Fusion & CoreMP7 versus standard microcontroller

I know this is the embedded forum, but with these parts especially the
Fusion the design has more to do with embedded system design than
FPGA. I just came back from the embedded systems conference and
participated on these product seminars and discussion.
The question I can not answer should I change over to one of the
softcore FPGA based design concepts.

I try to stick with one tool and not too many different kinds.
Especially these SoC design tools have some learning curve. I am
familiar with the Quartus, but I do not want to pay on these license
fees anymore for the NIOSII. The idea is that once familiarized with a
core and tool I can reuse the whole concept from one design to the
next one meaning I look more into the long run. But the more I compare
the less I can not say which one is the right choice.

This is what I got out of the show. The Actel system with the Fusion
is a complete new approach where all is on 1 big FPGA flash chip. You
have analog capability for normal monitoring functions resolution and
speed is limited somewhere in the 100MHz, a standard ARM7 core with
the GNU & eclipse toolset integrated by Actel. The lab I did at the
ESC showed how powerful this chip really is, because you pretty much
can do it all with one chip -on top no license fees for anything of
the tools.
The LatticeMico32 has with the Wishbone interface a big pool of
additional cores. But I do hear that not all of them are too useful.
Because external bus interface to the core is possible for sure speed
is not the limit, but it is a pure FPGA so you will need again several
supporting ICs.

And to confuse the whole discussion more her is another observation
point. Most of the designs I did in the last years was around the ARM7
from NXP. With their new products the LPC22xx and LPC23xx you have
most of the time all what you need at a single chip with really low
cost. It is not often, that I reach the timing limits with these
parts.
I am not sure if I should do a switch and if this softcore processors
in FPGA will really gain most of the embedded controller market in the
next years to come?
Is it worth to dive into the new Actel's Fusion concept?

On 6 apr, 04:12, betteron...@gmail.com wrote:
> I know this is the embedded forum, but with these parts especially the
> Fusion the design has more to do with embedded system design than
> FPGA. I just came back from the embedded systems conference and
> participated on these product seminars and discussion.
> The question I can not answer should I change over to one of the
> softcore FPGA based design concepts.
>
> I try to stick with one tool and not too many different kinds.
> Especially these SoC design tools have some learning curve. I am
> familiar with the Quartus, but I do not want to pay on these license
> fees anymore for the NIOSII. The idea is that once familiarized with a
> core and tool I can reuse the whole concept from one design to the
> next one meaning I look more into the long run. But the more I compare
> the less I can not say which one is the right choice.
>
> This is what I got out of the show. The Actel system with the Fusion
> is a complete new approach where all is on 1 big FPGA flash chip. You
> have analog capability for normal monitoring functions resolution and
> speed is limited somewhere in the 100MHz, a standard ARM7 core with
> the GNU & eclipse toolset integrated by Actel. The lab I did at the
> ESC showed how powerful this chip really is, because you pretty much
> can do it all with one chip -on top no license fees for anything of
> the tools.
> The LatticeMico32 has with the Wishbone interface a big pool of
> additional cores. But I do hear that not all of them are too useful.
> Because external bus interface to the core is possible for sure speed
> is not the limit, but it is a pure FPGA so you will need again several
> supporting ICs.
>
> And to confuse the whole discussion more her is another observation
> point. Most of the designs I did in the last years was around the ARM7
> from NXP. With their new products the LPC22xx and LPC23xx you have
> most of the time all what you need at a single chip with really low
> cost. It is not often, that I reach the timing limits with these
> parts.
> I am not sure if I should do a switch and if this softcore processors
> in FPGA will really gain most of the embedded controller market in the
> next years to come?
> Is it worth to dive into the new Actel's Fusion concept?

Hi,

those topics came also into my eyes, that's why at this time I am
experimenting (starting level) with Actel ProASIC3 (Arm enabled) and
FUSION FPGA's.

At this time maybe speed is a bad point for those ACTEL FPGA's. But
they will invest in it in the future I think. FPGA softcores are
interesting when you have to do parallel processing or you have some
specific hardware (vhdl, verilog or systemC written).

the future for fpga's will be a mixture HW/SW design with tools for
systemC synthesis.


They are promoting the ACTEL FUSION at this time specially at motor
control applications.

Kind regards,
Vince

Vince wrote:
> On 6 apr, 04:12, betteron...@gmail.com wrote:
> > I know this is the embedded forum, but with these parts especially the
> > Fusion the design has more to do with embedded system design than
> > FPGA. I just came back from the embedded systems conference and
> > participated on these product seminars and discussion.
> > The question I can not answer should I change over to one of the
> > softcore FPGA based design concepts.
> >
> > I try to stick with one tool and not too many different kinds.
> > Especially these SoC design tools have some learning curve. I am
> > familiar with the Quartus, but I do not want to pay on these license
> > fees anymore for the NIOSII. The idea is that once familiarized with a
> > core and tool I can reuse the whole concept from one design to the
> > next one meaning I look more into the long run. But the more I compare
> > the less I can not say which one is the right choice.
> >
> > This is what I got out of the show. The Actel system with the Fusion
> > is a complete new approach where all is on 1 big FPGA flash chip.

Well, not quite. The code memory usually cannot fit on chip,
and even the SRAM data memory might not fit in the FPGA -
so a design might decide to use SDRAM for both,  but a 1 chip
solution
it is not.

> those topics came also into my eyes, that's why at this time I am
> experimenting (starting level) with Actel ProASIC3 (Arm enabled) and
> FUSION FPGA's.
>
> At this time maybe speed is a bad point for those ACTEL FPGA's. But
> they will invest in it in the future I think. FPGA softcores are
> interesting when you have to do parallel processing or you have some
> specific hardware (vhdl, verilog or systemC written).
>
> the future for fpga's will be a mixture HW/SW design with tools for
> systemC synthesis.
>
>
> They are promoting the ACTEL FUSION at this time specially at motor
> control applications.

These SoftCPUs will remain in the niche applications - as soon as a
sufficently large volume market appears, a focused Microcontroller
will take over.
Present uC are growing in FLASH size and Core power quite quickly, so
the window
for where FPGA CPUs make sense moves as well.
They make sense when you cannot get a uC large enough -so now you have
off-chip memory in both options. They also make sense for (multiple)
tiny cores, where code will fit into the FPGA memory - but if a Std uC
exists, then usually it is better to choose that, and use
a smaller/cheaper FPGA (or even CPLD)  to augment the peripherals, as
needed.
-jg