Split from: AVR vs. TI MSP430

Go To Last Post
20 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi All,

 

I was just about to create a new post when I came across this OLD post from 2003.

Basically, I have a suite of products that use the MSP430F1611IPM and MSP430FG4618. I am running out of code space (10 bytes left) and looking at swapping out the micro's for new bigger devices.

Personal preference is AVR Mega series or XMega series.

How do these compare to the MSP430 series?

 

Just to through another micro into the mix, one of the projects has a PIC18F480. I would love to swap this out for a Mega or something as well.

 

Thoughts???

 

Thanks,

Neil.

Kind Regards,

Neil Wrightson.

Last Edited: Fri. Feb 8, 2019 - 11:09 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I have a suite of products that use ... MSP430FG4618

You're running out of code space on an MSP430 with 116kbytes of flash?

IMO: time to upgrade your thinking to one of the 32bit chips (SAMD, MSP432, etc.)

I mean, academically, I'd rate both MSP430 and AVR as "hurting" once you get above their native 16bit address space.

There aren't many plain "mega" parts with more than 128k of memory (and going to "only 128k" when you're running out at 116k seems foolish), unless code density REALLY improves, which I would doubt.  MSP430 is a reasonable 8bit CISC instruction set, AVR is reasonable 8bit RISC...

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

nwrightson wrote:
Personal preference is AVR Mega series or XMega series.

How do these compare to the MSP430 series?

  • Some AVR have an EBI (MSP430TM doesn't)
  • MSP430 may be a bit more compute efficient than AVR

dsPIC33C arrived mid'18 with Curiosity boards for it; the single core dsPIC33C has an EBI (16b address, 8b or 16b data, mux'd)

nwrightson wrote:
Just to through another micro into the mix, one of the projects has a PIC18F480.
MPLAB XC8 for PIC made a major change from v1 to v2 (clang, C99)

 


CPU Benchmark – MCU Benchmark – CoreMark – EEMBC Embedded Microprocessor Benchmark Consortium

 

via yesterday's e-mail :

Microchip Technology - Development Tools - 16-bit Curiosity Boards

 

MPLAB- XC Compilers | Microchip Technology

Clang C Language Family Frontend for LLVM

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just Add 

A MSP430 is a 16bit chip not 8.

 

So if the MSP430 is to small I also would go for an ARM, which I don't know that will really depend.

 

add:

One thing both MSP430 and AVR are good compared to ARM is low power, and sleep power, but newer ARM's are getting better.

Last Edited: Wed. Feb 6, 2019 - 12:22 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

westfw wrote:
... (SAMD, MSP432, etc.)
Renesas SynergyTM

westfw wrote:
I mean, academically, I'd rate both MSP430 and AVR as "hurting" once you get above their native 16bit address space.
MSP430TM has a 20-bit address space; FSF MSP430 GCC has a 20b pointer type.

IAR EWAVR has the 24b pointer type in a few memory models; AVR GCC has 24b types (XMEGA) but would take some effort to make a pointer by it.

westfw wrote:
There aren't many plain "mega" parts with more than 128k of memory ...
XMEGA384 models have 384KB program space and 32KB of local RAM (that's usually enough RAM for most applications)

 


http://www.ganssle.com/tem/tem365.html#article6

[Renesas SSP]

Need development tools? IAR's suite is provided along with their runtime analysis tools.

...

So, how free is free? The website indicates these are all totally free, with no licensing or royalty fees

...

[e-mail]

...

After getting that email I ordered a $35 Synergy dev board and will give it a whirl.

...

'16 MCU ranking by revenue : 1. NXP, 2. Renesas, 3. Microchip

Am not certain the ones at Microchip can top Renesas SSP.

https://gcc.gnu.org/onlinedocs/gcc-8.2.0/gcc/MSP430-Options.html#MSP430-Options

...

 

-mlarge

-mlarge

Use large-model addressing (20-bit pointers, 32-bit size_t).

 

...

https://www.iar.com/iar-embedded-workbench/#!?architecture=AVR

https://gcc.gnu.org/wiki/avr-gcc#Types

https://www.microchip.com/sitesearch/search/Product%20and%20Development%20Tools/ATxmega384

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

westfw wrote:
IMO: time to upgrade your thinking to one of the 32bit chips (SAMD, MSP432, etc.)
One might prefer to first select the OS, RTOS, framework, scheduler, etc then select the MCU second.

Zephyr's CPU are all 32-bit :

Supported Boards — Zephyr Project Documentation

The "unusual" CPU in that list is RISC-V; Microchip is well placed in RISC-V by the acquisition of Microsemi.

RISC-V Foundation | Instruction Set Architecture (ISA)

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

westfw wrote:

I have a suite of products that use ... MSP430FG4618

You're running out of code space on an MSP430 with 116kbytes of flash?

IMO: time to upgrade your thinking to one of the 32bit chips (SAMD, MSP432, etc.)

I mean, academically, I'd rate both MSP430 and AVR as "hurting" once you get above their native 16bit address space.

There aren't many plain "mega" parts with more than 128k of memory (and going to "only 128k" when you're running out at 116k seems foolish), unless code density REALLY improves, which I would doubt.  MSP430 is a reasonable 8bit CISC instruction set, AVR is reasonable 8bit RISC...

 

 

The MSP430F1611 as used in the current products has the following spec:

- 48KB+256B Flash Memory

- 10KB RAM

 

This is why I was looking at say the AVR Mega128 or XMega128.

 

 

Kind Regards,

Neil Wrightson.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Have you tried the MCU selector: https://www.microchip.com/maps/M...

Only you know exactly what kind of specs you need.

 

edit: yeah, I'd go for 32 bit for these specs. They are faster and cheaper, this means the manufacturer wants people to use them. Here are a couple from our sponsors:

https://www.microchip.com/wwwpro... (PIC32, i.e. MIPS32 architecture, I never used them but the specs seem nice)

https://www.microchip.com/wwwpro... (and this one is a Cortex-M0+. From my limited experience, the SAMD are nice chips, a bit complicated to configure, but that is kind of inevitable with 32 bit chips)

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

So you are still working you poor thing..... wink

John Samperi

Ampertronics Pty. Ltd.

www.ampertronics.com.au

* Electronic Design * Custom Products * Contract Assembly

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

PIC32 and some SAM can reach 125C though dsPIC can reach 150C except for the new dsPIC33C at 125C (radiation tolerance is proportional to maximum ambient temperature)

16b is a typical sweet spot for code and data density as 16b integers are usually enough precision for most applications; an MCU manufacturer may have a fixed-point math library.

 

dsPIC33EP512GM310 - 16-Bit - Microcontrollers and Digital Signal Controllers

New Digital Signal Controller (DSC) Accelerates DSP Performance for Time-Critical Control Applications | Microchip Technology

dsPIC33CK is Microchip’s highest performance single-core DSC in an ultra-small package

Fixed Point Math Library for PIC24 and dsPIC

via Software Libraries

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

El Tangas wrote:
... and this one is a Cortex-M0+. From my limited experience, the SAMD are nice chips, a bit complicated to configure, but that is kind of inevitable with 32 bit chips

 

This is exactly the reason I never will change to these 32 Bit chips: Complicated to configure! Unnecessarily difficult to configure!

It's so pleasant to work with 8Bit controllers that are simple in design but still completely sufficient for me and many purposes.

Last Edited: Fri. Feb 8, 2019 - 07:46 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

nwrightson wrote:
The MSP430F1611 as used in the current products has the following spec: - 48KB+256B Flash Memory - 10KB RAM   This is why I was looking at say the AVR Mega128 or XMega128.
You want to REDUCE the RAM??

 

The fact is that the flash to RAM ratio in AVR is not as high as ARM or (apparently) MSP. There is only one AVR that has more than 8K RAM and that is the Mega1284P which has a whopping 16KB. So if RAM is going to be an issue then AVR may not be the right choice.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

GermanFranz wrote:
...: Complicated to configure! Unnecessarily difficult to configure!
Hopefully, Jack Ganssle will write a review of Renesas SynergyTM SSP.

wrt Microchip PIC32, I have zero experience with MPLAB Harmony.

GermanFranz wrote:
It's so pleasant to work with 8Bit controllers that are simple in design but still completely sufficient for me and many purposes.
Concur

Amazing that 8b MCU have DMA, other somewhat complex peripherals, and are being improved (there's persistent demand for 8b MCU)

 


IAR Embedded Workbench® for Renesas Synergy™

MHC | Microchip Technology

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

clawson wrote:
There is only one AVR that has more than 8K RAM and that is the Mega1284P which has a whopping 16KB.
... and several XMEGA256 at 16KB plus the two XMEGA384 at 32KB.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

nwrightson wrote:
... Mega128 ...
and AT90USB1286 or AT90USB1287 though, as Cliff stated, you'll have more internal RAM with most 32b MCU.

AT90USB128 series external memory (XMEM) : one latch to demux the LSB of address and data

 

Teensy Technical Specifications (Teensy++ 2.0)

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@OP - have you looked at why you are running out of code space?  Can you just optimize current codebase and get more free space?  That way you don't end up with orphaned hardware with old CPU ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

andrewm1973 wrote:
That way you don't end up with orphaned hardware with old CPU ?
An example of such is AVR32 after it was legacy'd by Atmel.

AVR and MSP430TM are very low risk of legacy by the respective manufacturers or by foundations (FSF, LLVM)

PIC24 and dsPIC weren't selected by foundations.

PIC32 has a MIPS core so it should have a long life.

PIC - it's interesting that the ones at Microchip selected Clang for MPLAB XC8 PIC v2.

 

http://clang.llvm.org/docs/UsersManual.html#other-platforms

https://github.com/llvm/llvm-project/blob/master/clang/lib/CodeGen/TargetInfo.cpp

https://github.com/llvm/llvm-project/tree/master/llvm/lib/Target

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I was actually responding to OP saying

 

Basically, I have a suite of products that use the MSP430F1611IPM and MSP430FG4618. I am running out of code space (10 bytes left) and looking at swapping out the micro's for new bigger devices.

 

If he changes CPU then his suite of products now has a lot of "version 2" that is incompatible with version one.  I don't know how many products he has out there or if he has to support past customers.  If he does have to support old customers then maybe code space optimizing is a better idea.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

"Dare to be naïve." - Buckminster Fuller

Last Edited: Wed. Feb 27, 2019 - 12:46 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Build Larger, More Robust Applications with Microchip’s Expanded Dual- and Single-core dsPIC Digital Signal Controller (DSC) Family | Microchip Technology

New dsPIC33C devices meet market demand for more memory and functional safety features in automotive and wireless charging applications

Chandler, Arizona

March 13, 2019

...

Microchip’s new dsPIC33CH512MP508 dual-core DSC enables support for applications with larger program memory requirements. The dsPIC33CK64MP105 single-core DSC adds a cost-optimized version for applications that require smaller memory and footprint. 

...

A robust set of CAN-FD peripherals, along with the new support for 150°C operation, make these devices ideally suited for use in extreme operating conditions such as under-the-hood automotive applications.

...

The dsPIC33CH Curiosity development board is available now for $39.99 each.

The dsPIC33C PIM development boards mentioned above are available now for $25.00 each.

...

 

"Dare to be naïve." - Buckminster Fuller