Retrospective of 2014: Goodbye EDA, hello IoT!

It's this time of the year! Whether there is snow outside your window or not, the end of year is nigh, and it's time to reflect on what happened this year at Synflow, and to think about the future and where we want to go. We've (finally) got enough of the semiconductor industry's fear of change, and we're tired of the EDA mindset. We're saying hello to the much more interesting Internet of Things (IoT), its challenges and opportunities, and will focus on enabling IC design for the IoT!

Christmas Night (c) 2010 mobilestreetlife - Flickr

There's no best way to sum up a whole year, so I've just listed the things we did by chronological order below:

  • at the beginning of 2014 we secured funding for one year thanks to a 0% interest personal loan combined with a bank loan (not zero percent that one ^^).
  • we took a trip to the French Riviera to see semiconductor companies there. Two of them were particularly interested (one was Samsung Research France, now defunct). Another one, RivieraWaves led us to believe they were interested for months, only to be patronizing, and in the end, telling us that our product is not worth the effort since they were almighty Verilog experts.
  • we attended the DATE 2014 conference, which was a huge disappointment. Dan had warned us about conferences, but somehow we didn't think about it too much. Turned out we should have listened, sorry Dan!
  • we pitched at START WEST. No feedback, waste of time.

At this point, things were a bit unclear. The offices we were in at that time were in front of Renesas, who had been shut down for 7 months. Samsung was particularly interested if we moved near them to provide support for our technology. This is about the time when we were contacted by TheFamily. A trip to Paris later, meeting with Jean de la Rochebrochard and we were convinced and entered TheFamily.

Synflow opens a new office

Not much later Nicolas moved to the Riviera. We were hoping this would get us closer to signing a contract and would allow us to extend our network to the numerous companies there working with semiconductors. At least we were right that it was a much better place to be than Brittany (from many points of view).

Juan les Pins

We redesigned our website based on advice from Jean and our friends at Homadeus, and despite our initial skepticism, we set up a forum and started to focus on the community. I met with Liam Boogar at the Web2day conference, who suggested that we moved to open source, and after thoughts and discussion with Nicolas, we open sourced our compiler.

Technological achievements and roadmap

2014 is the year during which I completed the transition to textual description of networks and implemented specialization so you can write generic code and instantiate specialized versions of it (in the future this will allow us to support multiple instantiation and dependency injection). For the first half of the year, the VHDL code generator had been broken, I repaired it during the summer. Other noticeable improvements are proper clock domain support and checks, and customizable clocks and resets, supported with both VHDL and Verilog code generation.

Just before the end of the year I've added support for simulation once again, but this time based on C to make it easy to hook API calls to other systems (like co-simulation of legacy RTL or even hardware/software co-simulation). Oh and unlike Vivado HLS this supports bit-accurate simulation and arbitrary large integers (ok not 100% implemented yet, but still) with any C compiler!

Our roadmap keeps evolving, but for now we plan to make Cx more user-friendly for software programmers, with things such as automatic array implementation as ROM or RAM, and support for fixed and floating point. Since many operations in fixed point and floating point are much better implemented using a pipeline, we will introduce support for pipelining tasks based on a "pipeline" property (with manually specified number of pipeline stages). We have other ideas, so stay tuned! :-)

Pivot

Fast-forward to the end of the year. We shifted our focus to the Internet of Things and towards software programmers. Why? After all, we wasted spent almost three years trying to get professional hardware designers to use our product. Nicolas recently wondered about what the future holds for hardware engineers and had already warned semiconductor companies about the fear of change and its actual cost. In fact, we actually got more interaction with a handful of coders with a software background than with hundreds of hardware designers, despite having been focused exclusively on the latter (going to conferences, networking, sending emails, calling, having meetings, etc.).

For 2015, the plan is to get funding (again!), except that this time we have a strategy that is much, much more promising than selling to a few big semiconductor companies like pretty much any other traditional EDA company. This strategy has given more results in two months than what we did before during more than 30 months.

We are Synflow and we're saying "goodbye EDA, hello IoT"! We allow people, startups, and larger companies to design hardware and chips for the Things of the IoT. Cheers!