Here is an old investor presentation from Microchip. It touched on quite a few areas but what's interesting to me is this page:
It talks about how likely people are going to pick which 32-bit chips for their next design (less than 100Mhz only).
PIC32 led with 33% of the surveyed saying that they would consider PIC32. Very good picture for Microchip.
or is it?
what do you think?
I think the hidden rat is in the sentence " below 100MHz"
If that is the core clock and you go for a 32 bit micro you probably want the computing power and the Arms are going way above the 100MHz.
Also I am suspicious about who they queried, where these pro businesses or ?
You said it yourself: Investor presentation. It is filled with BS, as you would expect. They are coming up in 8 and 16 bits, because above them others stopped making it or switched to M0 or merged with other companies. They put "AMD Fusion, Athlon, Sempron, Turion, Opteron, Geode", TI OMAP and Arduino and the Tegra as under 100 Mhz. Atmel AT91xx is twice on that table.
I'm sure if they would show a graph about the increasing number of cat videos rendered on 16 bit MCUs the investors would be happy.
*isnt geode and OMAP and bunch of other platform NRND anyway? Sure this survey must have been filled by "experts".
If that is the core clock and you go for a 32 bit micro you probably want the computing power and the Arms are going way above the 100MHz.
Yeah. Many of the higher-end CM3/4 chips wouldn't qualify here for example.
Also, 33% of the 810 surveyed (=~260) said they would consider PIC32.
How many of them said that they would consider ARM, even with the <100Mhz handicap?
[Microchip] stock does not do that well
How do you figure? Microchip is the second best performer in my personally-picked portfolio, up over 80% since my purchases in ~2010. Plus 3% dividend. That's on par with DJIA, significantly better than STM and ONNN, similar to ATML (without as much roller coaster.) Only NXPI has done much better (and that, only recently (when I wasn't paying attention!))
https://finance.yahoo.com/echarts?s=MCHP+Interactive#symbol=mchp;range=5y;compare=^dji+atml+onnn+stm+nxpi;indicator=volume;charttype=area;crosshair=on;ohlcvalues=0;logscale=off;source=;
(Best performer in the portfolio: ARMH - up 620% since 2006 :-)
An interesting technical question: would you put PIC32 (MIPS core) ahead of other proprietary-core 32bit CPUs? (AVR32, Renesas, etc?) Because it's a "standard core"? Because it's Microchip?
Add up all the ARM chips, it comes to well above 100% (WTF!)
The question is "which would you consider?", in which case it's perfectly reasonable to consider more than one option.
Personally, I won't know what chips I'd consider for my next embedded project until I know what that project is and what its requirements are. It would be a pretty odd set of requirements that would have me using a PIC32 over, say, an STM32, though.
Sometimes customers prescribe specific parts that they want to use, which incidentally is the only reason I ever have used a PIC32 in a project. It wasn't a particularly good fit in terms of price or performance, but it had lots of I/O pins and could be programmed with a cheap PICKIT3, which was all that mattered at the time.
The question is "which would you consider?", in which case it's perfectly reasonable to consider more than one option.
Precisely. The page speaks with authority about the uphill battle Microchip faces in the 32-bit market and the future of the market.
Given this is an "investor" presentation, I was wondering if a mole for ARM slipped that page into the presentation,
. Or the management simply didn't understand the chart.
Either way, I would be scared as an investor.
[Microchip] stock does not do that well
How do you figure? Microchip is the second best performer in my personally-picked portfolio, up over 80% since my purchases in ~2010. Plus 3% dividend. That's on par with DJIA, significantly better than STM and ONNN, similar to ATML (without as much roller coaster.) Only NXPI has done much better (and that, only recently (when I wasn't paying attention!))
https://finance.yahoo.com/echarts?s=MCHP+Interactive#symbol=mchp;range=5y;compare=^dji+atml+onnn+stm+nxpi;indicator=volume;charttype=area;crosshair=on;ohlcvalues=0;logscale=off;source=;
How do I tell, that's easy, compare their stock to IXIC and you will see that it's way below the NASDAQ composite. Same for VTSMX (VG total stock market). If this is your second best performer I feel bad for you. I highly recommend this little book
http://amzn.com/0399536345 .
When I was considering 32 bit microcontrollers, the PIC32 was at the bottom of the list simply because ARM has a much larger ecosystem to draw on for code and development tools. ARM also lowers the likelihood of vendor lock in.
If I have to learn another instruction set architecture, I would much rather go with the more common ARM than the less common MIPS.
..ARM also lowers the likelihood of vendor lock in.
Two points here..
The ARM cores are the same but you cannot source your favourite ARM chip from another vendor, so locked in you are young Luke!
I dont work that close to the metal and Everything is done in C these days I personally wouldnt have a clue if the machine code generated was arm or mips or FutureMegaChipX, and I cant say I really care so long as the IDE, documentation and libraries are good, the price is low and the supply is good.
People I know pick MC micros because they are familiar with the peripherals (and MC peripherals are very good). Other than that PIC32 is pretty boring chip - old core, high power consumption, etc. On the other hand, Microchip parts have good availability and they sell any quantity directly, which is a huge plus these days.
..ARM also lowers the likelihood of vendor lock in.
The ARM cores are the same but you cannot source your favourite ARM chip from another vendor, so locked in you are young Luke!
The vendor lock-in is a lot more powerful when you have to change ISAs as well though. At least the various ARM chips are roughly comparable and have the same idiosyncrasies.
I dont work that close to the metal and Everything is done in C these days I personally wouldnt have a clue if the machine code generated was arm or mips or FutureMegaChipX, and I cant say I really care so long as the IDE, documentation and libraries are good, the price is low and the supply is good.
I invariably end up doing some work at the assembly level if only because of compiler issues.
People I know pick MC micros because they are familiar with the peripherals (and MC peripherals are very good). Other than that PIC32 is pretty boring chip - old core, high power consumption, etc. On the other hand, Microchip parts have good availability and they sell any quantity directly, which is a huge plus these days.
Microchip has great documentation and availability. The later makes up for them being a sole source as far as 8 and 16 bit PIC compared to their competitors in that space but for PIC32, ARM is an alternative to MIPS. I would really love to see Atmel as an alternative but they have not lived up to that.
At least the various ARM chips are roughly comparable
The cores are identical. But as a practical matter, you cannot buy a core only. What makes the difference are the peripherals and they are highly dissimilar from vendor to vendor and sometimes from chip to chip.
As long as you code in a high level language, the core is largely irrelevant for this discussion.
At least the various ARM chips are roughly comparable
The cores are identical. But as a practical matter, you cannot buy a core only. What makes the difference are the peripherals and they are highly dissimilar from vendor to vendor and sometimes from chip to chip.
As long as you code in a high level language, the core is largely irrelevant for this discussion.
I know this is technicality, but in fact, you can buy just the core. From ARM. Implementation is up to you.
..ARM also lowers the likelihood of vendor lock in.
The ARM cores are the same but you cannot source your favourite ARM chip from another vendor, so locked in you are young Luke!
The vendor lock-in is a lot more powerful when you have to change ISAs as well though. At least the various ARM chips are roughly comparable and have the same idiosyncrasies.
IMHO the problems are a lot bigger when switching between toolchains (compiler) than between processor architectures. Each C compiler has it's own slang and especially in microcontroller land you quickly run into compiler/toolchain specific C extensions (pragma's/attributes for example).
IMHO the problems are a lot bigger when switching between toolchains (compiler) than between processor architectures. Each C compiler has it's own slang and especially in microcontroller land you quickly run into compiler/toolchain specific C extensions (pragma's/attributes for example).
What is great is that most ARM toolchains are GCC-based and you can make do with a single compiler even for chips from completely different vendors.
I had a lot more trouble switching between C18 and XC8 on PICs than switching between NXP and ST Micro ARM micros (and keeping the same compiler). With ARM even if you have to change chip vendor, the toolchain remains pretty much the same.
the PIC32 compiler is also gcc, so while you might have some culture shock going from PICC8 to PIC32 gcc, subsequent moves to other gcc-based chips shouldn't result in additional shock.