Mathworks was successful in deploying a core tool, Matlab, and built a business model around it. If one follows the new support and improvements in their portfolio since last year, one can easily notice how disconnected the opensource software community is from the real world.
Their support to the AutoSar environment since the beginning (with new package embedded coder) has mostly given many startups around the world, the opportunity to collaborate in the AutoSar platform. Each of these startups was cautious and surveyed the EDA industry before engaging into the AutoSar platform. I was even contacted to give inputs of FEL’s solutions to satisfy this demand.
It is very interesting to see from a top level point of view how the vendor is working and how the user is making use of it to spin out astonishingly good design methodologies.
The opensource community will pronounce either “octave” or “scilab” everytime the word “matlab” roams around. But it is surprising no one has actively stepped in to port or spread the opensource word to those who published their models on Matlab’s File Exchange. One will be surprised that octave actually meets roughly 80% compatibility with those published models.
Matlab/Octave/Scilab are all a very good example of “what I called Vendor Lock Up tools”. I use these applications to model the behaviour of parts of my hardware and irrespective of their license, I’m locked up when my design needs to interface others simulators. I curse these tools as they are either not TCL based or have broken TCL API between releases.
If you look at Canonical, they are trying to adopt such a business model with their recent announcement on Cloud Computing for their next release. However, for the
- ubuntu community, such adoption does not give weight to “Linux for humans” and is not simple end user oriented. Go to an Ubuntu booth at any future OSS events, you will notice how disconnected their speech will be from their initial driven roadmap.
- commercial users: the commercial users (90% of the Semiconductor companies) have already denied a debian based linux distribution for their lack of innovation in the Virtualisation environment, lack of long term (20 years) commitment, lack of experience in the semiconductor industry to elevate vendors’ tools performance and the learning curve of such deployment never takes less than 2 years for the semiconductor industry. During those 2 years, the “deb” community has changed several times their opinions and for the commercial user experiences Vendor Lock Up and uncertainty. Assuming the cycle of ASIC project from specifications to tape-out is 6 months,the design center’s EDA/CAD department will be doing fulltime and assisting designers for vendors’ related tools and loosen their time on methodologies to reduce this development cycle.
Red Hat Enterprise Linux still remains the number one choice for the Semiconductor industry and the only one pulling the opensource wagon. With rumours stating that Windows 7 can seduce EDA vendors, I am very excited to follow this industry trend and how everyone is coping with others’ decisions. We have been assuming for a couple of weeks now, Synopsys is the now the world EDA leader. I’m curious to read the papers that they have presented at DVCon. Is it only for the sake of title or being an innovation leader ?
Chitlesh Goorah
