> The ARM9 should cover the TriCore performance range, but is normally not
> single chip.
> The AT91SAM9261 will run at 180 MHz and has the V5 DSP instructions +
> 160 kB SRAM. Can load from an SO-8 size dataflash so it is close to single
> chip.
> Should only be a couple of months to the first internal sample.
Sticking with the ARM core, this today from STm is notable
[Not quite a single chip, but almost..]
http://www.st.com/stonline/press/news/year2005/p1416h.htm
330MHz and 2MBytes of DRAM, plus smallish FPGA - and the
semantics games start about a first with 'embedded FPGA fabric' :)
which hinges on if you call it ARM+FPGA or FPGA+ARM.
It does seem a better mix of CPU/memory/ProgLogic than
earlier offerings.
[Seems Xilinx buying Triscend was not enough to prevent
ARM+FPGA from hitting the streets..]
-jg
Reply by Chris Hills●March 9, 20052005-03-09
In article <4DBWd.90$Y7.45@read3.inet.fi>, Tauno Voipio <tauno.voipio@ik
i.fi.NOSPAM.invalid> writes
>Mayank Kaushik wrote:
>> Okay...Lets put it this way..which ARM-based microcontroller is the
>> easiest to work with, the criterea for "easy" range from the way to
>> access internal components, example code availability, and the
>> existance of user groups etc.
>>
>
>Maybe Philips LPC2000 or Atmel AT91.
Or the ST or the...
Which ever you can find the right peripherals that has a nice dev board
and tools.
/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
\/\/\/\/\ Chris Hills Staffs England /\/\/\/\/\
/\/\/ chris@phaedsys.org www.phaedsys.org \/\/
\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/
Reply by Ulf Samuelsson●March 9, 20052005-03-09
>> Is that an OK from you to switch from the 8051 to the AVR then, he
>> he! :-)
>
> Heresy! I did say an increase in performance ;)
>
> Besides, who needs to change now, even Atmel has 1 clock AT89LP
> devices ?
An engineering manager at a local company once said:
"It is good that it is better, but it would be better if it was good"
but I dont mind making you a little happy!
--
Best Regards,
Ulf Samuelsson
ulf@a-t-m-e-l.com
This message is intended to be my own personal view and it
may or may not be shared by my employer Atmel Nordic AB
Reply by Jim Granville●March 9, 20052005-03-09
Ulf Samuelsson wrote:
>>..... It was a side comment, and I had assumed everyone would know the
>>Tricore was not binary compatible with ARM, perhaps unwisely ?.
>>
>> The point really was, that while the smaller ARM uC vendors argue
>>over whose 30/50/60MHz device really is fastest in whatever mode,
>>there is another stratum of single chip 32 bit embedded
>>microcontrollers emerging out there.
>> Will they replace ARM ? - of course not, but to someone who really
>>needs _performance_, the opcodes used really do not matter to them.
>
>
> Is that an OK from you to switch from the 8051 to the AVR then, he he! :-)
Heresy! I did say an increase in performance ;)
Besides, who needs to change now, even Atmel has 1 clock AT89LP devices ?
>
> The ARM9 should cover the TriCore performance range, but is normally not
> single chip.
> The AT91SAM9261 will run at 180 MHz and has the V5 DSP instructions +
> 160 kB SRAM. Can load from an SO-8 size dataflash so it is close to single
> chip.
> Should only be a couple of months to the first internal sample.
Nice part, but a nudge short of 2MBytes of on chip ECC flash.
-jg
Reply by Ulf Samuelsson●March 9, 20052005-03-09
> ..... It was a side comment, and I had assumed everyone would know the
> Tricore was not binary compatible with ARM, perhaps unwisely ?.
>
> The point really was, that while the smaller ARM uC vendors argue
> over whose 30/50/60MHz device really is fastest in whatever mode,
> there is another stratum of single chip 32 bit embedded
> microcontrollers emerging out there.
> Will they replace ARM ? - of course not, but to someone who really
> needs _performance_, the opcodes used really do not matter to them.
Is that an OK from you to switch from the 8051 to the AVR then, he he! :-)
The ARM9 should cover the TriCore performance range, but is normally not
single chip.
The AT91SAM9261 will run at 180 MHz and has the V5 DSP instructions +
160 kB SRAM. Can load from an SO-8 size dataflash so it is close to single
chip.
Should only be a couple of months to the first internal sample.
>
> -jg
--
Best Regards,
Ulf Samuelsson
ulf@a-t-m-e-l.com
This message is intended to be my own personal view and it
may or may not be shared by my employer Atmel Nordic AB
Reply by Jim Granville●March 9, 20052005-03-09
Ulf Samuelsson wrote:
>>... and if a 30-50MHz ARM core, with 32-512K of Code, and 64KRAM is
>>too wimpy for your embedded widget, you can always pop in the new
>>TC1796 TriCore from Infineon : 150 MHz DSP/FPU, 2MBytes/ECC FLASH,
>>192KB SRAM, Automotive Spec, - includes a 32 bit Co-processor for
>>interrupt/DMA style tasks...
>>
>>-jg
>
>
> I think any of the wimpy ARM7 parts will execute *ARM* code faster than the
> Tricore.
> Read the subject line...
..... It was a side comment, and I had assumed everyone would know the
Tricore was not binary compatible with ARM, perhaps unwisely ?.
The point really was, that while the smaller ARM uC vendors argue
over whose 30/50/60MHz device really is fastest in whatever mode,
there is another stratum of single chip 32 bit embedded
microcontrollers emerging out there.
Will they replace ARM ? - of course not, but to someone who really
needs _performance_, the opcodes used really do not matter to them.
-jg
Reply by Ulf Samuelsson●March 8, 20052005-03-08
> As to the availability, somebody marked the devices of the
> LPC2130-series not available that have been acknowledged in the LPC2000
> newsgroup being available for almost 3 months. LPC2138, LPC2132 and
> LPC2131. afaik the LPC2134 and LPC2136 are not released to the market
> yet but given pin compatibility with the LPC2138, everybody can start
> the design.
My mistake, the timing was actually referring to the LPC214x.
Realized that after sending the entry.
> Atmels SAM7 has claimed to have better peripherals, it seems there has
> been some oversight to provide enough pins to get them out. Comparing
> the 64-pin devices gives me almost 10 extra I/O pins on the Philips
> devices. There is a point though in making the JTAG pins dedicated but,
> who ever uses Philips has that option, who uses Atmel does not have the
> option to use the JTAG pins as I/O.
The addition of the on chip voltage regulator is a key feature which cost
some pins.
> Last but not least the measure about the most popular ARM devices?
> How many people (not volume) are using the device and are willing to
> share their wisdom. That is my measure. So the AVR from Atmel is very
> popular (e.g. in the newsgroups / Yahoo forums), so is the Microchip
> PIC but talking about ARM devices, my bet is on the LPC2000 family.
> A Yahoo forum with more than 1300 members and more than 500 on topic
> messages a month, that makes a device popular.
So I wonder how the 250+ projects for the AT91 in my tiny little
region (which normally consist of 5% of the total world market) compares to
that.
Very few (< 10%) use newsgroups, and only a couple use Yahoo.
Based on experience, I know also that there are probably plenty of projects
which
that I dont even know about.
If this is replicated worldwide, then there is 5000 projects.
Does that make the AT91 popular or not?
The fact that a newsgroup does not have so many entries does not
neccessarily make
the chip unpopular. It does mean that the user community is not organized.
It could mean that the local support for the chip is good and that
newsgroups is not needed.
I think you mistake activity on newsgroups for popularity or you define
popularity as activity
on newsgroups instead of number of people actually using the chips.
I also hear comments from distributors about which is part sells best
and comments from tool vendors, which ARM part their customers use
and that is positive (They could of course be lying their teeths out ;-)
> That is just my personal opinon.
Which you are entitled to.
>
>
> Meindert Sprang wrote:
> > "Ulf Samuelsson" <ulf@a-t-m-e-l.com> wrote in message
> > news:3928elF5q6uu9U1@individual.net...
> > > I think it would be interesting to measure how many will switch
> from
> > > the SAM7 to the Philips and how many will switch in the other
> direction.
> >
> > I, for one, would never use a Philips, since it has no memory
> protection
> > scheme.
> >
> > Meindert
> >
> >
> > > The AT91SAM7 is just about to enter production but we have already
> stolen
> > > quite a few Philips designs.
> > >
> > > > So if you need an ARM with lots of external memory, have a look
> at
> > > > Atmel, OKI, Samsung, Sharp, etc.. If you want a single chip
> version,
> > > > Philips offers the best selection of small devices with 48 and
> > > > 64-pins.
> > >
> > > If you look at a matrix of part you will notice that Philips has
> more
> > > devices
> > > but the range of Atmel flash parts is wider.
> > >
> > >
> > > 48 PIN PARTs
> > >
> --------------------------------------------------------------------------
> > --
> > > -------------
> > > 32kB AT91SAM7S32
> > > 64 kB
> > > 128 kB
> > > LPC2104/5/6
> > >
> > > 64 PIN PARTS
> > >
> --------------------------------------------------------------------------
> > --
> > > -------------
> > > 32kB
> LPC2131*
> > > 64kB AT91SAM7S64 LPC2132*
> > > 128kB AT91SAM7S128 LPC2114
> > > 256kB AT91SAM7S256 LPC2124
> > > 512kB AT91SAM7S512* LPC2138*
> > >
> > > 64 PIN PARTS WITH 2 X CAN
> > >
> --------------------------------------------------------------------------
> > --
> > > -------------
> > > 128kB
> LPC2119
> > > 256kB
> LPC2129
> > >
> > > 64 PIN PARTS WITH 2 X CAN
> > >
> --------------------------------------------------------------------------
> > --
> > > -------------
> > > 256kB
> LPC2194
> > >
> > > '*' not available yet, LPC213x is close though if I understand
> things
> > > correctly.
> > >
> > >
> > > 100 PIN PARTS WITH 2 X CAN
> > >
> --------------------------------------------------------------------------
> > --
> > > -------------
> > > 256kB AT91SAM7A3
> > >
> > >
> > > SMALLEST PART - TQFP48 LOWEST COST
> > > Atmels smallest part is a 48 pin part with 32 kB Flash and 8 kB
> SRAM
> > > Philips smallest part is a 48 pin device with 32 kB part and 16 kB
> SRAM
> > > Atmel should win round for lowest cost
> > >
> > > SMALLEST PART - TQFP48 MOST MEMORY
> > > Atmel has the AT91SAM7S32 with 32 kB + 8 kB SRAM
> > > Philips has the 128 kB LPC210x family - This is a 128 kB with
> 16/32/64kB
> > > SRAM
> > > Philips wins the round.
> > >
> > > SMALLEST PART - TQFP48 Analog
> > > Atmel has the AT91SAM7S32 with ADC
> > > Philips 48 pin parts do not have ADC onboard
> > > Atmel wins the round
> > >
> > > LARGEST PART FLASH MEMORY
> > > Atmels largest part in 64 TQFP is 256 kB Flash with 64 kB SRAM
> > > (AT91SAM7S256)
> > > Atmels pincompatible AT91SAM7S512 is some time away.
> > > but also have BGA with 512kb/2 MB Flash and 256 kB SRAM.
> > > This is a dual chip package and does not have so many features
> though
> > > Philips largest announced part is LPC2138 w 512kB Flash and 16 kB
> > > Atmel wins the round if BGA is acceptable, otherwise Philips (when
> > > available)
> > >
> > > LARGEST PART SRAM MEMORY
> > > Atmels largest part in 64 TQFP is 256 kB Flash with 64 kB SRAM
> > > (AT91SAM7S256)
> > > Philips has a single part with 64 kB SRAM and this has 128 kB Flash
> > > The larger parts have less memory. Even the LPC2138 w 512kB Flash
> has 32
> > kB
> > > It is even for 32 kB/64kB Flash
> > > Philips wins the 128 kB Flash
> > > Atmel wins the 256 kB Flash
> > >
> > > PERFORMANCE
> > > Philips has an 128 but wide flash bus running at 20 MHz
> > > Atmel has a 32 bit wide flash bus running at 30 Mhz
> > > Philips has higher performance in ARM mode
> > > Atmel has higher performance in Thumb Mode
> > > If you really need the performance, the AT91RM3400 will outperform
> > > the LPC by executing from its 96 kB SRAM at zero waitstates. This
> will
> > need
> > > an external SO-8 dataflash
> > > If you can afford the 30-40% increase in code size incurred by the
> ARM
> > mode
> > > then Philips is your choice. If not, go for Atmel.
> > > If you accept external memories, the peripherals of the AT91SAM7s
> > > series (except the analog) is available in the AT91RM9200 at 200
> MIPS.
> > >
> > > I/O
> > > Atmel has a 100 pin TQFP series single chip.
> > > Philips soes not have a 100 pin single chipper.
> > >
> > > PERIPHERALS
> > > Atmel peripherals are generally more advanced, coming from ARM9
> chips.
> > > This means they can handle more complex problems
> > >
> > >
> > >
> > > --
> > > Best Regards,
> > > Ulf Samuelsson
> > > ulf@a-t-m-e-l.com
> > > This message is intended to be my own personal view and it
> > > may or may not be shared by my employer Atmel Nordic AB
> > >
> > >
>
Reply by Meindert Sprang●March 7, 20052005-03-07
"An Schwob in USA" <schwobus@aol.com> wrote in message
news:1110233853.074401.23450@z14g2000cwz.googlegroups.com...
> Meindert,
>
> seems you looked at Philips the last time 18 months ago when the very
> first LPC2104/05/06 became available (more than one year ahead of any
> other small ARM device!!).
> Point well taken, the LPC210x don't have code protection, however, the
> flash devices (>10) all do. Let's call it an oversight of the pioneers
> with the face in the dust.
I indeed wasn't aware of the newer types. I'll look into these, thanks.
Meindert
Reply by An Schwob in USA●March 7, 20052005-03-07
Meindert,
seems you looked at Philips the last time 18 months ago when the very
first LPC2104/05/06 became available (more than one year ahead of any
other small ARM device!!).
Point well taken, the LPC210x don't have code protection, however, the
flash devices (>10) all do. Let's call it an oversight of the pioneers
with the face in the dust.
As to the availability, somebody marked the devices of the
LPC2130-series not available that have been acknowledged in the LPC2000
newsgroup being available for almost 3 months. LPC2138, LPC2132 and
LPC2131. afaik the LPC2134 and LPC2136 are not released to the market
yet but given pin compatibility with the LPC2138, everybody can start
the design.
Atmels SAM7 has claimed to have better peripherals, it seems there has
been some oversight to provide enough pins to get them out. Comparing
the 64-pin devices gives me almost 10 extra I/O pins on the Philips
devices. There is a point though in making the JTAG pins dedicated but,
who ever uses Philips has that option, who uses Atmel does not have the
option to use the JTAG pins as I/O.
Last but not least the measure about the most popular ARM devices?
How many people (not volume) are using the device and are willing to
share their wisdom. That is my measure. So the AVR from Atmel is very
popular (e.g. in the newsgroups / Yahoo forums), so is the Microchip
PIC but talking about ARM devices, my bet is on the LPC2000 family.
Talking about high volume delivery into the market, TI delivered double
digit million parts into ABS but are their devices very popular (yet)?
Many high volume designers use the ARM7 (and now ARM9) for ASICs in
hand-helds but does that make the devices popular?
A Yahoo forum with more than 1300 members and more than 500 on topic
messages a month, that makes a device popular.
That is just my personal opinon.
Meindert Sprang wrote:
> "Ulf Samuelsson" <ulf@a-t-m-e-l.com> wrote in message
> news:3928elF5q6uu9U1@individual.net...
> > I think it would be interesting to measure how many will switch
from
> > the SAM7 to the Philips and how many will switch in the other
direction.
>
> I, for one, would never use a Philips, since it has no memory
protection
> scheme.
>
> Meindert
>
>
> > The AT91SAM7 is just about to enter production but we have already
stolen
> > quite a few Philips designs.
> >
> > > So if you need an ARM with lots of external memory, have a look
at
> > > Atmel, OKI, Samsung, Sharp, etc.. If you want a single chip
version,
> > > Philips offers the best selection of small devices with 48 and
> > > 64-pins.
> >
> > If you look at a matrix of part you will notice that Philips has
more
> > devices
> > but the range of Atmel flash parts is wider.
> >
> >
> > 48 PIN PARTs
> >
> --
> > -------------
> > 256kB AT91SAM7A3
> >
> >
> > SMALLEST PART - TQFP48 LOWEST COST
> > Atmels smallest part is a 48 pin part with 32 kB Flash and 8 kB
SRAM
> > Philips smallest part is a 48 pin device with 32 kB part and 16 kB
SRAM
> > Atmel should win round for lowest cost
> >
> > SMALLEST PART - TQFP48 MOST MEMORY
> > Atmel has the AT91SAM7S32 with 32 kB + 8 kB SRAM
> > Philips has the 128 kB LPC210x family - This is a 128 kB with
16/32/64kB
> > SRAM
> > Philips wins the round.
> >
> > SMALLEST PART - TQFP48 Analog
> > Atmel has the AT91SAM7S32 with ADC
> > Philips 48 pin parts do not have ADC onboard
> > Atmel wins the round
> >
> > LARGEST PART FLASH MEMORY
> > Atmels largest part in 64 TQFP is 256 kB Flash with 64 kB SRAM
> > (AT91SAM7S256)
> > Atmels pincompatible AT91SAM7S512 is some time away.
> > but also have BGA with 512kb/2 MB Flash and 256 kB SRAM.
> > This is a dual chip package and does not have so many features
though
> > Philips largest announced part is LPC2138 w 512kB Flash and 16 kB
> > Atmel wins the round if BGA is acceptable, otherwise Philips (when
> > available)
> >
> > LARGEST PART SRAM MEMORY
> > Atmels largest part in 64 TQFP is 256 kB Flash with 64 kB SRAM
> > (AT91SAM7S256)
> > Philips has a single part with 64 kB SRAM and this has 128 kB Flash
> > The larger parts have less memory. Even the LPC2138 w 512kB Flash
has 32
> kB
> > It is even for 32 kB/64kB Flash
> > Philips wins the 128 kB Flash
> > Atmel wins the 256 kB Flash
> >
> > PERFORMANCE
> > Philips has an 128 but wide flash bus running at 20 MHz
> > Atmel has a 32 bit wide flash bus running at 30 Mhz
> > Philips has higher performance in ARM mode
> > Atmel has higher performance in Thumb Mode
> > If you really need the performance, the AT91RM3400 will outperform
> > the LPC by executing from its 96 kB SRAM at zero waitstates. This
will
> need
> > an external SO-8 dataflash
> > If you can afford the 30-40% increase in code size incurred by the
ARM
> mode
> > then Philips is your choice. If not, go for Atmel.
> > If you accept external memories, the peripherals of the AT91SAM7s
> > series (except the analog) is available in the AT91RM9200 at 200
MIPS.
> >
> > I/O
> > Atmel has a 100 pin TQFP series single chip.
> > Philips soes not have a 100 pin single chipper.
> >
> > PERIPHERALS
> > Atmel peripherals are generally more advanced, coming from ARM9
chips.
> > This means they can handle more complex problems
> >
> >
> >
> > --
> > Best Regards,
> > Ulf Samuelsson
> > ulf@a-t-m-e-l.com
> > This message is intended to be my own personal view and it
> > may or may not be shared by my employer Atmel Nordic AB
> >
> >
Reply by Meindert Sprang●March 7, 20052005-03-07
"Jim Granville" <no.spam@designtools.co.nz> wrote in message
news:422c0a6e$1@clear.net.nz...
> Meindert Sprang wrote:
> > "Ulf Samuelsson" <ulf@a-t-m-e-l.com> wrote in message
> > news:3928elF5q6uu9U1@individual.net...
> >
> >>I think it would be interesting to measure how many will switch from
> >>the SAM7 to the Philips and how many will switch in the other direction.
> >
> >
> > I, for one, would never use a Philips, since it has no memory protection
> > scheme.
>
> Their selector guides show memory protection on newer models.
I suppose they just had to, to hold their market share... :-)
Meindert