Reply by Everett M. Greene●December 3, 20052005-12-03
Dave Hansen <iddw@hotmail.com> writes:
> Jim Granville <no.spam@designtools.co.nz> wrote:
>
> [...]
> >stacking die - Q: if it's one package, but >1 die, what do you call it ?
>
> I've always heard it called MCM. Though to some that implies a
> "mini-pcb" implemented on a ceramic substrate.
The more things change, ...
Didn't connecting the bare die used to be called hybrid
packaging/packages (where "used to be" >= 20 years ago)?
Reply by Dave Hansen●December 2, 20052005-12-02
On Sat, 03 Dec 2005 07:38:28 +1300 in comp.arch.embedded, Jim
Granville <no.spam@designtools.co.nz> wrote:
[...]
>stacking die - Q: if it's one package, but >1 die, what do you call it ?
I've always heard it called MCM. Though to some that implies a
"mini-pcb" implemented on a ceramic substrate.
Regards,
-=Dave
--
Change is inevitable, progress is not.
Reply by Jim Granville●December 2, 20052005-12-02
Ian Bell wrote:
> Telenochek wrote:
>
>>I think I need to clarify/modify my original question:
>>
>>See, my personal opinion is that if you can find one IC that has all
>>the capabilities your system needs (maybe IC with 4 pins is enough or
>>maybe 100M reprogrammable gates is just enough, doesn't matter), then
>>you are better off selecting that IC rather than combining a bunch of
>>ICs on a PCB which combined give you the same capability as the one IC.
>>
>
>
> I think SoC is a bit like fusion - talked about a lot but not much real
> progress.
Correct - it's mostly a marketing phantom : and what they claim as
single chip, usually needs others to actually work...
"Oh, - but it's only one chip from _us_, so it is our One-Chip solution "
>
> I gave a series of seminars on SoC about 20 years ago and to be frank there
> has been little progress in the directions we anticipated. In practice is
> has all been overshadowed by the move from analogue to digital processing.
> Die sizes have increased and device packaging has only got worse.
Even Cell phones, where the biggest impetus for single chip appears, are
stacking die - Q: if it's one package, but >1 die, what do you call it ?
> Perhaps the only area where SoC has made any real progress in in really mall
> really low cost consumer items where COB is now commonplace.
Shouldn't that be called CoS, then :)
[LCDs do call it CoG]
-jg
Reply by Mark Borgerson●December 2, 20052005-12-02
In article <1133514258.209403.227380@g49g2000cwa.googlegroups.com>,
interpasha@hotmail.com says...
> Thank you all for your responses!
> There are some really nice things here that I didn't think about
> (like difficulty manufacturing precision analog circuitry, rf
> circuitry, fast memory & etc on the same chip).
>
> I think I need to clarify/modify my original question:
>
> See, my personal opinion is that if you can find one IC that has all
> the capabilities your system needs (maybe IC with 4 pins is enough or
> maybe 100M reprogrammable gates is just enough, doesn't matter), then
> you are better off selecting that IC rather than combining a bunch of
> ICs on a PCB which combined give you the same capability as the one IC.
>
> That's exactly what my definition of a SoC is: one chip has all the
> capabilities for your project.
> So whats the problem with generalized MCUs?
> The problem (as I see it) is that it doesn't matter how large the
> manufacturer product lineup selection is, most projects always require
> something extra that the chips are not going to have. So you get a
> bunch of chips with all the needed capabilities and you place them on
> your PCB, etc etc.
> But in the end (maybe distant future) aren't you much better off having
> all of these chips combined on a single IC?
>
> Which is why I was sort of thinking (in my original question) that
> future designers are much more likely to be building specialized ICs
> (SoC) for their projects rather than develop PCBs.
> Again, as I said earlier, if someone is selling a chip that has all the
> capabilities you need, then there is your SoC, you don't need to
> develop a large PCB, or develop your own IC/SoC :)
>
> Do you guys think that future designers will be transitioning to
> designing custom ICs for their projects rather than PCBs?
> Or using *highly customizable IC platforms* rather than develop
> PCBs?
I think a highly customizable IC will always have a higher current
requirement and cost more than two or three simpler ICs. It will
still be simpler to design the PC board for a 64-pin LQFP MCU
and another 64 pins of smaller parts that it will be to design a
board for a 208-pin BGA package----even if you are only using
half the pins.
If you put enough logic on the chip to cover 95% of the possible
uses, I would guess that 95% of the time the designer will be using
only half the logic, but paying for all of it.
>
> I sort of doubt that it will be FPGAs :
> Since the hardware configuration is stored on board, rather than being
> physically implemented, they are always going to be power hungry
> (unless the gates start consuming 10^(-16) Amps quiescent).
> And if you have a good design (that is out of the prototyping stage)
> with reasonable volume its probably better to have a *hardwired* but
> customizable/flexible chip.
>
>
Mark Borgerson
Reply by Ian Bell●December 2, 20052005-12-02
Telenochek wrote:
>
> I think I need to clarify/modify my original question:
>
> See, my personal opinion is that if you can find one IC that has all
> the capabilities your system needs (maybe IC with 4 pins is enough or
> maybe 100M reprogrammable gates is just enough, doesn't matter), then
> you are better off selecting that IC rather than combining a bunch of
> ICs on a PCB which combined give you the same capability as the one IC.
>
I think SoC is a bit like fusion - talked about a lot but not much real
progress.
I gave a series of seminars on SoC about 20 years ago and to be frank there
has been little progress in the directions we anticipated. In practice is
has all been overshadowed by the move from analogue to digital processing.
Die sizes have increased and device packaging has only got worse.
Perhaps the only area where SoC has made any real progress in in really mall
really low cost consumer items where COB is now commonplace.
Ian
Reply by Telenochek●December 2, 20052005-12-02
Thank you all for your responses!
There are some really nice things here that I didn't think about
(like difficulty manufacturing precision analog circuitry, rf
circuitry, fast memory & etc on the same chip).
I think I need to clarify/modify my original question:
See, my personal opinion is that if you can find one IC that has all
the capabilities your system needs (maybe IC with 4 pins is enough or
maybe 100M reprogrammable gates is just enough, doesn't matter), then
you are better off selecting that IC rather than combining a bunch of
ICs on a PCB which combined give you the same capability as the one IC.
That's exactly what my definition of a SoC is: one chip has all the
capabilities for your project.
So whats the problem with generalized MCUs?
The problem (as I see it) is that it doesn't matter how large the
manufacturer product lineup selection is, most projects always require
something extra that the chips are not going to have. So you get a
bunch of chips with all the needed capabilities and you place them on
your PCB, etc etc.
But in the end (maybe distant future) aren't you much better off having
all of these chips combined on a single IC?
Which is why I was sort of thinking (in my original question) that
future designers are much more likely to be building specialized ICs
(SoC) for their projects rather than develop PCBs.
Again, as I said earlier, if someone is selling a chip that has all the
capabilities you need, then there is your SoC, you don't need to
develop a large PCB, or develop your own IC/SoC :)
Do you guys think that future designers will be transitioning to
designing custom ICs for their projects rather than PCBs?
Or using *highly customizable IC platforms* rather than develop
PCBs?
I sort of doubt that it will be FPGAs :
Since the hardware configuration is stored on board, rather than being
physically implemented, they are always going to be power hungry
(unless the gates start consuming 10^(-16) Amps quiescent).
And if you have a good design (that is out of the prototyping stage)
with reasonable volume its probably better to have a *hardwired* but
customizable/flexible chip.
Reply by PeteS●November 30, 20052005-11-30
Jan Homuth wrote:
> For high volume, low cost applications the PIC and other chips will survive.
> Larger Applications (i.e. 16 or 32 bit bit MCU with lots of peripherals and
> additional circuitry) will likely have an alternative to be moved to FPGA,
> using soft IP cores.
>
> Have a look at www.alium.com -> Nexar.
>
> They are on this course.
> That is: use an FPGA, add a soft IP core (anything from 8051 core to
> PPC/ARM) and some soft IP library modules and you have a self designed
> processor. Target specific compilers are part of this Embedde Deveopment
> system so you do not have to do everything in VHDL.
>
> You do not have to be a FPGA or VHDL epert to work with this system.
>
> I guess it is just a matter of time when this concept will be accepted as a
> serious alternative to conventional MCU solutions.
>
>
> grtnx
> /janh
>
Unfortunately, the power consumption of FPGAs is still way too high in
many applications. I can get 32 bit processing with peripherals at a
fraction of the power of an FPGA with a soft core.
To a great extent, it's inherent in the design, using SRAM LUTs. Unless
and until this is addressed, FPGA SoC will probably not reach critical
mass.
Cheers
PeteS
Reply by Jan Homuth●November 30, 20052005-11-30
For high volume, low cost applications the PIC and other chips will survive.
Larger Applications (i.e. 16 or 32 bit bit MCU with lots of peripherals and
additional circuitry) will likely have an alternative to be moved to FPGA,
using soft IP cores.
Have a look at www.alium.com -> Nexar.
They are on this course.
That is: use an FPGA, add a soft IP core (anything from 8051 core to
PPC/ARM) and some soft IP library modules and you have a self designed
processor. Target specific compilers are part of this Embedde Deveopment
system so you do not have to do everything in VHDL.
You do not have to be a FPGA or VHDL epert to work with this system.
I guess it is just a matter of time when this concept will be accepted as a
serious alternative to conventional MCU solutions.
grtnx
/janh
"Telenochek" <interpasha@hotmail.com> schrieb im Newsbeitrag
news:1133201249.400996.183070@g43g2000cwa.googlegroups.com...
> I am wondering if the SoC (ARM/AMBA architecture) (where a
> whole system with upgradeable hardware modules/ IP cores can be stuffed
> inside a single chip) will make all kinds of generalized
> microcontrollers (like PIC) obsolete.
> When PIC microcontrollers are used, they often need external hardware
> to help them, DSP blocks cannot be integrated into the chip at will
> (its all up to Microchip, whatever they decide to include in a chip).
> Just seems to me like SoC will eventually replace every MCU based
> system, because of the processing power, and application-specific
> flexibility in hardware. And almost all systems can use extra
> processing power, capabilities & etc.
> Maybe a traffic light with video camera and remote alerts for
> speeders + array radar sensing of speeding cars & reporting their
> position via GPS. I'm not saying that developing a supercomputing
> traffic light is a very high priority task, just using it as an
> illustration of stuffing more capability into into a simple system.
>
> What would be the problem will be with replacing almost all MCU-based
> systems with SoC?
>
Reply by Guy Macon●November 29, 20052005-11-29
Telenochek wrote:
>Jim Granville wrote.
>
>>Price, and operational details, like battery life, Size etc.
>
>Actually, size of a SoC is much smaller than of a system composed
>of individual hardware pieces put together on a board.