He gave me the “you are a crazy person” look - the COSSAP acquisition was still fairly recent as I recall and it was about five years after Mentor bought Microtec Research. Software acquisition was not going to be the popular choice at that point. But I explained that many of his customers would wind up using Simulink (or Matrixx as it was then) if they weren’t already. And there was a developing ecosystem around it, although a number of those companies had to be rescued by FPGA makers and other hardware companies with an interest in the tools.

Today, oddly enough, I’d probably give a different answer, largely on the basis of cost versus future revenue. The problem is recommending another company. One remarkable aspect about embedded software is that there really isn’t a leader with a handle on the kinds of markets that matter to a company such as Synopsys.

The RTOS companies have, strange as it may seem, limited insight into what customers are doing in many of their markets. Those with a good insight are often operating in specialty markets such as military or automotive that don’t carry over into consumer. Some just seem to be confused. And none of them have a very good story on multicore development, which is surely the one place that they should be focusing. The people doing a lot of the running there are actually hardware players or have more of a hardware background - ARM, Tensilica, Critical Blue - plus the people who actually ship hardware and have an interest in shipping N cores, where N is a large number, such as Clearspeed, Intel and nVidia.

And it’s worth noting that Synopsys has an operation that is software focused - the virtual prototyping group. Does the company need to add an RTOS offering to that? I think not - in fact, it might kibosh deals it already has.

Not sure Synopsys needs to buy an RTOS vendor. Maybe ARM should. Most HW blocks as well as many Tensilica processors that are used as HW blocks (we call them dataplane processors) don’t need an OS, just firmware.

Paula

Wind River is a company that immediately came to my mind as well, being as they are an RTOS leader. They may also prefer to go after a smaller player like Express Logic which has a lighter weight RTOS and which they can morph more easily into their organization without the baggage of a large sales force.

Another interesting possibility would be Tensilica. Chris Rowen is from Synopsys, so the DNA is there. If one could figure a way to automatically analyze C-code and build a Tensilica Xtensa architecture optimized for the instruction stream of that code, that would be very powerful.