Posted by rtstofer July 6, 2007
> > It's odd that you should bring up concurrency. I find that to be the
> > most fascinating part of FPGAs. You create a bunch of finite state
> > machines that are interlocked through semaphores and the machines all
> > run in parallel.
>
> Parallelism is quite natural in hardware.
>
> Leon
>

Sure but the part I found interesting is that sequential reading code
isn't sequential when under a clock:

Assume A = 3

Under a clock the statements

A = A + 1;
B = A;

results in a pipeline and after the clock A is equal to 4 (no surprise
here) but B is equal to 3, the previous value of A, not the current value.

The whole idea of reading code from top to bottom and the implied
sequence having something to do with execution sequence is right out
the window!

I really enjoy designing things with FPGAs. Even debugging is more
challenging.

Richard

An Engineer's Guide to the LPC2100 Series

Posted by Leon July 6, 2007
----- Original Message -----
From: "rtstofer"
To:
Sent: Friday, July 06, 2007 6:02 AM
Subject: [lpc2000] Re: Tools to help port from LPC2294 to FPGA??
>> Designing this kind of logic scares me a bit. As someone else said,
>> integrating the peripherals is likely to be the hard part. In the
>> world I am used to just getting known good silicon to work is often
>> challenge enough, let alone coding them too. It is a really different
>> mindset. After all of these years, I find the notion of working out
>> timing in an environment with *true* concurrency intimidating.
>>
>> -- Dave
>> It's odd that you should bring up concurrency. I find that to be the
> most fascinating part of FPGAs. You create a bunch of finite state
> machines that are interlocked through semaphores and the machines all
> run in parallel.

Parallelism is quite natural in hardware.

Leon
Posted by sig5...@hotmail.com July 6, 2007
I went through this months ago. Using softcore CPUs in FPGAs is becoming very common. After looking at the various options, I went with the Altera system. You are correct that connecting all of the perfs you need to run can be the time consuming debugging part. That is why it is very important to pick a system that has as many of these sub soft components already built and tested within their system.

I have used several ARM MCUs for designs, but not really interested in using that inside a FPGA. A softcore MCU needs to be optimized for FPGA synthesis. ARM was created for custom silicon compiling. Not at all the same thing. If you don't care whether your FPGA uses 40,000 cells or 20,000 cells it may not matter to you, but whether the part costs $15 or $30 usually matters to me. I generally have to pick a softcore CPU that fits the needs at hand. It may be as small as 800 logic cells, or as big as 8000. All depends on what you need, but don't forget FPGA cells are not free.

There are tiny 8 bit softcore CPUs like Picoblaze, Pacoblaze, NIOS-I, i8080. Then you have larger 32 bit CPUs such as NIOS-II in Altera. It has a ton of pre built perfs such as running DDR mem, etc. The Megacore functions of Altera provide a wide array of functionality with little debugging. Very good DSP compilers as well.

It is not that difficult using softcore CPUs in FPGAs. However it becomes more of an optimization issue. You don't want to waste the FPGA cells on a 32 bit CPU if a tiny 8 bit CPU will do the job. The smaller the logic cell requirement, the cheaper the FPGA.

Chris.
._,___
Posted by rtstofer July 6, 2007
> Designing this kind of logic scares me a bit. As someone else said,
> integrating the peripherals is likely to be the hard part. In the
> world I am used to just getting known good silicon to work is often
> challenge enough, let alone coding them too. It is a really different
> mindset. After all of these years, I find the notion of working out
> timing in an environment with *true* concurrency intimidating.
>
> -- Dave
>

It's odd that you should bring up concurrency. I find that to be the
most fascinating part of FPGAs. You create a bunch of finite state
machines that are interlocked through semaphores and the machines all
run in parallel.

FPGAs are a trip!

Richard
Posted by pork_u_pine2000 July 5, 2007
--- In l..., bob engle wrote:
>
> hello,
>
> OOPS
> that is actel
>
> bob engle
> embedded solutions
>
...

Yes, if you either have an investment in high level ARM7 code related
to this product, or are especially comfortable with the ARM
architecture, you should have a look at Actel's Core M7 (I believe...)
series of chips. They allow you to build on that architecture yet
leave you room to drop in other cores for peripherals as needed. They
come in a range of sizes, and I think that the cost of the ARM core is
included in the chip price, so no separate IP licensing is required.
Like everybody else, most of the major tools are 'free' to use up to
10^6 gates.

As I am most familiar with ARM cpu's but would like to learn more
about FPGA design, this is a route that has appealed to me for
some time now. I'm of the school that believes that this will be a
critical skill set in the near future.

Other issues related to incorporating FPGA's into a design don't go
away, though. Power consumption, time to market in an unfamiliar
environment, design validation issues, etc. I believe the risk on a
project is likely to increase greatly.

Designing this kind of logic scares me a bit. As someone else said,
integrating the peripherals is likely to be the hard part. In the
world I am used to just getting known good silicon to work is often
challenge enough, let alone coding them too. It is a really different
mindset. After all of these years, I find the notion of working out
timing in an environment with *true* concurrency intimidating.

-- Dave
Posted by bob engle July 2, 2007
hello,

OOPS
that is actel

bob engle
embedded solutions
Leonard Rohnert wrote:

> Dave (and Leon),
>
> Thanks for the reply.
>
> I'll answer your question (and the one from Leon) as much as I can. Our
> customer is looking for flexibility in the design (wants to deal with some
> specialzed sensors, and probablly move some functions from SW to HW). The
> customer is always right - and prices of FPGAs are getting smaller (we
> think
> we'' get other customers who want to do this in the near future: we
> think it
> is worth spending a bit of my time to get into this).
>
> Apologies to Tom and others if this seems "off topivc" (as I said first
> post, we are porting from an LPC design).
>
> regards
> LenRt
>
> On 7/2/07, pdqlogic > > wrote:
> >
> > Hi Leonard,
> >
> > You are right that the tools we supply will not help you port the
> > LPC2294 functionality to a FPGA. Instead they convert the output of
> > the Xilinx FPGA design tools & provide a convenient means for
> > downloading the design to the FPGA.
> >
> > I'm interested why you want to move to FPGA - I presume you have some
> > real-time requirement in your application that the LPC2294 can't
> > handle ? It sure is nice to have the flexibility to implement parts
> > of your application in programmable hardware rather than trying to
> > squeeze every clock cycle from your CPU :)
> >
> > ARM have produced a core which is designed specifically for FPGA &
> > may be of interest to you. Take a look at
> http://www.arm.com/products/
> > CPUs/ARM_Cortex-M1.html
> >
> > Regards
> > Dave
> >
> > --- In l...
> , "Leonard
> > Rohnert"
> > wrote:
> > >
> > > I've been asked to find the costs of swapping one of our LPC2294
> > prototypes
> > > onto FPGA. If the move gets the OK, this is going to be a complete
> > SoC
> > > design (with a soft core). I've done various ARM7 (and PIC)
> > designs but
> > > this will be my first FPGA project.
> > >
> > > In the archives for this group there is some discussion about some
> > free
> > > tools from pdqlogic http://www.pdqlogic.com/index.html.
> This
> > company has a
> > > neat board but I don't think the tools will help do the porting I
> > need to
> > > do (could be wrong?).
> > >
> > > I couldn't find anything else in this archive : - surely someone
> > else has
> > > done this?
> > >
> > > I've googled RapidiTTy (www.tte-systems.com). This seems to work
> > with xilinx
> > > FPGAs (only?), and comes with some kind of MIPS core in the
> > package. The
> > > www site says you can swap code between ARM and FPGA. If you've
> > tried this
> > > stuff, I'd appreciate any feedback (and costs - I couldn't get
> > prices on
> > > the tte site, and their sales guys don't seem to reply to mail on
> > > Sundays...).
> > >
> > > Thanks
> > > LenRt.
> > >
> > >
> > >
> > >
> >
> >
> >
>
>
Posted by bob engle July 2, 2007
hello,

atmel has both the arm7 and cortex cores for its
flash based fpga. the licence for the core does not
add to the cost of the part.

bob engle
embedded solutions
Leonard Rohnert wrote:

> Dave (and Leon),
>
> Thanks for the reply.
>
> I'll answer your question (and the one from Leon) as much as I can. Our
> customer is looking for flexibility in the design (wants to deal with some
> specialzed sensors, and probablly move some functions from SW to HW). The
> customer is always right - and prices of FPGAs are getting smaller (we
> think
> we'' get other customers who want to do this in the near future: we
> think it
> is worth spending a bit of my time to get into this).
>
> Apologies to Tom and others if this seems "off topivc" (as I said first
> post, we are porting from an LPC design).
>
> regards
> LenRt
>
> On 7/2/07, pdqlogic > > wrote:
> >
> > Hi Leonard,
> >
> > You are right that the tools we supply will not help you port the
> > LPC2294 functionality to a FPGA. Instead they convert the output of
> > the Xilinx FPGA design tools & provide a convenient means for
> > downloading the design to the FPGA.
> >
> > I'm interested why you want to move to FPGA - I presume you have some
> > real-time requirement in your application that the LPC2294 can't
> > handle ? It sure is nice to have the flexibility to implement parts
> > of your application in programmable hardware rather than trying to
> > squeeze every clock cycle from your CPU :)
> >
> > ARM have produced a core which is designed specifically for FPGA &
> > may be of interest to you. Take a look at
> http://www.arm.com/products/
> > CPUs/ARM_Cortex-M1.html
> >
> > Regards
> > Dave
> >
> > --- In l...
> , "Leonard
> > Rohnert"
> > wrote:
> > >
> > > I've been asked to find the costs of swapping one of our LPC2294
> > prototypes
> > > onto FPGA. If the move gets the OK, this is going to be a complete
> > SoC
> > > design (with a soft core). I've done various ARM7 (and PIC)
> > designs but
> > > this will be my first FPGA project.
> > >
> > > In the archives for this group there is some discussion about some
> > free
> > > tools from pdqlogic http://www.pdqlogic.com/index.html.
> This
> > company has a
> > > neat board but I don't think the tools will help do the porting I
> > need to
> > > do (could be wrong?).
> > >
> > > I couldn't find anything else in this archive : - surely someone
> > else has
> > > done this?
> > >
> > > I've googled RapidiTTy (www.tte-systems.com). This seems to work
> > with xilinx
> > > FPGAs (only?), and comes with some kind of MIPS core in the
> > package. The
> > > www site says you can swap code between ARM and FPGA. If you've
> > tried this
> > > stuff, I'd appreciate any feedback (and costs - I couldn't get
> > prices on
> > > the tte site, and their sales guys don't seem to reply to mail on
> > > Sundays...).
> > >
> > > Thanks
> > > LenRt.
> > >
> > >
> > >
> > >
> >
> >
> >
>
>
Posted by Devaraj Ayavoo July 2, 2007
Hi all,

I have actually used the RapidiTTy toolset for some of my embedded development, so will try to clarify some of the discussions on core swapping.

Basically, RapidiTTy provides 2 softcores (PH and TTE1) which are based on the MIPS architecture. These softcores come with their own header files that allows me to interface the various peripherals like UART and IO pins through my C program. So when I use RapidiTTy to generate a UART example for ARM7 (LPC 2294), I can then convert the project to also work on the softcore (and vice versa). The conversions are limited to the examples that are supported in the tool. I doubt the tool can actually support other cores available in the market.

However, the feature that I found most interesting was that by using the existing JTAG interface and cables, I could program, debug and even obtain timing analysis (execution time and task period) from the code runing on the develpoment board. It saves me the hassle of using LabView to make my measurements. The ARM debugging interface on RapidiTTy is quite straightforward to use and it saves me the hassle of trying to set up Open OCD, GDB etc.

Leonard, it might be worth trying out the 30-day evaluation version of RapidiTTy (http://www.tte-systems.com/downloads.php) to see if this tool would fit your needs. Its free :-)

Leonard Rohnert wrote:
Dave (and Leon),

Thanks for the reply.

I'll answer your question (and the one from Leon) as much as I can. Our
customer is looking for flexibility in the design (wants to deal with some
specialzed sensors, and probablly move some functions from SW to HW). The
customer is always right - and prices of FPGAs are getting smaller (we think
we'' get other customers who want to do this in the near future: we think it
is worth spending a bit of my time to get into this).

Apologies to Tom and others if this seems "off topivc" (as I said first
post, we are porting from an LPC design).

regards
LenRt

On 7/2/07, pdqlogic wrote:
>
> Hi Leonard,
>
> You are right that the tools we supply will not help you port the
> LPC2294 functionality to a FPGA. Instead they convert the output of
> the Xilinx FPGA design tools & provide a convenient means for
> downloading the design to the FPGA.
>
> I'm interested why you want to move to FPGA - I presume you have some
> real-time requirement in your application that the LPC2294 can't
> handle ? It sure is nice to have the flexibility to implement parts
> of your application in programmable hardware rather than trying to
> squeeze every clock cycle from your CPU :)
>
> ARM have produced a core which is designed specifically for FPGA &
> may be of interest to you. Take a look at http://www.arm.com/products/
> CPUs/ARM_Cortex-M1.html
>
> Regards
> Dave
>
> --- In l... , "Leonard
> Rohnert"
> wrote:
> >
> > I've been asked to find the costs of swapping one of our LPC2294
> prototypes
> > onto FPGA. If the move gets the OK, this is going to be a complete
> SoC
> > design (with a soft core). I've done various ARM7 (and PIC)
> designs but
> > this will be my first FPGA project.
> >
> > In the archives for this group there is some discussion about some
> free
> > tools from pdqlogic http://www.pdqlogic.com/index.html. This
> company has a
> > neat board but I don't think the tools will help do the porting I
> need to
> > do (could be wrong?).
> >
> > I couldn't find anything else in this archive : - surely someone
> else has
> > done this?
> >
> > I've googled RapidiTTy (www.tte-systems.com). This seems to work
> with xilinx
> > FPGAs (only?), and comes with some kind of MIPS core in the
> package. The
> > www site says you can swap code between ARM and FPGA. If you've
> tried this
> > stuff, I'd appreciate any feedback (and costs - I couldn't get
> prices on
> > the tte site, and their sales guys don't seem to reply to mail on
> > Sundays...).
> >
> > Thanks
> > LenRt.
> >
> >
> >
>


---------------------------------
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for your freeaccount today.
Posted by Leonard Rohnert July 2, 2007
Dave (and Leon),

Thanks for the reply.

I'll answer your question (and the one from Leon) as much as I can. Our
customer is looking for flexibility in the design (wants to deal with some
specialzed sensors, and probablly move some functions from SW to HW). The
customer is always right - and prices of FPGAs are getting smaller (we think
we'' get other customers who want to do this in the near future: we think it
is worth spending a bit of my time to get into this).

Apologies to Tom and others if this seems "off topivc" (as I said first
post, we are porting from an LPC design).

regards
LenRt
On 7/2/07, pdqlogic wrote:
>
> Hi Leonard,
>
> You are right that the tools we supply will not help you port the
> LPC2294 functionality to a FPGA. Instead they convert the output of
> the Xilinx FPGA design tools & provide a convenient means for
> downloading the design to the FPGA.
>
> I'm interested why you want to move to FPGA - I presume you have some
> real-time requirement in your application that the LPC2294 can't
> handle ? It sure is nice to have the flexibility to implement parts
> of your application in programmable hardware rather than trying to
> squeeze every clock cycle from your CPU :)
>
> ARM have produced a core which is designed specifically for FPGA &
> may be of interest to you. Take a look at http://www.arm.com/products/
> CPUs/ARM_Cortex-M1.html
>
> Regards
> Dave
>
> --- In l... , "Leonard
> Rohnert"
> wrote:
> >
> > I've been asked to find the costs of swapping one of our LPC2294
> prototypes
> > onto FPGA. If the move gets the OK, this is going to be a complete
> SoC
> > design (with a soft core). I've done various ARM7 (and PIC)
> designs but
> > this will be my first FPGA project.
> >
> > In the archives for this group there is some discussion about some
> free
> > tools from pdqlogic http://www.pdqlogic.com/index.html. This
> company has a
> > neat board but I don't think the tools will help do the porting I
> need to
> > do (could be wrong?).
> >
> > I couldn't find anything else in this archive : - surely someone
> else has
> > done this?
> >
> > I've googled RapidiTTy (www.tte-systems.com). This seems to work
> with xilinx
> > FPGAs (only?), and comes with some kind of MIPS core in the
> package. The
> > www site says you can swap code between ARM and FPGA. If you've
> tried this
> > stuff, I'd appreciate any feedback (and costs - I couldn't get
> prices on
> > the tte site, and their sales guys don't seem to reply to mail on
> > Sundays...).
> >
> > Thanks
> > LenRt.
> >
> >
> >
> >
>
Posted by tvragp2004 July 2, 2007
Hi Leonard,

1. you could look at the Altera NIOSII board. I am told they are good
although i have never tried them.

2. I did use this: MANIK 32 Bit on Xilinx available free from
http://www.niktech.com/index.htm.

I fully agree with Leon Heller,in fact i feel, you will spend more
time implementing a CPU soft core on FPGA than using a uP IC like LPC
or TMS470.

regards
ananth

--- In l..., "Leonard Rohnert" wrote:
>
> I've been asked to find the costs of swapping one of our LPC2294
prototypes
> onto FPGA. If the move gets the OK, this is going to be a complete SoC
> design (with a soft core). I've done various ARM7 (and PIC)
designs but
> this will be my first FPGA project.
>
> In the archives for this group there is some discussion about some free
> tools from pdqlogic http://www.pdqlogic.com/index.html. This
company has a
> neat board but I don't think the tools will help do the porting I
need to
> do (could be wrong?).
>
> I couldn't find anything else in this archive : - surely someone
else has
> done this?
>
> I've googled RapidiTTy (www.tte-systems.com). This seems to work
with xilinx
> FPGAs (only?), and comes with some kind of MIPS core in the package.
The
> www site says you can swap code between ARM and FPGA. If you've
tried this
> stuff, I'd appreciate any feedback (and costs - I couldn't get prices on
> the tte site, and their sales guys don't seem to reply to mail on
> Sundays...).
>
> Thanks
> LenRt.
>
>