Jump to content
Fred

A new MSP430 coming [MSP432 ARM]

Recommended Posts

@@jpnorair  I believe you discounted Atmel before, but they claim the SAML21 is the LP champ:  http://blog.atmel.com/2015/03/31/atmels-sam-l21-mcu-for-iot-tops-low-power-benchmark/

Any chance you will add them to the "shootout" or can you eliminate it early because it is M0+ or another reason?

 

The [extremely sophisticated] benchmarks I have will port easily to MSP432 and STM32L4.  Atmel will be later in the year, but I have low hopes for Atmel, and I might not bother.

 

Warning: Rant ahead.

 

Anyway, Atmel uses more BS in their datasheet and marketing than TI does, and TI uses *slightly* more than ST.  

  • Atmel's marketing numbers are unrealistic, best case figures where all the peripherals are off, the integral DC-DC is running, the chip is running at 12 MHz only, and it is running specially crafted code from a small region of SRAM.  Practically, it does 100uA/MHz.  Look at the datasheet, it's all there, albeit somewhat hidden.  Moreover, it's CM0+, so you really should multiply all figures by 66% to compare to CM4: 100 --> 166.
  • TI's numbers are more honest.  A benchmark library is running from FLASH, the clock is 48 MHz, peripherals are mostly on, but the DC-DC is running with input 3.0V.  TI provides a great amount of information about what the power figures are when running with LDO, and they are still quite good (~160uA/MHz).
  • ST is running Dhrystone benchmarks with all the peripherals going, clock at 80 MHz, code from flash, 128 KB of SRAM active, and there is no DC-DC on the device, so it's just LDO.  112 uA/MHz.  I have to guess they have implemented a mixed-size process with the core at 65nm, because they are a "quantum" ahead of the TI and Atmel offerings.

I do wireless IoT, at "low" frequencies (sub 1-GHz), and with low power.  I tend to shun DC-DC converters because they kick up a lot of noise that does affect my radios (I've measured, it can be seriously bad).  It takes a lot of design time to cut-down the noise of DC-DC converters, and it's not always possible (if you can control the switching frequency in software, it's a whole lot better though).  Moreover, the input voltage limit of MSP432 and SAML21 is 3.7 and 3.6V respectively, so IT'S USELESS. It needs to be 4.3V at least, so we can use Li-Ion.  Otherwise, I need to make my own step-down from the Li-Ion, and I'm inclined to use my own DC-DC, with a totally vetted analog design and low EMI, that feeds 1.8V to my system.

 

I really don't need the M4F, but the truth is that I really do prefer the M3/M4 to the M0+.  M0 is a "red-haired step child."  It doesn't fit into a good standard (it's ARMv6+ vs the gold-standard ARMv7).  Compilers are worse.  It only has 8 registers that can be batched to the stack, whereas all the regs in M3/M4 can be (so threading sucks on M0).  It has a crippled NVIC, and on and on.  Basically, it's a low cost hack.  For low power, you actually do better with M3/M4.  For performance, M3/M4 also.

 

M4F gives the MSP432 (and STM32L4) a really good entry point into sophisticated signal processing apps.  Hell yeah, we're going to do FFT on these things.  I think it is ridiculously cool to do FFT -- fast -- with 4-8 mA.  Not everyone here realizes the kinds of possibilities this opens-up.  You might not think you need FFT, but the best Reed-Solomon algorithm uses FFT, and RS is awesome for error correction and data integrity in lossy IoT networks.  So it's a completely legit thing to talk about.

 

I want to reiterate that I only hate Atmel because they make me hate them by being disingenuous.  If they changed their ways, I would have no issue.  I have "no dog in this fight."

 

Anyway, I can tell you now that any performance difference between MSP432P4 and STM32L4 will almost certainly be outweighed by the difference in the development packages, features, and peripherals.  But I still think it is worth the time to explore all the little things that don't get marketed, which actually do have a huge impact on low-power RTOS functionality.  These will be in the official review, coming sometime in May or June I would guess.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...