The Internet is already an integral part of today's business and technical (chip design, SW development) environments. Intranets are the backbone of IT in modern enterprises. They link the computers, which in turn run the applications that keep a business alive. Synopsys uses the Internet for a variety of purposes including making corporate information available, marketing, online support via knowledge bases and virtual consultants and electronic software transfer.

As far as the future use for CAE is concerned, the main impediments to solve are licensing, the legal aspects of security and use across national borders and, on the technical side, speed/bandwidth and appropriate software. The technical problems in particular are being addressed by the Next-Generation Internet, which predicts point-to-point speeds between 100 Mbits per second to 1 Gbit per second by 2002. At those speeds, the Internet would be faster than today's intranets.

On the application side, the Internet2 initiative is looking at revolutionary innovations in the areas of learningware, digital libraries, tele-immersion and virtual laboratories. Each of these four areas has direct applications to CAE. Our conclusion is that the Internet is not only a reasonable but also an increasingly essential tool for CAE.

Computer-aided design (CAD) tools are routinely used for electronic component and system design. New challenges in electronic design automation (EDA) are related to the large number of tools (and user interfaces) that need to be applied to a design, and to the geographical dispersion of designers, tools and design data. In the last few years, software technologies have been moving fast to enable the development of distributed, user-friendly applications across the Internet. Real-world distributed EDA applications are not evolving as fast as practical needs and enabling technologies.

Experience in using of the Web for EDA was gained from PPP, a design environment for low-power digital circuits that integrates EDA tools running on heterogeneous, distributed platforms under a common user interface based on the World Wide Web (WWW). Users may access the features offered by a large and diverse set of remote CAD tools using their own WWW browsers. PPP was conceived to show the feasibility of WWW-based tool integration in the EDA area and to provide a real-world case study for future research in this field. It uses only well-established software technologies for the base implementation of PPP.

Lev Markov, Mentor Graphics Corp.
Whether the Internet is a reasonable tool for CAE depends on the particular application. CAE has several unique features, which have to be addressed quite differently. First of all, designers deal today in their design environments with a number of very large, complex tools, some of which are used in the interactive fashion. For such environments, the Internet can not bring a lot of benefits. Also, the levels of abstraction and CPU intensity of these tools have a significant effect on their Internet opportunities. The higher the level of a tool's advancement, or of its CPU complexity, the lower its chances for Internet application. On the other hand, design reuse is a critical factor for modern designs, and the Internet can bring a lot of value in this matter. In order to make it successful several issues like IP copyrights, IP provider liability, privacy and data security, and various types of standards have to be resolved. On some of these issues we are quite advanced, but on others we are far from even starting.

Also, there are a number of potential Internet applications related to design data management, as well as to online training and customer support. These applications are the obvious candidates for the Internet environments.

The Internet is already supporting CAE in some ways and it is important to look at this issue in more detail, segmenting the problem, to understand it well. One possible segmentation is into four different areas:

• Support — pre- and post-sale support, technical product support, bug submission and tracking
• "Framework" — online documentation, tool flow and other methodology aids, and job control and tracking
• Delivery — Internet delivery of tools, "push" technology, library updates, pay-per-use pricing models
• Tool architecture — changing tools to integrate the Web into the middle of the tool
• The Internet, specifically the Web, has made huge inroads into the Support part of the EDA business. All the major EDA suppliers utilize the Web heavily. Cadence has an open Web site with information for prospective customers, including a Web-zine, as well as a private area for current customers where they can get more in-depth information about product status.

There are a lot of issues about the environment in which EDA tools exist, here called the Framework. The Internet has started to make inroads into on-line documentation, with most companies at least moving towards Web-based documents. Job control and methodology/flow tools are a natural place where the Web and Java based tools could be first applied. There are some companies working this angle now.

The Delivery of EDA tools is certainly doable over the Internet. The problem is one of business model more than technology and this area will take some time to work itself out. Incremental library updates will probably be an area that leads in the use of the Web. You can see some evidence of this from the IP suppliers, such as Xilinx, providing access to some forms of IP over the Web.

Perhaps the most difficult area is in the architecture of the EDA tools themselves. With such large complex tasks to perform, tool vendors will have to analyze carefully what it means to adopt the Internet. If you look into an application you will find an architectural structure that was created with file systems and sockets within an individual machine. To break this apart using Web or other protocols will require a careful performance analysis. Separating at least parts of the GUI from the performance critical engine is one partition that seems feasible.

Within Cadence we heavily rely on our intranet for our design documents and we see our customers doing the same thing. Design data and code tends to still be located on a local file system (that is accessed through NFS or equivalent). The weakest link for use of the Internet today is its performance. Until it becomes similar to file access (which is going to be difficult), the Internet will continue to be carefully applied within EDA. The strongest feature is interoperability and ubiquitousness. Since everyone has a Web browser, there is little or no learning curve. The Web will continue to make inroads into support and framework areas based on this strength.

Five years from now, I expect the Internet to dominate Support, Framework, and Delivery and to be integrated where appropriate within Tool architecture. The higher the performance of the Internet in the future, the more inroads we will see into Tool architecture. Agents (independent software programs that can migrate over the Net to perform local tasks) will start to make themselves felt in the Framework and Delivery areas in particular. As agent technology matures, expect to see designers begin to rely on electronic assistants that are relatively unimaginative but very persistent and running jobs, collecting results, and summarizing the state of a large design.