Hacker News new | past | comments | ask | show | jobs | submit login
Microchip Technology to Buy Atmel for Nearly $3.6B (nytimes.com)
118 points by mkeeter on Jan 21, 2016 | hide | past | favorite | 61 comments



Well, this isn't a particularly insightful comment, but my feelings about this are "fark :[". The AVR definitely has something Nordic about it. I first learned C when programming an AVR Butterfly eval board to take photos like [0] using [1]. I've found the AVR architecture to be much cleaner than 8 or 16 bit PICs. Atmel, for the most part, embraced GPL and GCC. Microchip has tried to squirrel GCC optimization features away from being truly FOSS.

[0]: https://c1.staticflickr.com/1/53/115619170_4445c593cd_o.jpg

[1]: https://c1.staticflickr.com/1/56/112758047_dc5b4bdc3b_o.jpg


IIRC Microchip's 8-bit PIC line had poor/nonexistent GCC support because their terrible instruction set provided no way to maintain a data stack. Their own compiler just allocated everything statically.

Granted, it's (thankfully) been over a decade since I had to use a Microchip microcontroller, and they've presumably improved since. But Atmel got it right from the start. GCC-AVR produced beautiful assembly and their ISA was a joy to use. RIP Atmel.


Calling AVR a "joy to use" is pushing it a little far. There are things to like about AVR chips as a package, but the AVR instruction set is pretty janky. I don't understand why they're so popular compared to the MSP430.


I found the AVR instruction set pretty straight forward. Granted, it was my first forray into assembly and I learned it around 17 years ago at age 14. Maybe everything seems easy in hindsight at that age.

I've never programmed MSP430s in assembly, but I have for PICs and found their instruction set to be more funky than AVR.


So first a couple caveats: (1) I'm sure AVR is a hell of a let better than PIC, (2) I come at this from a really weird place (exclusively emulators and compilers), and (3) I'm not talking about the AVR parts themselves, which might be more cost-effective for a given project.

That said:

* Harvard architecture (split I & D memory)

* 8 bit registers

* Not at all orthogonal, and particularly painful for pointer addressing

* The stack pointer is annoying to work with

* Address wrap at physical memory bounds, rather than the bounds of address space

I could pick a bunch more nits that would only really be relevant to someone writing an emulator (complicated instruction decode, IO addressing, &c) but those are my big complaints.

I find MSP430 much more pleasant to work in.


Those are all true (though that an 8-bit microcontroller has 8-bit registers is… unsurprising). However I am referring to the instructions themselves, which were well-engineered to match the needs of a compiler. The result is very compact assembly for most C programs that you would write on an 8-bit microcontroller (including things like 16-bit arithmetic).


1MIPS @ 1MHZ made it popular.


Care to elaborate?



PICs used to be less power hungry though and had ROM versions.


??? Sounds like FUD to me.


FUD?? Why? The did have ROM version, and, being made from slower silicon had lower leakage.


I don't know that I would say Atmel has embraced GCC. Consider the case of the AVR32: Atmel's port is in a fork of GCC, and has not been merged upstream. Considering how long they've been maintaining their fork, I'm not sure it ever will be merged.


I believe they were working on upstreaming their changes at the time, but as the sister comment said, AVR32 never really took off and is now basically dead.

I still have a couple of AVR32 NGW100 dev boards, they were really nice to work with.


I think AVR32 is dead (also UC3, AP7 died years ago). I've not seen any development the last years. I guess they don't want to compete with their own ARM's.


In a few meet and greets, the Atmel app engineers mentioned that they always steer people to ARM parts unless they have a requirement that absolutely doesn't fit it, then they push AVR32 parts (and seemingly hesitantly).


Out of curiosity, what does "something Nordic" refer to?


I interpreted it as:

Elegant, clean design (certainly compared to the old 8-bit PICs). Good taste. FOSS-friendly.

Microchip might be a Texan oilman in this comparison. Just keep pumping them PICs out don't listen to those commies with their 'GCC' and their 'lee-nux' - sounds foreign and suspicious.

This is obviously a tongue-in-cheek characterisation but I think that's what the OP was getting at.


The AVR architecture was designed in Norway[1].

[1]: https://en.wikipedia.org/wiki/Atmel_AVR#Brief_history


Iirc, Atmel is from Norway


Total guess - maybe Nordic Semiconductor were also kicking the tyres deciding if they wanted to buy Atmel?


Dude, be careful what you post. These are the days when 10-year-olds can't afford to make spelling mistakes [1].

[1] http://www.bbc.com/news/uk-england-lancashire-35354061


Great... more consolidation of major competitors to fix market problems like great features and prices from competition. Just what we needed.


I believe some consolidation helps them reduce costs and improve efficiency. I guess every market consolidates into two to three big players eventually.


It's already at 2-3 big players. Further consolidation gets us closer to monopoly tactics. Let's just say tech consolidation is usually bad for the consumer. I see no reason that two, neck-to-neck competitors becoming one company will help us in long-term. We've benefited from those costs and product efficiency was great so far.

I'm sure their costs and efficiency will improve. As usual. ;)


What is the 3rd player right now? AFAIK it's PIC or Atmel. I guess there are the various ARM boards but those tend to be more powerful than little 8 bit things.


ST Microelectronics has a line of microcontrollers that are (largely) compatible with the Atmel AVR instruction set.


Never heard of those. Do you have a link?


http://www.st.com/web/catalog/tools/FM116/SC959/SS1532/LN184...

They advertise Arduino compatibility, so I'm assuming that means compatibility with the AVR instruction set, but I could be wrong about that.


Thank you.

But that's just Arduino, i.e. high level compatibility. You have a setup() and a loop() and the same way to do I/O, but not the same underlying instruction set.

The chips on these guys are regular ARM cortex uCs.

They might be just barely fast enough to do real-time emulation of the AVR INSNs though ;)


Renesas is a huge microcontroller supplier that no one ever hears about for some reason.

Other major competitors are NXP/Freescale, ST, TI, and Infineon.


I literally just posted that and deleted it in a comment. It had something else that was inaccurate. I was about to repost it just saying Renesas had SOC's, SuperH chips, and their on fab then saw you comment. :)


TI still makes MSP430s.


As well as a variety of ARM microcontrollers, including the confusingly named MSP432:

http://www.ti.com/tool/MSP-EXP432P401R


Texas Instruments maybe?


It's in the article...


On the upside. Less companies means fewer platforms and less fragmentation for developers to worry about.


Developers routinely rely on this fragmentation for more options in the feature set vs. price product space. Such a merger will most certainly lead to some edge case configurations go missing. Lowered prices may be the only actual upside - if it materializes in the light of shrinking competition.


Exactly. This market demands that to find the right fit for a product and best value in performance/IO/watts/dollars. More options is always good for us. Anyone wanting to learn less can stick to one product family.


Atmel's and Microchip's architectures are so different that I wouldn't expect any consolidation of platforms.


Less companies means less wooing for end users and thus - a general quality drop in the long run.


Nah, there are lots of markets that have natural incentives against monopolies. For example, classic web hosting type things will probably never consolidate for a number of reasons ranging from the size and difficulty of the configuration space to being unable to retool when new technology comes along and new players enter.

Chip design and manufacture, however, I can easily believe converge toward a monopoly, mostly because research efforts compound like crazy.


And NRE costs. Lets not forget that. More consolidation every node.


Right, chip fabrication pipelines. Your lithography setup costs $15B, but once you have it set up you can manufacture anything. Ouch.


That's true for fabs but I was talking HW development. Mainly, the EDA tools and the masks. It's why some companies can specialize in offering USB or whatever on certain nodes. It's just that expensive to develop something like that.


I see a disconnect. Somebody knowledgeable, please comment.

here it goes :

Atmel had a deal with Dialog Semiconductor worth 4.6 Billion Dollars on September 15 already, refer [0] [1].

Microchip manage to break that deal with 137 Million dollar termination fee, but with a deal worth of 3.6 Billion Dollars.

That's 1 Billion dollars "less" deal worth agreed by Atmel to be sold.

Sounds unbelievable!

MCHP managed to do this by giving more cash for Atmel shareholders? (..and not to forget 137 M of termination fine.)

1 Billion dollar (which was imaginary) disappeared into thin air??

If it was not of real value, then how / why Dialog over valued it?

Whoever understands all this, please comment.

[0] : http://www.dialog-semiconductor.com/content/dialog-semicondu... [1] : http://fortune.com/2015/09/21/dialog-buying-atmel/


I think you kind of answered it yourself. Microchip probably offered a higher portion of the deal in cash and atmel shareholders find the cash to be worth more than dialog stock.


I think the value of the Dialog deal plummeted since a lot of it was to be paid in Dialog stock and their stock price dropped substantially since September.


This popped up earlier (w/ 20 comments as of now): https://news.ycombinator.com/item?id=10939576


This is surprising news to me. An analog with the microprocessor market would be Intel buying AMD!

I love Microchip's application notes. My first controller was a PIC16F877A and apart from the architecture quirks (and mediocre C compiler support) it is a good processor for beginners. In that sense, this acquisition nicely complements MT's own strengths.

I'm still not sure how I feel about the monopolization of the uC though.


I tried both PIC and AVR when first getting into embedded development. I very quickly grew a strong distaste for the PIC instruction set (their docs were admittedly very nice).

AVR was just so clean by comparison. It's a sad day, microchip really needed a good competitor.


The whole swarm of cheap ARM micros that are now around or below the 1$ per chip, plus with free and sane C compiler ?


From my own experience, this might be a good time for Atmel employees to polish their resumes.


I know someone at Atmel, and many employees are only still there because of promised severance packages.


Polishing of resumes is a lot less effective if you're not in software.


This is a huge relief compared to the proposed acquisition by Dialog. Microchip is vastly more open. Crazily enough, You have to apply for a copy of a data sheet from Dialog instead of it being freely available on their website for everyone to download..


I commented in the other thread https://news.ycombinator.com/item?id=10939576:

>People in this thread keep talking like this is the "death" of Atmel. I don't see it changing much other than being guided by Microchip management to meet higher profit margins. If you think that Atmel's compilers and IDE's are going to be dropped overnight in favor of MPLAB X and Microchip compilers, you are almost surely mistaken; there's a big gap between architectures so it would be a lot of work to get Microchip tools to support Atmel parts.


Cool, maybe now it'll be possible to obtain Atmel chips in quantity without ridiculous lead times.


I wounder if this will this bring the death of the original Arduino.


I suspect it won't - even if AVR started to disappear the Arduino guys have already got a handful of boards out using ARM Cortex M0 and Intel Curie chips. They're quite cleanly integrated with the same IDE and the boards are in roughly similar form factor, so from a dev's perspective if they moved to ARM/Curie completely it wouldn't be too tough I reckon. Not sure if there's any supply chain or manufacturing issues that could complicate things though.


I hope not, but then again, the conflict inside the Arduino project (Arduino LLC vs Arduino SRL) is a threat, too.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: