April 16, 2007
Virtual Hardware Models - Carbon Design Systems
Please note that contributed articles, blog entries, and comments posted on EDACafe.com are the views and opinion of the author and do not necessarily represent the views and opinions of the management and staff of Internet Business Systems and its subsidiary web-sites.
Jack Horgan - Contributing Editor

by Jack Horgan - Contributing Editor
Posted anew every four weeks or so, the EDA WEEKLY delivers to its readers information concerning the latest happenings in the EDA industry, covering vendors, products, finances and new developments. Frequently, feature articles on selected public or private EDA companies are presented. Brought to you by EDACafe.com. If we miss a story or subject that you feel deserves to be included, or you just want to suggest a future topic, please contact us! Questions? Feedback? Click here. Thank you!

Would you discuss Carbon’s product for IP distribution?

We want to enable companies to distribute Carbon models to their customer in order to get an early start on writing software. Let’s say it is a semiconductor company who wants to give their customers the ability to either start evaluating their IP early or to start writing software drivers. Those are the typical cases. What they need is a model of their chip or their system, essentially an executable spec that is cycle accurate that they can provide which people can use standalone or on their own platform. We are trying to create a model ecosystem where once these companies have their chip or system validated they are able in a secure manner to provide this to their customers for early access. What do they need to distribute a model? In the past the only thing you had was encrypted RTL models for particular simulators. What was the problem with that? If you had an RTL model and encrypted it for a particular simulator, it meant that your customers had to be using the same simulator to decrypt and run the model. It was not universal. It was very vendor specific. Secondly, since it was an RTL encrypted model running on an RTL simulator, it was inherently slow. Tens or maybe hundreds of Hertz for any kind of chip design. It was just too slow for people to develop software on. It was not fast enough. It was vendor specific. That’s what held that use paradigm back. It took a technology like Carbon to come along and solve these problems. We had the ever fast mode. By the way the models do not have to be event accurate. RTL simulators are event accurate. Software drivers do not need event accuracy, they need cycle accuracy. That’s the other thing; it (the solution) had to be at the right level of abstraction. Carbon came along with its technology. The models are cycle accurate, fast, available for all the common platforms and secure. Secure is different than what the other simulation vendors have done in the past. They encrypted their models but that was never a widely accepted method for people distributing IP. People are always concerned about encryption methods. Would people be able to decrypt it and find the original source? That was another lynchpin. The models had to be secure but not secure by the old means of vendor encryption. What does Carbon offer? They generate cycle accurate models. It is machine object code. There is no source code to find in the model. It is derived from and cycle accurate to the RTL but there is no source to recover from the machine object. So we solved the security problem. It is fast, cycle accurate meaning the right level of abstraction, and it is secure. Also Carbon plugs into all the popular platforms either running standalone or running in any SystemC or ESL platform
environment. What else is needed? There has to be a business model that allows people to distribute these models.

The business model has to allow customers to distribute this on a broad basis. Over a number of years Carbon has come to a business model that makes sense for its customers. We have a program that allows people to buy a license to deploy models. They can deploy their modes by buying into a Carbon program. They can distribute their models. It is not relegated on a part number basis. They can distribute their models to any of their customers, as many as they want. It’s like an annual kind of license. They get the license from Carbon to deploy the models and they can distribute them to whomever they want.

Does the customer’s customer have to pay anything for these models?

We certainly do not impose anything on the customer. Different vendors don’t obligate one way or the other on whether they charge for these models or not. I would say by and far our customers do not charge their customers for models. I would not say it is an absolute. Our customer certainly can charge if that is their business model. We do not require that. Our history has been that our customers use this as an enabler of their business. So they typically provide these models for free.

How many customers do you have?

Carbon has on the order of 25 to 30 customers worldwide. These are big name customers like AMCC, NEC and Toshiba. To give you a more interesting metric out of the top 10 semiconductor companies in the world 5 of them are our customers.

The mechanical world has talked about concurrent engineering, simultaneous engineering for more than 20 years. I have slides that show the benefits of doing software simulation in terms of faster time to market, lower development cost, and better products. On your website there is a comparison between your approach and the traditional approach.

I have been in EDA since 1983. I have seen similar diagrams and pictures from vendors through the years. I think what is different now is that if you contrast what people did in the past, typically when they talked about concurrent engineering, software/hardware engineering the hardware view was that you had to wait until you had the full chip done in terms of RTL. Then you dump that into a emulator, a QuickTurn type of box. Then you would start to develop software on that emulated hardware either before silicon or during the fab of that first prototype hardware. The people felt that they were doing things concurrently then they had been doing in the past of just waiting for silicon. People definitely got value out of emulation technology. The problem with that is that they had to wait until the full RTL was done, meaning stable RTL just before they were ready to go to fab before they could drop it onto the emulator. Because it took time to bring up the emulator you had to have things pretty well right before you put it on the emulator otherwise you were wasting your time. Emulation was very late in the design process. There were certainly a lot of companies, tools and customer’s money put into using that technology. There was value in that concurrent design methodology. What has changed? It is not to say that people do not use emulation technology. Today I would say that there is less use of the traditional emulators. The FPGA prototypes are in the same vein, in terms of mapping to a custom or vendor prototype. What has changed is that Carbon allows you to create a software virtual prototype. The whole idea of concurrent engineering can start a lot earlier. Now we are using hardware/software concurrent engineering a lot earlier in the process, all the way back to the architectural phase. The difference is that early on you have mostly heterogeneous, meaning you have all different levels of models, from multiple source. You can derive a lot of value from this heterogeneous model of your system, if you use the Carbon models through the entire process,
early on through hardware development and through software development. Firmware development can start a lot earlier because you do not have to wait until you have the final system ready to tapeout or go to the emulator. You can start a lot earlier. That’s the pull for the last 5 years for ESL platforms. Start earlier, verify and validate things as early as possible before tapeout. There has been a gradual shift to these platforms. Carbon makes these platforms complete. People now have a complete modeling environment that they did not have before.

If you look at the various pitches from the ESL vendors about the value of doing early simulation, the pitch is very attractive. Why hasn’t ESL been more successful?

There is a relatively simple answer to that. Maybe it is self-serving. The first if not the second question to these vendors would be: The benefits sound great. The analysis tools look very nice. But how do I bring in my RTL? Whether their design is 25%, 50% or 80% existing RTL, there is some percentage of RTL there that the customer requires to bring in to start a new design. None of these ELS platforms have a little button that says RTL import. If you do not have that, then you have a major hole in your platform. Carbon recognized that early on. If we can fill that hole, then we can not only become part of every major ESL platform, we can fulfill the mission of ESL. All
of these vendor’s pitches are by and far the same pitch, concurrent design, do it earlier, get it done quicker. We can really fulfill the promise. What Carbon is really doing to bring RTL into these environments in a natural way, automatically generated models whether in Verilog, VHDL or both languages. He has requirement to use that for his next design as well as his new designs is going to be implemented in RTL. If you are able to compile that into high speed cycle accurate models for these platforms then you have a complete environment. Without you just don’t.

« Previous Page 1 | 2 | 3 | 4 | 5  Next Page »

You can find the full EDACafe event calendar here.

To read more news, click here.

-- Jack Horgan, EDACafe.com Contributing Editor.


Review Article Be the first to review this article


Featured Video
Peggy AycinenaWhat Would Joe Do?
by Peggy Aycinena
Acquiring Mentor: Four Good Ideas, One Great
More Editorial  
Sr. staff ASIC Design Engineer -2433 for Microchip at San Jose, CA
Manager, Field Applications Engineering for Real Intent at Sunnyvale, CA
SENIOR ASIC Design Engineer for TiBit Communications at Petaluma, CA
Upcoming Events
DeviceWerx - 2016 at Green Valley Ranch Casino & Resort Las Vegas NV - Nov 3 - 4, 2016
2016 International Conference On Computer Aided Design at Doubletree Hotel Austin TX - Nov 7 - 10, 2016
ICCAD 2016, Nov 7-10, 2016 at Doubletree Hotel in Austin, TX at Doubletree Hotel Austin TX - Nov 7 - 10, 2016
Electric&Hybrid Aerospace Technology Symposium 2016 at Conference Centre East. Koelnmesse (East Entrance) Messeplatz 1 Cologne Germany - Nov 9 - 10, 2016
S2C: FPGA Base prototyping- Download white paper

Internet Business Systems © 2016 Internet Business Systems, Inc.
595 Millich Dr., Suite 216, Campbell, CA 95008
+1 (408)-337-6870 — Contact Us, or visit our other sites:
AECCafe - Architectural Design and Engineering TechJobsCafe - Technical Jobs and Resumes GISCafe - Geographical Information Services  MCADCafe - Mechanical Design and Engineering ShareCG - Share Computer Graphic (CG) Animation, 3D Art and 3D Models
  Privacy Policy