs, including real-times graphics processors used in early Military
flight simulations. This book was sort of the "bible" of that industry.
A lot of the ideas should translate into FPGA implementations.
-Steve
________________________________________
From: f... [f...] On Behalf Of rtstofer [r...@pacbell.net]
Sent: Thursday, January 09, 2014 3:35 PM
To: f...
Subject: [fpga-cpu] Re: Beginning with FGPA
--- In f..., "reubs@..." wrote: >
> Hello,
> I am a senior in high school and have been studying ALU and CPU design
independently. I have just read "Bebop Bytes Back" and have just purchased a
book titled "HDL Chip Design". Starting with FPGA's was recommended to me,
but I was overwhelmed after trying to choose a board. Would anyone be able to
recommend a board for a beginner, one with a generous amount of tools (switches,
7-segment display etc.) but one that does not take away from the experience by
having many of the difficult aspects of building a CPU previously completed.
Thank you very much and I appreciate any input.
>
The CPU design is the fun part. If you want to pursue it in a structured manner,
consider "Microprocessor Design Using Verilog HDL". The author builds a Z80
microprocessor using Excel spreadsheets. It is a very structured approach.
Nice...
There is another book "Computer Architecture" by Caxton C Foster that is
available at Alibris.com. The book is 40 years out of date. Except... There is a
complete description and logic design for a 16 bit 4k word minicomputer that
might have been sold in the late '60s or early '70s. It's a
trivial machine but the design is complete. A little rework in the start/stop
logic and the thing could be easily ported to a Spartan 3 Starter Board.
There used to be a web site blue.hotsolder.com that had the complete design. It
doesn't seem to be available so, if you are interested, contact Al Williams
at hotsolder.com and see what's happening. I have a copy of the
distribution.
His design adds a BUNCH of new instructions and registers. In addition, and this
is a BIG deal, he wrote an assembler for it. In the end, these projects are
worthless without a way to program them.
That BeMicro project even has a C compiler: http://www.altera.com/b/nios-bemicro-evaluation-kit.html
For $50, there is a lot of education in this project. Unfortunately, it all
works! There isn't a lot to learn and it's certainly not like rolling
your own.
Just get real good at designing finite state machines. This whole FPGA thing is
all about FSAs. And some combinatorial logic...
Richard
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
--- In f..., "reubs@..." wrote: >
> Hello,
> I am a senior in high school and have been studying ALU and CPU design
independently. I have just read "Bebop Bytes Back" and have just purchased a
book titled "HDL Chip Design". Starting with FPGA's was recommended to me,
but I was overwhelmed after trying to choose a board. Would anyone be able to
recommend a board for a beginner, one with a generous amount of tools (switches,
7-segment display etc.) but one that does not take away from the experience by
having many of the difficult aspects of building a CPU previously completed.
Thank you very much and I appreciate any input.
>
The CPU design is the fun part. If you want to pursue it in a structured
manner, consider "Microprocessor Design Using Verilog HDL". The author builds a
Z80 microprocessor using Excel spreadsheets. It is a very structured approach.
Nice...
There is another book "Computer Architecture" by Caxton C Foster that is
available at Alibris.com. The book is 40 years out of date. Except... There
is a complete description and logic design for a 16 bit 4k word minicomputer
that might have been sold in the late '60s or early '70s. It's a
trivial machine but the design is complete. A little rework in the start/stop
logic and the thing could be easily ported to a Spartan 3 Starter Board.
There used to be a web site blue.hotsolder.com that had the complete design. It
doesn't seem to be available so, if you are interested, contact Al Williams
at hotsolder.com and see what's happening. I have a copy of the
distribution.
His design adds a BUNCH of new instructions and registers. In addition, and
this is a BIG deal, he wrote an assembler for it. In the end, these projects
are worthless without a way to program them.
That BeMicro project even has a C compiler: http://www.altera.com/b/nios-bemicro-evaluation-kit.html
For $50, there is a lot of education in this project. Unfortunately, it all
works! There isn't a lot to learn and it's certainly not like rolling
your own.
Just get real good at designing finite state machines. This whole FPGA thing is
all about FSAs. And some combinatorial logic...
Richard
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
--- In f..., ben wrote: >
> On 1/9/2014 3:53 AM, AndrPrado wrote:
> > Don't use AHDL, it's bad.
> > Use VHDL or Verilog, they are industry and academic standard. Yes you do
> > know the code it is going to generate, read the Pong P. Chu Book "RTL
> > Hardware desing using VHDL"
>
> I really don't see how it is bad?
> Not portable is the only bad feature I see.Good documentation
> solves that. FPGA's all seem to be upgraded with non standard features
> any how so your compiler has to be revised for every new chip.
There is a tendency to want portability over just about everything else. Code
reuse is paramount. In every language.
Unfortunately, the author used his own descriptive language 'sfl' and
it appears that there may not be a 'free' translator. I really like
the language but I hate that I can't use it.
It's an elegant language and design. But it won't work for me...
Richard
To post a message, send it to: f...
To unsubscribe, send a blank message to: fpga-cpu-unsubscribeYahoo Groups
Links
On 1/9/2014 3:53 AM, AndrPrado wrote: > Don't use AHDL, it's bad.
> Use VHDL or Verilog, they are industry and academic standard. Yes you do
> know the code it is going to generate, read the Pong P. Chu Book "RTL
> Hardware desing using VHDL"
I really don't see how it is bad?
Not portable is the only bad feature I see.Good documentation
solves that. FPGA's all seem to be upgraded with non standard features
any how so your compiler has to be revised for every new chip.
If I want to have a netlist generated say with FORTRAN IV it my
choice. Like they say with STANDARD'"you have so many to chose from."
Is there even a standard netlist format even?
Ben.
To post a message, send it to: f...
To unsubscribe, send a blank message to: f...Yahoo Groups Links
Don't use AHDL, it's bad.
Use VHDL or Verilog, they are industry and academic standard. Yes you do
know the code it is going to generate, read the Pong P. Chu Book "RTL
Hardware desing using VHDL"
On Thu, Jan 9, 2014 at 6:18 AM, wrote:
> ---In f..., wrote:
>
> On 1/8/2014 10:01 PM, gnuarm.2006@... wrote:
>
> > You might want to consider a small, simple, inexpensive board to start
> > with. You can always get a fancier board later. There are several
> > boards offered at Kickstarter that can be stand alone or combined with a
> > CPU board like the rPi or the Arduino. Here is one that is stand alone
> > and only costs $55.
> >
> >
>
http://www.kickstarter.com/projects/1106670630/mojo-digital-design-for-the-hobbyist?ref=live
> >
> > As to design methods, not using simulation shows a lack of understanding
> > of how to use an HDL or even FPGAs really.
>
> I see it as writing good software. I don't claim to do so
> but OLD SCHOOL PROGRAMIMG requires proper code or hardware
> documentation. If the module has a bug, then A) typo,
> B) Mistake in logic, C) Undefined behavor for undefined logic.
> I don't want to debate FPGA design in this thread, if you wish to do
that,
> let's start another thread.
> > I am a professional and have
>
> > done many, many designs and I never skimp on simulation. As a result
> > there are many times that my design works 100% the first time I fire it
> > up on the bench.
>
> Right now I need to understand the FPGA board I am using. Once that is
> done I can then spend time on my logic rather trying to fit other logic
> into my design. I got the DE1 because it has static ram, easy to get
> up and runing. It was worth the extra $20.I still need to learn the
> tools, but having good docmentaon is needed. The DEx boards don't have
> it and expect your $55 board has less.
>
> I'm a little confused. Are you the OP? I was responding to the
OP's
> request for a board for a beginner. If you need more that's fine.
I'm
> just not sure who asked the original question.
> > If you are doing a home brew project then it doesn't
>
> > matter so much that it works 100% I guess.
>
> With a homebrew, you get that last % when you feel like, not at
> 8:45 pm a day before some deline.
>
> You seem to have a fairly negative attitude about development. I use
> simulation because it is the *easiest* way to debug an FPGA. I can see any
> signal inside the device I wish and I can repeat all sorts of tests without
> any equipment. But like I said, let's open a new thread rather than
hijack
> this one.
>
> >
> > On the other hand, no one is perfect. I forgot a very basic design rule
> > on a test fixture recently and as a result had a bug that took a lot of
> > work to track down. lol
>
> The hardest bug I have, is software silently removing logic often do to
> a module change.
>
> Development software doesn't remove logic silently and it doesn't
remove
> logic that is actually used. The only logic that is removed is logic that
> does not connect to an output. Without a connection to the outside world
> there is no way for it to affect anything so why keep it in. The tools
> *always* give notes or warnings about removed logic. The problem is there
> are so many such warnings that are to be expected because logic is removed
> that is not needed such as carries out of an adder chain when that carry
> out is not described in the code. The adder drags the carry chain into the
> design and the carry out is then removed because it is not used. I may not
> be properly remembering this exactly, but that is the problem I have, too
> many warnings that must be scanned for important warnings.
>
> BTW, I used to hate Google groups with a passion. Yahoo groups has risen
> to the same level. I have no idea why they had to "update" their web pages
> and make them so much harder to use.
>
> Rick
>
>
>
--
Att
Andr
Reply by gnua...@arius.com●January 9, 20142014-01-09
I see it as writing good software. I don't claim to do so
but OLD SCHOOL PROGRAMIMG requires proper code or hardware
documentation. If the module has a bug, then A) typo,
B) Mistake in logic, C) Undefined behavor for undefined logic.
I don't want to debate FPGA design in this thread, if you wish to do that,
let's start another thread.
> I am a professional and have
> done many, many designs and I never skimp on simulation. As a result
> there are many times that my design works 100% the first time I fire it
> up on the bench.
Right now I need to understand the FPGA board I am using. Once that is
done I can then spend time on my logic rather trying to fit other logic
into my design. I got the DE1 because it has static ram, easy to get
up and runing. It was worth the extra $20.I still need to learn the
tools, but having good docmentaon is needed. The DEx boards don't have
it and expect your $55 board has less.
I'm a little confused. Are you the OP? I was responding to the OP's
request for a board for a beginner. If you need more that's fine.
I'm just not sure who asked the original question.
> If you are doing a home brew project then it doesn't
> matter so much that it works 100% I guess.
With a homebrew, you get that last % when you feel like, not at
8:45 pm a day before some deline.
You seem to have a fairly negative attitude about development. I use simulation
because it is the *easiest* way to debug an FPGA. I can see any signal inside
the device I wish and I can repeat all sorts of tests without any equipment.
But like I said, let's open a new thread rather than hijack this one.
>
> On the other hand, no one is perfect. I forgot a very basic design rule
> on a test fixture recently and as a result had a bug that took a lot of
> work to track down. lol
The hardest bug I have, is software silently removing logic often do to
a module change.
Development software doesn't remove logic silently and it doesn't
remove logic that is actually used. The only logic that is removed is logic
that does not connect to an output. Without a connection to the outside world
there is no way for it to affect anything so why keep it in. The tools *always*
give notes or warnings about removed logic. The problem is there are so many
such warnings that are to be expected because logic is removed that is not
needed such as carries out of an adder chain when that carry out is not
described in the code. The adder drags the carry chain into the design and the
carry out is then removed because it is not used. I may not be properly
remembering this exactly, but that is the problem I have, too many warnings that
must be scanned for important warnings.
BTW, I used to hate Google groups with a passion. Yahoo groups has risen to the
same level. I have no idea why they had to "update" their web pages and make
them so much harder to use.