Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)? - Page 2

Do you have a question? Post it now! No Registration Necessary

Translate This Thread From English to

Threaded View
Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
|> opencores is up for sale, and there is no one who cares to "take it."
|>
|> What does that say?

That it is not profitable to give free hosting, especially not to a
high-traffic site?

I don't see how you relate the *site* opencores.org to the open cores
on that site. The site might change to closedip.com through the sale,
but the open cores (as in GPL, BSD or whatever "open" license) wouldn't
vanish and most likely reappear e.g. on Sourceforge or somewhere else.

Rainer

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Rainer,

The way I read the articles is different from how you read them(?):

http://www.eetimes.com/rss/showArticle.jhtml?articleID20%0000302&cid=RSSfeed_eetimes_newsRSS

http://edageek.com/2007/06/26/opencores-for-sale /

Sounds like they will "go away" unless they find someone to buy them.

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
On a sunny day (Thu, 12 Jul 2007 08:21:15 -0700) it happened austin

Quoted text here. Click to load it

So, the only 'value' is the Verilog and VHDL code, not so much the 'open cores'
label.
The correct place where to put it is on a large archive such as
ftp://sunsite.unc.edu
if it is released under GPL, a simple email to the maintainer would probably
solve that.
Lost of people make money of open source Redhat, Suse, etc...
If they try to cash in on 'open cores' as a label it will be a really
disappointing
experience.
It only costs money to run a server without irritating ads:-)


Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Jan,

Seems there are other announcements in this area, too:

http://www.eetimes.com/rss/showArticle.jhtml?articleID20%1000153&cid=RSSfeed_eetimes_newsRSS

I would absolutely just love it if open source IP for FPGAs could be a
vibrant and healthy business!  But it seems that there is enough
difference between the open software world, and the open hardware world
to make it not to be.

So, the question is:  what is needed to make the model work?

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
On a sunny day (Thu, 12 Jul 2007 10:48:18 -0700) it happened austin

Quoted text here. Click to load it

FPGA's in DIL 48 pin packages available in local and online shops at low
prices.



Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?

Quoted text here. Click to load it

That nobody is interested in the brand 'Opencores'. But according to
the magazine articles you refer to, they attract a lot of engineers
working with programmable logic and ASIC. SO the content of the
website is attractive enough. It would be a great buy for an FPGA
vendor to push their own products.

--
Reply to nico@nctdevpuntnl (punt=.)
Bedrijven en winkels vindt U op www.adresboekje.nl

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?

Quoted text here. Click to load it

Certainly it would look like a good fit for Lattice to host this.
Then they can develop an eco system around their own OpenCores,
whilst allowing others space to live as well.

I'd imagine the upper echelons of Xilinx would gag on their
own legal-red-tape, even thinking about this idea  ;)

Companies _can_ have too many lawyers (but never too many engineers :)

-jg



Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Jim,

It isn't the lawyers that would have a problem with this:  it is the
regular rank and file engineers.  We understand that most of what is
there is (sorry) very poor quality, and that would reflect badly on
Xilinx.  Very badly indeed.

No company would put there name on such a mess:  it is too much of a
liability.  Would you tell your boss that all your IP came from such a
source?  And expect to stay employed?

Think of it:  do you ever get an email that says "I used that code,
understood it completely without even reading anything, and it worked,
and I am sooo happy."

Yes, sure.

Rather, you get (tens of) thousands of complaints....just read the
newsgroup!  Problems that are non-problems, problems that are easily
solved by reading the data sheet, problems solved by a few simple
experiments, it matters not:  most people see a problem, and call up the
help desk.  If you are their biggest supplier, you get blamed for
everything else (and we do often solve problems that do not even belong
to us).

"Your chip is breaking my power supply" (* a real complaint!).  We
analyzed and fixed their power supply for them.

Don't get me wrong:  I love getting tens of thousands of web cases.  It
allows us to measure how we improve, and measure how satisfied our
customer's are.  With 350,000 seats of software and tens of thousands of
engineers designing every hour of every day, the web cases are a
"heatbeat" and tell us immediately how well we are doing what we
promised to do.

And this is for products, code, manuals, and services where we spend a
great deal of time on, and pay a lot of attention to quality.

I was told many years ago by the "Bell System" that one half of all
returned boards had no problems (the problem was in the imagination of
the technician).  This holds as true today as it did then:  most (more
than 1/2) of all reported problems are not really a problem (in a well
run organization with a great product).  Yes, all reported issues are
taken seriously, and followed through on.  Every time.  That is how you
get better.

Imagine what would happen with a cornucopia of stuff of totally unknown
quality, and inadequate documentation!  In other words, real garbage?

Lawyers?  That is the least of our concerns:  how about staying in business?

You can't impress anyone (for long) with having more garbage than your
competitor.

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32,Leon)?
  I'll take this as a 'no' then ? :)

  The topic is not about buying the cores, but about providing a
hosting/forum shell, where such cores can "percolate away",
and I'm sure a company with the right culture could make this work.

  I did not expect Xilinx to embrace the idea, but it does seem
a better fit for Lattice.

-jg


austin wrote:

Quoted text here. Click to load it


Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32,Leon)?
Jim,

Hosting something, in effect, anonymously, is OK.  The problem is what
if someone finds out you are the host, and then makes life "difficult?"
  "There be trolls out there ..."

Hosting 'open cores' would be something that would need to have benefit.
  To Xilinx.

Imagine trying to manage an open cores site...

No, Jim, I don't think we will host any open cores.  We might not even
continue to host what we do now, in the future (depends on the trolls).

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Quoted text here. Click to load it


All,

I don't know if we (whether user or contributor to opencores
project) and You (FPGA/ASIC company) can live better with
or without opencores site.

I only hope, if really the opencores site will go off, the
contributors will put their project to other open-source sites
(i.e. sourceforge but any other will be ok).

Just a little remind... once upon a time there were few
commercial C/C++ compilers. Today many commercial compilers
doesn't exist more and many firms use gnu C/C++  and gnu
assembler/linker [maybe/surely You too (FPGA companies) use it].

GNU C/C++ seems to be a clear example of good open source
product that has more chance to have a very long life.

Is there any other project that can boast 22 year of life
without being obsolete ? Yes maybe yes but in IT is a very
good target both for commercial and free projects!

How this happen ? I don't know... maybe/surely company having
interest in it give him money because so they (the company)
can use it and so they can build product to sell.

And yes, there are a lot of opensource projects that disappear
after few monthis.

I don't want to state all the cores/software should be
free/opensurce, but the I think that we all needs both kind of
cores/software (both open and commercial).

Maybe in the future companies that today strongly defend the
commercial licenses could make more money because more users
of open cores are available to buy their product or because
using open source they build more product to sell.

And Maybe in the future peoples that strongly defend open source
could decide to buy commercial product because had good experience
with other related free products.

Then the problem is not whether are better commercial or open
sourced product but if are better good or bad product (both
opensource and commercial).

Sandro

P.S. Sorry for my poor english. I hope concepts were clear anyway


Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
|------------------------------------------------------------------------|
|"[..]                                                                   |
|                                                                        |
|Additionally, a massive amount of work goes into testing and            |
|re-optimizing every core when the technology node changes.  Who will pay|
|for that?  Who will warrant or guarantee operation?  Who supplies the   |
|test bench vectors to verify proper operation?                          |
|                                                                        |
|Austin"                                                                 |
|------------------------------------------------------------------------|

Hello,

Testing does not guarantee correctness.

Regards,
Colin Paul Gloster

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
But it helps....

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
|------------------------------------------------------------------------|
|"[..]                                                                   |
|                                                                        |
|Additionally, a massive amount of work goes into testing and            |
|re-optimizing every core when the technology node changes.  Who will pay|
|for that?  Who will warrant or guarantee operation?  Who supplies the   |
|test bench vectors to verify proper operation?                          |
|                                                                        |
|Austin"                                                                 |
|------------------------------------------------------------------------|

with a malformed From header (sorry, that was my fault):
|---------------------------------------|
|Testing does not guarantee correctness.|
|---------------------------------------|

|-----------------|
|"But it helps....|
|                 |
|Austin"          |
|-----------------|

Agreed.

C. P. G.

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Thanks everyone, for the replies. My question was vague and poorly
worded because I don't have a good grasp on what constitutes a
"FPGA CPU-core"and an "ASIC CPU-core."  Microblaze and Nios-II
obviously are architected around the unique device-features of their
respective vendors.  But when it comes to the Opensource offerrings,
I didn't see the same trend (maybe I didn't look close enough?)

Picoblaze -- is that safe for ASIC-use?  I was under the impression
Xilinx's license-agreement limits it to Xilinx devices only.



Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Quoted text here. Click to load it

I'm not sure about the Picoblaze licensing.
Actually, there are at least 2 distinct version of Picoblaze available
from Xilinx site. There is a Virtex-2/Spartan-3 optimized version
(KCPSM3) that involves Xilinx-specific components (buffers, registers
etc), and another version with some differences in the architecture
(slightly older instruction set, 8 registers instead of 16) originally
devised for CPLDs.
However this description is completely portable across different FPGA
vendors. For example i have ported this one to XC3S200 with good
results (around 170 slices, maybe less i don't recall the exact
results).

It would be meaningful for Xilinx to license this more obsolete
version in GPL or LGPL sense. Anyone aware of the exact licensing
issues with the two different PicoBlaze versions, please jump in.

Nikolaos Kavvadias




Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Quoted text here. Click to load it

Since there seems to be some confusion on the PicoBlaze license,
the btter choices might be PacoBlaze,
or the quite similar LatticeMico8 ?

-jg


Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Jim,

I was corrected by Ken Chapman, that to download PicoBlaze, one has to
acknowledge they have read and agree to the "use restrictions."

Peter and I had a chat about this subject:  any IP that is specific to a
device (that would be Lattice, Altera, or Xilinx) is optimized for their
device, and would be suitably poor in any other technology.  So, if you
want something that is technology agnostic, you would end up buying
something from a 3rd party, and paying them (or your own team) to make
different versions of it that are all code and cycle identical, and
technology independent (? I am presuming this is possible:  it may not be!).

So, that is why if I designing an ASIC I buy an 'XYZ' (insert ARM, PPC,
MIPS, or whatever you like here) core:  I know what I am getting, I can
get it for 130nm, 90nm, 65nm, etc.;  I can emulate it in an FPGA (if I
have to); I can ask questions and get answers.

If I want to get the most for my money, I decide whose chip I am going
to use, and I use their tools and IP.  I "hope" that my c code can be
compiled and run on another vendor if I decide I must switch to another
vendor (for whatever reason).

The one time I had a major project of many uP's on many pcb's, and we
changed from Intel to Motorola (gasp!), it was not trivial to port all
the code (written in c) from one to the other platform....I wouldn't
want to do it today, either.  But at least, it was possible, and it
could be done (and we did it successfully).

As for support of old code:

http://www.xilinx.com/ise/logic_design_prod/classics.htm

Xilinx offers the last best and debugged version of each code build, for
free, to customers who need to maintain an old design.

Of course, we can not provide the old Windoze version, but a trip to the
Saturday Flea Market at Foothill (or equivalent) will get you all the
old Microsquat stuff you want.  As well, we no longer have rights to
distribute nor support certain old schematic tools, or simulators, but
as customers, you have the rights to use the old versions to fix old
stuff (they have archives, too).

So, be my guest, pick a processor: pick the one that you will get the
most use from, be the most efficient to implement in the technology
targeted, make the best use of code already written, and will be
maintainable for the lifetime of your product lines.

Austin

Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Quoted text here. Click to load it

So what i understand here is that If i want something technology
agnostic, i will get it from opencores or develop it myself (my budget
is limited). Note that there is usually only a 2%-10% of the overall
code that needs to be differentiated across vendors (i'm referring to
soft core i'm writing which works for a 2-processor system on an X-
FPGA but would work with little work on an A-, L- etc one).

I've written a 32-bit 5-stage pipeline RISC core (no coprocessor for
exceptions accounted) that maps its brains out on block RAM for all
storage (except pipeline registers) and only takes up 35% of the
XC3S200 slices and runs at decent speed (~50MHz). It is a great speed
given that most parts are written at almost behavioral level (the ALU,
PC update logic, branch unit, load-store unit, all this stuff). And i
also don't remember the source code line count but should be quite
small.

Anyway, i stand my ground, the KCPSM2 is beautiful on the XC3S200, 170
to 96 slices make no big difference to me.
And it is device-agnostic. Still i can't understand why people are
restricted to use X reference designs that are provided AS-IS.

Quoted text here. Click to load it

This is different to changing implementation medium for a given soft
core. This is not a proper example if you refer to what i understand.

Nikolaos Kavvadias


Re: Anyone really use those opensource CPUs (OR1K, Lattice Mico32, Leon)?
Noah,

You should sell your agnostic IP.

I am sure Ken is flattered to hear how "elegant" you think his soft
processor is.

My example was just to illustrate that changing your "standard uP" is a
non-trivial, and expensive thing to do.

Austin

Site Timeline