The first review of the Esprit/COSY project was held in Brussels on the 3 July 1998.

This is the objective of the COSY project. It really corresponds to a need for the future both for electronic consumer and industrial products. Most of application domains are concerned by such an evolution.

The COSY project will deliver a complete design methodology and tools supporting it for SOCs based on the use of functional and architectural IPs. According to my knowledge, such a result does not exist today. Very few tools exist for IPs reuse because it is a very new approach and standards such as VSI are not completely formalised. Existing tools are very specific and solve some problem sub-sets as they consider mostly one dedicate architecture with its specific components and not a generic architecture.

Companies involved in constructing complex embedded systems from silicon subsystem components are the potential end-users. High-volume and very short time-to-market constraints are probably periquisites for the need of such a technology. It means that " small " companies and/or companies developing prototypes and specific user products are not currently concerned by the results of this project. The potential market appears large : consumer electronics, computers, telecommunications, automotive electronics, medical equipement, etc.

Success in such developments imposes to follow a well-defined design process according to policies for designers, component vendors and component manufacturers. This gives less opportunity for invalid components, leads to reduce time to market, produces right-first-time designs, manufactures low-cost integrated products.

As the design methodology and the tools developed in the project are perceived to be new developments having no current competition, the COSY results are intended to revolutionize the way in which electronic system design is carried out today. COSY aims at reducting the gap between the customer and a solution supplier. Time-to-market will be reduced. Components and so sub-system or even system designs will be reusable in the form of IPs for other customers. Therefore, it contributes to constitute the backbone of an IP market. Solutions will be better and more optimized as COSY proposes a solution to easily explore and evaluate different hardware/software solutions.

It is not clear today what will be the result : a stand-alone product, a collection of tools which can be individually integrated in a CAD framework. However, the participation of the major EDA vendor CADENCE is appropriate for a complete success.

In the first phase of the project, each partner clarifies its background technologies for the other partners and useful for the achievement of the project. At the end of this phase, the whole product will be normally clearly identified and specified. This will lead to clarify what tools (independent or not) are under development or will be developed, their role in the complete design flow, their functionalities, their rationale for potential users.

The competition will really exist as the market of embedded systems and SOCs will tremendously increases in the next future. Today, it is difficult from my position to have an opinion on competitors and emerging similar products. The consortium has been built to be on the leading edge : the first worldwide EDA vendor (tools and services provider), two very strong european companies PHILIPS and SIEMENS. This industrial force is strongly connected to 2 well-known research groups : UPMC and PT.

The consortium has to clarify what will be the product at the end of the project and the added-value of COSY.

CEL as the first EDA vendor offers tools, environments as well as design services. The ALTA group as a Business Unit of CADENCE is well placed to identify the potential market of its new tool named FELIX. Key partners have been already identified with PHILIPS and SIEMENS as two of them. This product released during 1998 will give a fast feedback on its interest. The largest current interest comes from wireless communication which is more advanced in Europe than in US or Japan. Therefore the Cadence’s objective is to deeply penetrate the european market.

PHILIPS is a worldwide leading supplier of ICs for consumer electronic equipements. The company intends to exploit the results of the COSY project to develop generic CPU/RISC architectures for future semiconductor devices as SOCs ; to design, manufacture and sell system solutions for applications in set-top boxes and for Digital Video Broadcasting. The market is estimated very high in these domains.

If the IP market really emerges, the risk of substitution products is very low. The main risk is the non-acceptance by the market of the COSY solution, which may prefer to develop their own design methodology and integrate a set of the appropriate and specific tools selected on the market to constitute their own development platform and so be more independent of EDA vendors.

A complete workpackage WP7 is dedicated to Exploitation plan & Dissemination with 2 man-years effort. The strategy is to address this aspect both by each partner individually and through a coordinating team for the joint exploitation of the outcome of the projects.

The Task concerning Exploitation Plan is decomposed into three phases : In the first period, definition of an initial exploitation plan ; then an intermediate exploitation plan ; and at the end the final exploitation plan.

The Task concerning Dissemination aims at inform customers and companies. Two major events have been proposed : an OMI/COSY workshop ; DAC/COSY special Session. These events normally are planned to attract interest, to inform customers and capture early their feedbacks.

Customer market of the IP-based design methodology: Is there a wide acceptance of the IP approach? Is the IP (VSI) standard enough formalized and stabilized to invest time in IP development ?

Customer perception of the different tools: Are tools open enough and interoperable with others for external companies to be able to specialize them for their own needs?

Company cost investment: Will it be possible and quite easy to reasonably foresee the cost of investment for such a solution?

The two partners PHILIPS and SIEMENS are in the project to help answering these questions by showing that the return on investment is very positive for themselves and so for other companies.

The objectives are clearly stated in the Project Programme. The result will be a design flow and methodology, the necessary supporting tools and models for system design, the re-use of hardware and software IPs.

The project corresponds to a real need. Designing with few tools or inappropriate tools is not acceptable for more and more complex embedded systems. The reusability of third-party standard components is compulsory for efficiency, cost and time reduction. The solution for components is to be virtual, so the need of IPs and tools to support their use.

The partners appear complementary in the project : Research teams and industrials for the design flow, the methodology and the specification of the supporting tools ; an EDA vendor for tools development and deployment on the market ; two companies to specify the need, demonstrate the usefulness of the solution. Each partner also brings its background technologies to create an initial starting point.

It is always difficult to assess the potential return. However, it is felt that this project is being undertaken by the parties with the intention of making a real, saleable product which fits with each of their declared corporate strategies. The return on this project development will be estimated progressively through the development of the exploitation plan.

The periodic progress report is clear and gives a good overview and the progress of each task. Deliverables are also clear, concise and respond to their objectives. The decomposition of the project into 7 workpackages seems relevant to the final objective. The first workpackage WP1 covers the first year with the objective of technology assessment for all partners and specification of the methodology and supporting tools. If at a first glance, this period appears long, I consider it judicious to create a real and fruitful synergy between partners. This is a necessary condition to obtain a complete and coherent solution instead of loosely-related products and solutions.

Tasks were carefully constructed to ensure progress towards the overall project target. The project is following correctly the task decomposition.

With the participation of research teams, particularly PT as an associate partner with CEL, the first part is built on relevant concepts. The approach is quite pragmatic. Each partner has made the effort to present his technology on the basis of these concepts. These concepts will be also useful to specify the methodology and the supporting tools available at the end of the project, and also to disseminate outside the project.

The method followed by the consortium sounds technical and professional.

Firstly, the result of Task 1.3 will be a good element to measure the reachability of the project goal. This task produces Deliverable 1.3.1 which has to contains the Requirements for COSY. This means that the document will be able to " virtually " reflect the solution at the end of the project. At the time of the review, this document is normally established (Deadline on 23/04).

Secondly, the demonstration of tools efficiency is also important for the success of the project. Therefore it is important to consider the demonstrators as soon as possible. Then the quality of the applications considered for demonstration has to be appropriate to prove the adequation of the solution. The adequation criterium is not defined by the partners only but mainly by other companies.

The report explains that up to 80% of the objectives of the first period have been reached. Some tasks have been delayed and so 2 deliverables have been postponed :

It seems that the shift for these deliverables is not essential for the rest of the project.

The plan of some tasks has been also updated : T1.2.6, T1.4.3, T1.10.1, T7.1.1 ; some started earlier, some are a bit delayed. Normally the following deliverables will be elaborated according to the plan.

Therefore, the project is progressing a bit lower than expected. This is due to the need and the time necessary to establish the strong cooperation between partners. We can suppose that this step is now achieved and the consortium will be efficient to work on the final requirements on COSY and then the development of the tools.

The small modifications on the plan are justified in the document. Some tasks started earlier to obtain results which have been identified as essential in other projects.

Additional effort has been allocated to Project Management that was underestimated for the sart of the project.

Deliverable D.1.2.2 has been given at the review meeting.

• the assessment of the Felix technology by Philips ;
• the assessment of the Felix technology by Siemens ;
• a Q&A list concerning Felix Alpha Rel 5.

In Part 1, Philips relates the use of Felix to perform functional and performance verification for the DVB/VLD benchmark. The document recalls the example with the functional description, the architectural template and the 5 different mappings.

All the benchmarks have not been done. Functional simulation is done for 2 benchmarks and no performance simulation due to time and to bugs in the current Felix software. The document describes the 3 tasks to be done : capture the behavior, capture the system architecture, map the behavior onto the architecture. The document mentions a set of difficulties and requirements needed for Philips. The main difficulty is the necessity to convert a legacy code into the " Fire and Exit " style of the Felix computation model.

In Part II, Siemens reports the use of Felix to perform the simulation of the Data Transfer Manager example (DTM). The solution is first recalled, then the document describes the use of the Felix tool. Simulation times are given for different benchmarks. It seems that Siemens had less problems to use the Felix software. Is it due to the nature of the example or for other reasons?.

Part III presents a Question and Answer list by Cadence. Questions are asked by Philips and Siemens. They are classified in categories : Modeling, Performance Analysis, Delay Script language, Mapping, Simulation. The questions give a flavor of the problems met by system designers. The answers give a quite clear overview of the designers’ problems and the possibilities and limitations of the tool.

• Initial effort in setting up the working teams and coordinating the project ;
• The introduction of a new application domain and its related benchmark for Siemens (Hard Disk Controller); 
• Delays in the development of the Felix platform for IP based System Design by Cadence.

The presentation during the review meeting followed the hereafter agenda :

The review presentation gives us a good impression : quality of slides and brievity of explanations. The successive presentations help us to highlight the project objectives and the means to achieve these objectives. The presentation of the ongoing specification tasks shows us the work currently done which will be presented at the next review.

CEL is enhancing the Felix computation model to describes simultaneously dataflow oriented and control-oriented problems. By defining his demonstrator, Philips has a clear objective to meet in the DVB domain. The demonstrator presented by Siemens seems not clearly related to the objective of COSY.

No specific comments.

No specific comments.

Nothing is written concerning the synergy with other projects. It may be considered as a lack or we can consider that this is too early. I think it is interesting to know if other projects are partly similar and if some synergy may be possible and planned in the future to increase results.

There is a strong desire by the consortium members to disseminate when possible. The progress report contains the effort already done in information dissemination in the WP7 activity.

The review meeting has been very useful to better understand the project, its objective, its state of progress, the consortium, the work to be done and why. The suggested recommendations have been briefly commented at the end of the meeting.

1- Requirements for COSY

This step and the corresponding deliverable D1.3.1 produced by T1.3 appear important for me because it will reflect what will be the results at the end and what will be the outcome for other companies in Europe. Avoid delays for that and produce a high-quality document. In this document the demontrators must also be very clearly specified with the identification of relevant problems not easily solved today and the corresponding results expected at the end of COSY.

Benchmarking must be more formalized : criteria and measurements before and without COSY tools and progressively during the project. A clear list of metrics to monitor for the 3 dimensions (performance, design, cost) must be identified and selected in order to prove the achievement of better and faster results with the COSY solution.

The partners must continue to complete the benchmarks and add results to the metrics. Selection of appropriate benchmarks for the objective of the project. Measurement of the time needed by designers today without COSY tools.

3- Siemens demonstrator

The demonstrator presented by Siemens does not show that it will be useful to prove the added value of the COSY project. The example seems quite obvious. Integrated chips exist today for the control of a steppermotor. The presentation does not indicate the use of the Felix tool for the production of the solution. We recommend : to modify or/and complexify the example ; specify the COSY tools which will be used to produce the solution and prototype. The following requirements must be satisfied :

• Use COSY tools starting from Felix as the front-end tool ;
• Produce a prototype or (system on a chip) better than existing solutions (criteria to be specified) ;
• The solution must include a mixture of data-flow and control flow to exploit and validate the enhancement of the Felix tool ;
• Normally build the solution by using functional and architectural IPs.

4- Felix tool

Cadence has to clarify the specific added value of the COSY project for the Felix product. What is specific for Felix related to the objective of COSY and not considered without COSY?.

5- Synergy with other projects

Other european projects are currently concerned by CoDesign. Therefore it appears important all along this project to relate the work done and to be done with what will be done in these other projects : similarities, differences, complementary results, etc.

The project management appears professional and efficient.

The requirements of the COSY results have to be clear : what design flow, what tools and to do what, the interoperability of them. It is important to immediately think with the customer in mind, the customer perception of the different tools, the customer acceptance.

No real major problem. Some recommendations indicated above to improve the project.