Upset at Microchip Technologies (What are they thinking)

[ivaylo]

I've received free Maxim samples this way. Huge 3 feet long box, bubble wrap, another smaller box inside, more bubble wrap and then three tiny chips in the middle. Came all the way from Asia somewhere. I clearly don't understand this shipping thing (I understand they use bulk pricing and all but if I try to ship that thing it's gonna cost me an arm and a leg). Unless a disgruntled employee did it.

[EEVblog]

Absolutely nothing wrong with the packing you described.

Technically, if you take ESD seriously, there is something wrong with it. It is not best practice, and would not hold up to a stringent ESD audit. I know companies that have been black-listed as a supplier for doing that.

In practice, no, par for the course.

Dave.

[mikeselectricstuff]

Silicon companies have to charge for their compilers for three reasons. 
1. Accounting Trolls - because even in a company who's job it is to make and sell silicon there are still bean counters who don't understand the business.

Microchip CEO Steve Sanghi personally confirmed this to me. The reason they have to at least try and break even on tools is because if they didn't, there would be too much pressure (from the board/shareholders) to shut the division down and/or outsource it etc. Be careful what you wish for.

Dave.

It does make you wonder about the quality of people on the board  if they can't understand things like this though...
 

[firewalker]

It does make you wonder about the quality of people on the board  if they can't understand things like this though...
 

MBA people?

Alexander.

[blewisjr]

Silicon companies have to charge for their compilers for three reasons. 
1. Accounting Trolls - because even in a company who's job it is to make and sell silicon there are still bean counters who don't understand the business.

Microchip CEO Steve Sanghi personally confirmed this to me. The reason they have to at least try and break even on tools is because if they didn't, there would be too much pressure (from the board/shareholders) to shut the division down and/or outsource it etc. Be careful what you wish for.

Dave.

This does indeed make sense.  Stakeholders in a company can be quite a nuisance when it comes to small things like this.  This does however say something about their company structure that many companies try to avoid but they are indeed super successful as a whole profit wise.  Despite the stakeholders pushing for the charge I still think there could be a better way to market their compilers and still break even.  It is not like they update the compilers every day with new chips because they don't come out with new chips every day.  Even if it was as simple as allowing the free version to at least optimize to Level 1 so that the code is not 50% more bloated in larger cases.  50% is quite ridiculous of a limitation.  Level 1 would at least bring that down to 25% more bloat then Level 3.  Still attractive enough to buy the Pro version at some point but not hindering to the hobbyist.  Sure shoot for a bigger chip but bigger chip = more $$ and many hobbyists learning are bound to blow stuff up and to get to the bigger chips you are also looking at the architecture getting more complex which can make learning harder for the hobbyist.

[mikeselectricstuff]

Silicon companies have to charge for their compilers for three reasons. 
1. Accounting Trolls - because even in a company who's job it is to make and sell silicon there are still bean counters who don't understand the business.

Microchip CEO Steve Sanghi personally confirmed this to me. The reason they have to at least try and break even on tools is because if they didn't, there would be too much pressure (from the board/shareholders) to shut the division down and/or outsource it etc. Be careful what you wish for.

Dave.

This does indeed make sense.  Stakeholders in a company can be quite a nuisance when it comes to small things like this.  This does however say something about their company structure that many companies try to avoid but they are indeed super successful as a whole profit wise.  Despite the stakeholders pushing for the charge I still think there could be a better way to market their compilers and still break even.  It is not like they update the compilers every day with new chips because they don't come out with new chips every day. 

With Microchip it seems like it's not far off - I've always been puzzled how it can be worth doing so many variants. 

Even if it was as simple as allowing the free version to at least optimize to Level 1 so that the code is not 50% more bloated in larger cases.  50% is quite ridiculous of a limitation.  Level 1 would at least bring that down to 25% more bloat then Level 3.  Still attractive enough to buy the Pro version at some point but not hindering to the hobbyist.  Sure shoot for a bigger chip but bigger chip = more $$ and many hobbyists learning are bound to blow stuff up and to get to the bigger chips you are also looking at the architecture getting more complex which can make learning harder for the hobbyist.

I think Microchip have got this strategy wrong. Code-size limits make a lot more sense IMO
Producing unnecessarily bloated code means that someone evaluating them against the competition could easily come to falsely bad conclusions about speed and code size efficiency.
Compiler quality is a much bigger issue on PICs than most other common architectures as most of the competition have been designed around high level languages, so even an avarage compiler can work OK - PICs (at least the 8-bit ones) really need a clever compiler to get the best out of them.

[alm]

I think Microchip have got this strategy wrong. Code-size limits make a lot more sense IMO
Producing unnecessarily bloated code means that someone evaluating them against the competition could easily come to falsely bad conclusions about speed and code size efficiency.

I agree. Code size and performance are important factors when evaluating compilers or MCU platforms. Making themselves look bad is a stupid strategy in my opinion. Code size limitations is also what most of the other compiler vendors, at least in the ARM world, use.

[AlfBaz]

When they released the pic32 compiler it was code size limmited... After incessant whinging on the forum the next version came out with optimisation limits. Funny thing was if you downloaded the compiler source code (it was a gcc port) you could see the source file with the licensing limits, making it a breeze to hack

[blewisjr]

I agree code size is a good choice.  Give say level 1 or 2 optimization and maybe 3 - 4 kbytes of a limitation.  If you go beyond that odds are you are beyond beginner and well into intermediate.  If you push the limit you know have 3 choices port to ASM, mix ASM and c or buy the compiler.  This makes buying more convenient because odds are you are working on something cool that helps justify the purchase for a hobbiest that is.  Professionals would know right off the bat if they need it and work would buy it.

[SteveyG]

The correct reason to be ranting at Microchip is poor silicon engineering, unclear datasheets and libraries with inherent software bugs in them. About those I could go on for a whole day...

Microchip have some of the best written datasheets out there

[EEVblog]

It does make you wonder about the quality of people on the board  if they can't understand things like this though...

Define "quality". You can be a smart high quality board member that only cares about short term profits and your golden handshake 

Remember, board members have one duty that overrides all others, a fiduciary responsibility to shareholders, most of whom crave short term profits.

Dave.

[EEVblog]

Still attractive enough to buy the Pro version at some point but not hindering to the hobbyist.  Sure shoot for a bigger chip but bigger chip = more $$ and many hobbyists learning are bound to blow stuff up and to get to the bigger chips you are also looking at the architecture getting more complex which can make learning harder for the hobbyist.

Remember they are not in the business to pander to hobbyists, they derive almost no income from that. Only some perceptual concept of hobbyists turn into loyal commercial customers in the future etc.

Dave.

[blewisjr]

Still attractive enough to buy the Pro version at some point but not hindering to the hobbyist.  Sure shoot for a bigger chip but bigger chip = more $$ and many hobbyists learning are bound to blow stuff up and to get to the bigger chips you are also looking at the architecture getting more complex which can make learning harder for the hobbyist.

Remember they are not in the business to pander to hobbyists, they derive almost no income from that. Only some perceptual concept of hobbyists turn into loyal commercial customers in the future etc.

Dave.

I agree here Dave but like you said hobbiests become loyal customers.  Ironically I just picked up an arduino ultimate maker kit.  I am rather disappointed with how hard it is to find the information to attack avr from the perspective I want.  I have to give microchip credit there easy to find what you need.  Essentially I just wanted to use the arduino as a proto dev board and then use that to learn the architecture of the avr chip.  I want to do this through ASM and finding info beyond some small tutorials is next to impossible on the Atmel site.

At least now I have a crapload of components to use with my PIC.

[jucole]

  Also lets not forget there horridly over priced C compiler.

As far as I can recall XC8, XC16, and XC32 are free these days including commercial purposes.  It starts out as a 60 day Pro trial and reverts to the lower levels of compiler optimizations (which are still decent).

Silicon companies have to charge for their compilers for three reasons. 
1. Accounting Trolls - because even in a company who's job it is to make and sell silicon there are still bean counters who don't understand the business.
2. Some customers won't consider a compiler legitimate unless it costs lots of money.
3. Some want guaranteed support with a direct line which a good chunk of what they are paying for.

There is a free c compiler called "Small Device C Compiler"   http://sdcc.sourceforge.net/

It supports the PIC16F1829 along with these..





It would be nice for someone to try and benchmark all the various compilers for some of the popular make of chips to see how they get on.

[andersm]

Essentially I just wanted to use the arduino as a proto dev board and then use that to learn the architecture of the avr chip.  I want to do this through ASM and finding info beyond some small tutorials is next to impossible on the Atmel site.

What kind of information are you looking for? If you go to the product page of the device and hit the "Documents" tab there's a ton of documentation including the assembly language reference and development techniques app notes.

[blewisjr]

I am basically looking for an instruction reference.  The rest can be figured out from the data sheet.

[alm]

AVR Instruction Set Manual