Which board to buy? Status of open source tools?

Tom Hawkins wrote:
> 
> jon@beniston.com (Jon Beniston) wrote in message news:<e87b9ce8.0405070500.7c9c4bcc@posting.google.com>...
> > > Also, what's the status of open source tools?  (I like to tinker on
> > > the software end of things, too, and I have a bad case of PowerBook
> > > envy, and I haven't seen any software available other than x86 and
> > > big-iron UNIX).  Can they go from VHDL/Verilog all the way to
> > > downloading the file to the chip?  I see that compilation (Icarus
> > > verilog) and downloading (found it in this group's FAQ) work, but what
> > > about place and route?  Icarus' docs say it's a no-go for this; is
> > > there anything else, or must I use the Xilinx tools?  (I guess I have
> > > a Xilinx bias -- it's all I've used, and I don't see much else for
> > > cheap development boards.)
> >
> > Why bother with Open Source stuff when various incarnations of
> > commerical tools are free to use?
> 
> Because in the long term, an open-source tool chain can out perform a
> commercial tool.  Case in point: how many Unix/C developers compile
> with a commercial alternative to GCC?

I have been evaluating tools for the ARM MCU and all my info says that
gcc is a less efficient tool than the commercial ones.  I guess it only
stands to reason that the commercial tools must be better, why else
would they sell well enough to keep the companies in business?  


> Unfortunately, the track record of open-source FPGA implementation
> tools has been dismal.  Icarus is the only open-source tool I know
> that attempts Xilinx synthesis.  I believe it handles synthesis and
> technology mapping, but has little, if no optimization.

I think John Jakson summed it up pretty well, open source FPGA tools are
not practical for a number of reasons that have been discussed here many
times before.  Some people seem to think that FPGA and ASIC tools are
the same as compiliers, but in reality they are very different.  


> I did start a Virtex2 packer/placer awhile back.  I got as far as
> parsing the post-synthesis EDIF netlist and building a graph of FPGA
> primitives.  The goal was to pack and place the CLBs, then update the
> UCF with LOC constraints.  I still have the code lying around in case
> anyone is interested.
> 
> BTW, Confluence can compile it for Mac X.  It will solved your design,
> simulation, and verification problems, but for FPGA compilation,
> you're still locked into commercial tools.

-- 

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

"rickman" <spamgoeshere4@yahoo.com> wrote in message
news:40A01F19.EDA80D9F@yahoo.com...
> Tom Hawkins wrote:
> >
> > jon@beniston.com (Jon Beniston) wrote in message
news:<e87b9ce8.0405070500.7c9c4bcc@posting.google.com>...
> > > > Also, what's the status of open source tools?  (I like to tinker on
> > > > the software end of things, too, and I have a bad case of PowerBook
> > > > envy, and I haven't seen any software available other than x86 and
> > > > big-iron UNIX).  Can they go from VHDL/Verilog all the way to
> > > > downloading the file to the chip?  I see that compilation (Icarus
> > > > verilog) and downloading (found it in this group's FAQ) work, but
what
> > > > about place and route?  Icarus' docs say it's a no-go for this; is
> > > > there anything else, or must I use the Xilinx tools?  (I guess I
have
> > > > a Xilinx bias -- it's all I've used, and I don't see much else for
> > > > cheap development boards.)
> > >
> > > Why bother with Open Source stuff when various incarnations of
> > > commerical tools are free to use?
> >
> > Because in the long term, an open-source tool chain can out perform a
> > commercial tool.  Case in point: how many Unix/C developers compile
> > with a commercial alternative to GCC?
>
> I have been evaluating tools for the ARM MCU and all my info says that
> gcc is a less efficient tool than the commercial ones.  I guess it only

What do you mean by "less efficient" ?  There are big differences between
run-time efficiency and develop-time efficiency, and the balance choosen
between these will vary according to user and project.  There is little
doubt that there are commercial compilers that produce smaller and faster
code than gcc for many processors (including arm), but developer efficiency
can be a serious issue too.  I write C code for five different processors,
amongs other work - it is simply not economically feasible to buy
top-of-the-range commercial compilers for all of these, while gcc works fine
for me.  I also find it in many ways to be a more advanced compiler than
some compilers I have used (especially in terms of compile-time error
checking).  There are plenty of cases where commercial compilers would make
most economic sense overall, but it is far from every case.

> stands to reason that the commercial tools must be better, why else
> would they sell well enough to keep the companies in business?
>

There is one, and only one, reason why any company stays in business - it is
able to persuade enough people that its product is worth the money.  It may
or may not be the case that the product actually *is* worth the money (to
the given customer), but the immediate technical qualities of the product
are only one part of the sale - there are many other relevant issues such as
support and upgrade paths, and many irrelevant issues such as the
persuasiveness of the salesfolk, the guilibitity of the purchasing people,
and the quality of their flashy power-point slides and free pens.  I'm not
trying to say that commercial compilers are not worth the money - I'm just
pointing out the naivity of the "it costs more so it must be worth more"
argument.

rickman <spamgoeshere4@yahoo.com> writes:

> Tom Hawkins wrote:
> > 
<snip>
> > tools has been dismal.  Icarus is the only open-source tool I know
> > that attempts Xilinx synthesis.  I believe it handles synthesis and
> > technology mapping, but has little, if no optimization.
> 
> I think John Jakson summed it up pretty well, open source FPGA tools are
> not practical for a number of reasons that have been discussed here many
> times before.  Some people seem to think that FPGA and ASIC tools are
> the same as compiliers, but in reality they are very different.  
> 

I think there's also the point that the intersection of the sets
"compiler users" and "can write (and is interested in writing)
software" is much larger than that for "FPGA place and route tool
users" and the "writing software" sets.

Cheers,
Martin

-- 
martin.j.thompson@trw.com
TRW Conekt, Solihull, UK
http://www.trw.com/conekt

rickman <spamgoeshere4@yahoo.com> wrote in message news:<40A01F19.EDA80D9F@yahoo.com>...
> Tom Hawkins wrote:
> > 
> > jon@beniston.com (Jon Beniston) wrote in message 
[snip]
> > >
> > > Why bother with Open Source stuff when various incarnations of
> > > commerical tools are free to use?
> > 
> > Because in the long term, an open-source tool chain can out perform a
> > commercial tool.  Case in point: how many Unix/C developers compile
> > with a commercial alternative to GCC?
> 
> I have been evaluating tools for the ARM MCU and all my info says that
> gcc is a less efficient tool than the commercial ones.  I guess it only
> stands to reason that the commercial tools must be better, why else
> would they sell well enough to keep the companies in business?

Certainly when a compiler targets a specific architecture, it can do a
better job.  But in terms of the number of supported platforms and its
decent code optimization, it's hard to beat GCC.
  
> 
> 
> > Unfortunately, the track record of open-source FPGA implementation
> > tools has been dismal.  Icarus is the only open-source tool I know
> > that attempts Xilinx synthesis.  I believe it handles synthesis and
> > technology mapping, but has little, if no optimization.
> 
> I think John Jakson summed it up pretty well, open source FPGA tools are
> not practical for a number of reasons that have been discussed here many
> times before.  Some people seem to think that FPGA and ASIC tools are
> the same as compiliers, but in reality they are very different.

Sure the algorithms are different, but there are many similarities: 
Like a compiler, most logic optimizations are NP-complete, therefore
many of the same strategies can be applied.   Both a compiler and a
synthesizer deal with dataflow analysis and resource allocation. 
Instruction reordering is similar to retiming.  Memory and execution
timing tradeoffs are somewhat analogous to tradeoffs with area and
performance.

I agree the closer you get to implementation, the less pratical an
open source tool becomes.  But where is the line drawn?  Clearly
synthesis falls on our side.  Multi-chip partioning is certainly
possible.  And with the capabilities of UCF, I would argue placement
is a canidate as well.
  
> 
> 
> > I did start a Virtex2 packer/placer awhile back.  I got as far as
> > parsing the post-synthesis EDIF netlist and building a graph of FPGA
> > primitives.  The goal was to pack and place the CLBs, then update the
> > UCF with LOC constraints.  I still have the code lying around in case
> > anyone is interested.
> > 
> > BTW, Confluence can compile it for Mac X.  It will solved your design,
> > simulation, and verification problems, but for FPGA compilation,
> > you're still locked into commercial tools.
>  
> -- 
> 
> Rick "rickman" Collins
> 
> rick.collins@XYarius.com
> Ignore the reply address. To email me use the above address with the XY
> removed.
> 
> Arius - A Signal Processing Solutions Company
> Specializing in DSP and FPGA design      URL http://www.arius.com
> 4 King Ave                               301-682-7772 Voice
> Frederick, MD 21701-3110                 301-682-7666 FAX

David Brown wrote:
> 
> "rickman" <spamgoeshere4@yahoo.com> wrote in message
> news:40A01F19.EDA80D9F@yahoo.com...
> > Tom Hawkins wrote:
> > >
> > > jon@beniston.com (Jon Beniston) wrote in message
> news:<e87b9ce8.0405070500.7c9c4bcc@posting.google.com>...
> > > > > Also, what's the status of open source tools?  (I like to tinker on
> > > > > the software end of things, too, and I have a bad case of PowerBook
> > > > > envy, and I haven't seen any software available other than x86 and
> > > > > big-iron UNIX).  Can they go from VHDL/Verilog all the way to
> > > > > downloading the file to the chip?  I see that compilation (Icarus
> > > > > verilog) and downloading (found it in this group's FAQ) work, but
> what
> > > > > about place and route?  Icarus' docs say it's a no-go for this; is
> > > > > there anything else, or must I use the Xilinx tools?  (I guess I
> have
> > > > > a Xilinx bias -- it's all I've used, and I don't see much else for
> > > > > cheap development boards.)
> > > >
> > > > Why bother with Open Source stuff when various incarnations of
> > > > commerical tools are free to use?
> > >
> > > Because in the long term, an open-source tool chain can out perform a
> > > commercial tool.  Case in point: how many Unix/C developers compile
> > > with a commercial alternative to GCC?
> >
> > I have been evaluating tools for the ARM MCU and all my info says that
> > gcc is a less efficient tool than the commercial ones.  I guess it only
> 
> What do you mean by "less efficient" ?  There are big differences between
> run-time efficiency and develop-time efficiency, and the balance choosen
> between these will vary according to user and project.  There is little
> doubt that there are commercial compilers that produce smaller and faster
> code than gcc for many processors (including arm), but developer efficiency
> can be a serious issue too.  I write C code for five different processors,
> amongs other work - it is simply not economically feasible to buy
> top-of-the-range commercial compilers for all of these, while gcc works fine
> for me.  I also find it in many ways to be a more advanced compiler than
> some compilers I have used (especially in terms of compile-time error
> checking).  There are plenty of cases where commercial compilers would make
> most economic sense overall, but it is far from every case.

I am not interested in debating the whole topic of open source tools. 
You have your *opinion* and I have mine.  But your *facts* are not self
evident.  Feel free to use any tool you wish.  And that includes all the
open source FPGA tools... :)


> > stands to reason that the commercial tools must be better, why else
> > would they sell well enough to keep the companies in business?
> >
> 
> There is one, and only one, reason why any company stays in business - it is
> able to persuade enough people that its product is worth the money.  It may
> or may not be the case that the product actually *is* worth the money (to
> the given customer), but the immediate technical qualities of the product
> are only one part of the sale - there are many other relevant issues such as
> support and upgrade paths, and many irrelevant issues such as the
> persuasiveness of the salesfolk, the guilibitity of the purchasing people,
> and the quality of their flashy power-point slides and free pens.  I'm not
> trying to say that commercial compilers are not worth the money - I'm just
> pointing out the naivity of the "it costs more so it must be worth more"
> argument.

Again, you confuse the facts with your opinion.  If I pay for a tool,
then by definition it is worth the money I paid.  You can assume that
products are sold by Machiavellian sales people to Dilbert purchasers,
but in the real world everyone decides for themselves what is best.  End
of story.  

-- 

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

Tom Hawkins wrote:
> 
> rickman <spamgoeshere4@yahoo.com> wrote in message news:<40A01F19.EDA80D9F@yahoo.com>...
> > Tom Hawkins wrote:
> > >
> > > jon@beniston.com (Jon Beniston) wrote in message
> [snip]
> > > >
> > > > Why bother with Open Source stuff when various incarnations of
> > > > commerical tools are free to use?
> > >
> > > Because in the long term, an open-source tool chain can out perform a
> > > commercial tool.  Case in point: how many Unix/C developers compile
> > > with a commercial alternative to GCC?
> >
> > I have been evaluating tools for the ARM MCU and all my info says that
> > gcc is a less efficient tool than the commercial ones.  I guess it only
> > stands to reason that the commercial tools must be better, why else
> > would they sell well enough to keep the companies in business?
> 
> Certainly when a compiler targets a specific architecture, it can do a
> better job.  But in terms of the number of supported platforms and its
> decent code optimization, it's hard to beat GCC.

I am not going to debate the open source tool question.  I don't care
what you use.  But it is not a *fact* that GCC "out performs" commercial
tools.  


> > > Unfortunately, the track record of open-source FPGA implementation
> > > tools has been dismal.  Icarus is the only open-source tool I know
> > > that attempts Xilinx synthesis.  I believe it handles synthesis and
> > > technology mapping, but has little, if no optimization.
> >
> > I think John Jakson summed it up pretty well, open source FPGA tools are
> > not practical for a number of reasons that have been discussed here many
> > times before.  Some people seem to think that FPGA and ASIC tools are
> > the same as compiliers, but in reality they are very different.
> 
> Sure the algorithms are different, but there are many similarities:
> Like a compiler, most logic optimizations are NP-complete, therefore
> many of the same strategies can be applied.   Both a compiler and a
> synthesizer deal with dataflow analysis and resource allocation.
> Instruction reordering is similar to retiming.  Memory and execution
> timing tradeoffs are somewhat analogous to tradeoffs with area and
> performance.

There may be a few similarities, but you are only looking at technical
issues.  The real problem has to do with the way the FPGA devices keep
changing.  It is a major job for the vendors to keep up with their own
devices, it is not practical for open source tools to always be a
generation or two behind.  

If it is so practical, why has it not been done properly?  As an earlier
poster stated, some projects have been started, but none have produced
anything useful and many have just been dropped.  

> I agree the closer you get to implementation, the less pratical an
> open source tool becomes.  But where is the line drawn?  Clearly
> synthesis falls on our side.  Multi-chip partioning is certainly
> possible.  And with the capabilities of UCF, I would argue placement
> is a canidate as well.

So where are the tools?  Rather than rehash the issue here, google this
group and read what has already been posted over the last two years.  

-- 

Rick "rickman" Collins

rick.collins@XYarius.com
Ignore the reply address. To email me use the above address with the XY
removed.

Arius - A Signal Processing Solutions Company
Specializing in DSP and FPGA design      URL http://www.arius.com
4 King Ave                               301-682-7772 Voice
Frederick, MD 21701-3110                 301-682-7666 FAX

"Anna Acevedo" <acevedo@xilinx.com> escribi&#4294967295; en el mensaje
news:409FD261.DF96BD2C@xilinx.com...
> Digilent Inc  (www.digilentinc.com) has several low cost boards mainly
> used in academia but available to all. Soon they will be releasing a
> Spartan-3 board and in a couple of months a Virtex-II Pro board.
>
> Anna Acevedo - Manager
> Xilinx University Program
>


hi, is there any Virtex-II Pro (with only one PPC core embedded,
XC2VP4-5FG456C), the clock, config ROM, JTAG (maybe also USB) and some RAM
board out there?

cause i've seen some cards from Memec (there's one at about 400dollars with
EDK and 200 without it), but they dont have any RAM, and other boards have
too much peripheals that i dont need and then they have like 20 user I/O
pins available, cause they have keyboard, display, leds, parallel port, A/D,
D/A, video, and a lot of other stuff. If i wanted those peripheals later i
guess i'd get a daughter board, but in the beggining i'd like as much I/O as
possible

another one is the one for gameboy (i dont know it's name) it seems to have
everything i want, just the FPGA is not a Virtex-II Pro

any pointers will be appreciated




> Joel Hardy wrote:
>
> > Howdy all,
> >
> > I'm looking to get an FPGA development board sometime soon.  When I
> > was in college, I played with what was probably the Xess XSB-300E (fun
> > stuff: we made a  PCMCIA interface and plugged it into an iPaq... but
> > that's another story).  That's WAY outside of my price range (<
> > $200US), but the XSA-50 doesn't look too bad.  Any comments on that
> > one?  Also, I found this
> > (www.nuhorizons.com/products/xilinx/spartan3/development-board.html),
> > which looks considerably better to me, especially since my main
> > prospective project would benefit from a large amount of fast RAM.
> > Any advice?  Is there anything else in the sub-$200 I should look at?
> >
> > Also, what's the status of open source tools?  (I like to tinker on
> > the software end of things, too, and I have a bad case of PowerBook
> > envy, and I haven't seen any software available other than x86 and
> > big-iron UNIX).  Can they go from VHDL/Verilog all the way to
> > downloading the file to the chip?  I see that compilation (Icarus
> > verilog) and downloading (found it in this group's FAQ) work, but what
> > about place and route?  Icarus' docs say it's a no-go for this; is
> > there anything else, or must I use the Xilinx tools?  (I guess I have
> > a Xilinx bias -- it's all I've used, and I don't see much else for
> > cheap development boards.)
> >
> > Thanks for the help!
> > - Joel Hardy
>

"rickman" <spamgoeshere4@yahoo.com> wrote in message
news:40A0F55E.2E475D3B@yahoo.com...
> David Brown wrote:
> >
> > "rickman" <spamgoeshere4@yahoo.com> wrote in message
> > news:40A01F19.EDA80D9F@yahoo.com...
> > > Tom Hawkins wrote:
> > > >
> > > > jon@beniston.com (Jon Beniston) wrote in message
> > news:<e87b9ce8.0405070500.7c9c4bcc@posting.google.com>...
> > > > > > Also, what's the status of open source tools?  (I like to tinker
on
> > > > > > the software end of things, too, and I have a bad case of
PowerBook
> > > > > > envy, and I haven't seen any software available other than x86
and
> > > > > > big-iron UNIX).  Can they go from VHDL/Verilog all the way to
> > > > > > downloading the file to the chip?  I see that compilation
(Icarus
> > > > > > verilog) and downloading (found it in this group's FAQ) work,
but
> > what
> > > > > > about place and route?  Icarus' docs say it's a no-go for this;
is
> > > > > > there anything else, or must I use the Xilinx tools?  (I guess I
> > have
> > > > > > a Xilinx bias -- it's all I've used, and I don't see much else
for
> > > > > > cheap development boards.)
> > > > >
> > > > > Why bother with Open Source stuff when various incarnations of
> > > > > commerical tools are free to use?
> > > >
> > > > Because in the long term, an open-source tool chain can out perform
a
> > > > commercial tool.  Case in point: how many Unix/C developers compile
> > > > with a commercial alternative to GCC?
> > >
> > > I have been evaluating tools for the ARM MCU and all my info says that
> > > gcc is a less efficient tool than the commercial ones.  I guess it
only
> >
> > What do you mean by "less efficient" ?  There are big differences
between
> > run-time efficiency and develop-time efficiency, and the balance choosen
> > between these will vary according to user and project.  There is little
> > doubt that there are commercial compilers that produce smaller and
faster
> > code than gcc for many processors (including arm), but developer
efficiency
> > can be a serious issue too.  I write C code for five different
processors,
> > amongs other work - it is simply not economically feasible to buy
> > top-of-the-range commercial compilers for all of these, while gcc works
fine
> > for me.  I also find it in many ways to be a more advanced compiler than
> > some compilers I have used (especially in terms of compile-time error
> > checking).  There are plenty of cases where commercial compilers would
make
> > most economic sense overall, but it is far from every case.
>
> I am not interested in debating the whole topic of open source tools.
> You have your *opinion* and I have mine.  But your *facts* are not self

Much of what I wrote was not opinion, but fact - for the particular case in
question (i.e., my professional compiler usage).  As I said, the situation
will be different for different people - that's also fact.  As for
opinions - we all have plenty of them!

> evident.  Feel free to use any tool you wish.  And that includes all the
> open source FPGA tools... :)
>

Choice is an excellent thing - in the case of fpga development, I choose to
use an open-source tool (Confluence - once I get the hang of it :-), a
closed source paid-for tool (Quartus), a paid-for (but with some source)
component (Nios), and open-source compiler and debugger software
masquerading as closed-source (Gnu-pro compiler toolset for the Nios).  I
like to be able to pick and choose.

Incidently, isn't it interesting that both Altera and Xilinx have
standardised on gcc for the compiler for their soft-processor cores?

>
> > > stands to reason that the commercial tools must be better, why else
> > > would they sell well enough to keep the companies in business?
> > >
> >
> > There is one, and only one, reason why any company stays in business -
it is
> > able to persuade enough people that its product is worth the money.  It
may
> > or may not be the case that the product actually *is* worth the money
(to
> > the given customer), but the immediate technical qualities of the
product
> > are only one part of the sale - there are many other relevant issues
such as
> > support and upgrade paths, and many irrelevant issues such as the
> > persuasiveness of the salesfolk, the guilibitity of the purchasing
people,
> > and the quality of their flashy power-point slides and free pens.  I'm
not
> > trying to say that commercial compilers are not worth the money - I'm
just
> > pointing out the naivity of the "it costs more so it must be worth more"
> > argument.
>
> Again, you confuse the facts with your opinion.  If I pay for a tool,
> then by definition it is worth the money I paid.  You can assume that
> products are sold by Machiavellian sales people to Dilbert purchasers,
> but in the real world everyone decides for themselves what is best.  End
> of story.
>

By your definitions, you must be a perfect customer.  Are you suggesting
that you have never bought something that was not a good buy?  I'm sure you
make careful decisions about purchasing an expensive development tool - most
of us do that (at least, those of us who are fortunate enough to be able to
make such purchasing decisions ourselves).  But there is a long way between
the extremes of the mythical perfect customer and the mythical perfect
salesman and idiot customer combination.

"Jon Beniston" <jon@beniston.com> wrote in message
> Why bother with Open Source stuff when various incarnations of
> commerical tools are free to use? Check out for the Xilinx WebPack:
>

Because I'm using FreeBSD/KDE ?

I really sympathize with the desire to get tools that are "open" or at least
interoperable. I had the misfortune of participating in the "CAD Framework
Initiative" back in the early 90's when briefly, the major powers of CAD
believed they could agree to some open interfaces so that each others tools
could be mixed and matched. Some of that work lives on in things like EDIF
but for the most part vendors don't want to do anything that might make it
possible for someone to compete with them.

--Chuck