The OMI program has started with the first call for proposals at the end of 1991 and has from 1992 onwards been managing an ever increasing number of projects and has been building a portfolio of enterprises and scientific institutes that are part of the OMI community. The OMI program aimed to identify and take account of technological advances and trends in the market for microcomponents such as the move to greater on-chip systems integration, the need for simpler architectures, the wish to avoid redesign through portability, and the need for easy upgrading. OMI aimed to converge with accepted technology and standards rather than compete with them, and to help Europe build critical mass by recognizing and consolidation its strengths. The need to embrace all sectors of the microprocessor systems industry and for compatibility across platforms requires OMI to be open, both technically and organizationally, and this philosophy governs the program.
The program has centered on several programs: ANDF (Architectural neutral distribution format), the Eurocell library and interfaces (ELI), operating systems, software implementation methodology, microprocessors, and application demonstrators.
RISC as an emerging technology: The CISC technologies had established themselves solidly as the technology of choice for the desktop and various other business computing and information management platforms. The restricting licensing policy of Intel as the dominant player in this marketplace has reinforced over the last 5 years the position of Intel and a few other non-European companies as the dominant suppliers. The RISC technology has been gaining momentum in the application areas where less of a base of legacy applications were available or where performance justified migration. Notably the workstation market that emerged in the 80s became the domain of the RISC architecture, as well as the rapidly growing environment of embedded applications. It would be overstating the situation to say that RISC is in the process of replacing CISC but it is true to say that RISC is taking an increasing share of the microcomponents market.
The "RISC shipment overview" shows the shipment pattern of the various RISC microprocessor architectures over the last 5 years. It should be kept in mind that the microcomponents market is due to its trend of integration on the chip with other logic functions difficult to completely and correctly analyse and the table does not as a result include processor cores that were part of ASICs that were sold as a standard products nor does it include ASIC designs that are built into finished industrial products based on licensing arrangements between the technology supplier and the maker of those application products. As a result the numbers could for certain designs not give the true state of affairs but is nevertheless indicative of the trend. This table points to a high growth rate of ARM, but even greater for MIPS and the Hitachi SH. One should take into account that most of the ARM technology finds its way into industrial products through licensing arrangements where the ARM core becomes an integral part of the ASIC that is integrated into the product. The number of processors that find its way into the market via this route is not included into the table and is as well true for some of the other designs. As the degree to which this is true is not easy to determine one has to be careful to draw simple conclusions from this table.
Integration of logic functions: The integration of logic functions on the chip has been a market trend driven by the progress of manufacturing technology that lets the "law of Moore" do its work by doubling the amount of transistors on a unit of silicon each 18 months. This is making more space available on the die and the application industry is creating the demand by the uptake of the technology for a higher level of integration and lower power requirements in designing products for the mobile individual in the information society. As a result processor cores, peripheral logic, analog circuits, DSP logic and various other elements of application logic are being integrated on single chips, increasing significantly the demand for design tools and knowledge for the application industry to hold and strengthen their position
Open systems: Open systems have been boosted by the activities that have germinated around the internet. The internet started as an interconnection method of isolated desktop systems. Now it has gone from the academic environment into industry and is finally entering the homes of individuals in large numbers. It has started to create a new computing paradigm. Software development and standardization efforts have lead to a development environment where instead of static elements of information that are exchanged over the internet, more and more active software components are entering this domain. The battle ground for the next few years starts to be defined and will be fought on two fronts. Firstly the concept of a virtual computer which is at the heart of the JAVA technology will potentially open the door for general purpose computing on a much larger variety of microcomputer architectures as is the case today. Secondly the paradigm of a world connected and exchanging active components in a heterogeneous environment (the web is the computer) could seriously challenge the position of shrink wrapped monolithic applications. It could open the door to a virtual applications environment that will reshuffle the deck in the software market.
Both elements will accelerate the trends toward a more heterogeneous microprocessor environment and demand for ASICs for all the information appliances that the webbed world could service.
Embedded systems: The embedded applications have become part of our daily lives as an invisible force. We can say that the modern human being is encountering more microdevices today in a normal day as human beings. In telecommunications, office automation, consumer electronics, automotive and various other industrial processes of today, microcomponents are the hidden workers serving us from dawn to dusk. The embedded market is not ruled by frozen allegiances to certain technologies as there is much less of a legacy issue. Most products are designed for specific applications, and performance and cost requirements are often the factors that determine commercial viability. As a result in this exploding market there is much room for staking out market share driven by software and hardware innovation.
Application orientation: There is an application orientation trend that emerged in the overall computing arena from the computer systems market down to the development of microdevices. In the 80s after first the PC and later the technical workstation were introduced, each of them started to contribute to a general democratization of computing. The availability of abundant compute power through the PC, to a much larger group of users was forcing both standardization and user friendliness. The technical workstation opened the use of high-level design tools to almost each and every engineer and created a similar drive towards user friendliness and standardization. Although the level of standardization leaves still much to be desired it has led to an environment where the overall focus from end-users is moving away from compute-technical questions to solving practical application questions. In this paradigm it is often more important to master the industrial application knowledge than the computer knowledge. The company that has a long experience in the application and is able to absorb the ICT knowledge is often better placed than the company that has strong ICT knowledge but is a newcomer to the application area. Europe has strong areas of application orientation and can in this new paradigm recoup some of its lost territory.
Microprocessors have been at the heart of the work done by OMI and many projects have been aimed at helping the European industry to recover lost territory and establish beachheads in a number of application areas from where to expand. It has not attempted this through a frontal attack on the general purpose microprocessor but has worked with a number of RISC oriented designs and has nurtured in the projects seamless co-operation between the software and the hardware design companies and the application developers to create designs and uptake at the same time. Many projects have had SGS Thomson as a participant, who is one of the strong European participants is the market for microcomponents. After the acquisition of Inmos by SGS -Thomson they have become the owner of a promising design that has proven already its success in a number of application areas as can be seen further in the document. Siemens semiconductor is another of the big European players in this market and has also been an active participant in a number of projects. They have both 16 and 32 bit designs that are showing its success in telecommunications and automotive applications. Philips semiconductor as the number one company in Europe in the microcomponents market has been a participant in a number of projects but has been mostly involved in projects that were oriented toward standardization of software and methodology as well as the design of ASICs. Philips is in the use of microprocessors and microcontrollers mostly focusing on U.S designs through the large presence of Philips semiconductor in the United States. Promising emerging designs from Europe are the ARM of Advanced Risc Machines that has through design wins in products such as the Apple newton started to establish its reputation. Since then the success has continued as can be seen in the section about Advanced Risc Machines. Hyperstone from Germany is a relatively new participant who does not yet show in the ranking as their momentum started only recently to build, and shipments are still modest. Hyperstone has an innovative RISC design with DSP functionally combined with RISC and should be an increasing participant in this market.
and software methodology is another important element to be a viable player in the market for microprocessors, microcontrollers and embedded applications and is a domain where OMI has spawn a number of projects. The aim of OMIs program is to advance the open systems state-of-the-art in a rapid and cost-effective fashion. Among the benefits envisaged are improved partitioning and modularity, which will reduce the complexity of porting to new processor architectures. Scalability and better support for distributed computer architectures are aimed to cope with the increasingly distributed computing environment. OMI projects are also producing operating systems for embedded applications development. A number of companies and scientific institutes have been active in the design and development of operating systems technology, interconnection protocols, software design methods, compiler developments, and design verification software.
The OMI projects have also aimed at the increasing complexity of designing software for a multitude of different processor platform and the initiatives in the area of ANDF (architecture neutral distribution format) has been helping to create an environment where much less resources have to be allocated to porting and verification procedures. ANDF starts from the premise that software is built in the format of binary code for a virtual machine that is than by means of platform oriented compilers converted into machine specific code. This approach simplifies the task of the developer into development for one target machine (the virtual machine) and leaves the second step to the processor specific implementors. This approach is similar to the one that has been development by SUN Microsystems for the implementation of the JAVA virtual machine. OMI did not succeed to establish ANDF as a standard nor was any of its partners able to establish itself using this approach as SUN managed to do. Elements of this design approach however were adopted by software companies working within OMI projects.
Companies that have been active in the software projects are Chorus Systèmes from France, Etnoteam from Italy, Tasking software from Holland, Eonic systems, Iona and smaller ones such as Advanced Bytes and Rights, Performance research and Visual tools. The software work has been done in concert with the hardware designers and the application implementors. A few of those companies are present in a further section of the document.
Library and interconnection is another strategic area of activity of the OMI projects. The concept of an Eurocell library has emerged as an OMI equivalent to the macrocell library concept that is a common practice among the semiconductor manufacturers, foundries and design houses. Where the industry uses the macrocell library mostly in a proprietary sense limiting it to its own sphere of influence OMI is aiming at utilization across a broad industry sector. OMI is aiming to deal with the IPR issues in such a way that macro cells would be available as the result of projects and would be disseminated as freely as possibly to stimulate the uptake by other companies. The concept has resulted in two interconnect technologies for integration between macrocells: PI-bus and HIC (Heterogeneous interconnect). HIC covers the on- and off-chip interconnection in all ranges of microprocessors as well as communication interconnect in backplanes, between cabinet and locations.
: In many basic software or hardware design projects there is an important element aimed to prove and exploit the technology through application demonstrators.
Prototype systems are used to demonstrate developing technology to users and OMI participants. Application demonstrators are also used to show how a technology will be used in its target commercial environment, a requirement that OMI has laid down for most of its projects. From the outset, regular workshops have been held to develop common perspectives between technologists and users in the systems industries, and feedback is encouraged from the start of a project.
Dissemination and exploitation of a project is a vital element in the evaluation of projects and is often a point of discussion along the duration of the project between the project partners and external reviewers and OMI project officers. The issue of IPR and commercial viability are no easy problems but because of the strong objective of OMI to assure as much uptake as possible from its projects there is continuous attention to this area.
Exploitation and dissemination. The OMI sponsored projects are under close supervision of the responsible project officers within ESPRIT who monitor from the first evaluation of a proposal to the completion of a project that the project fulfills basic criteria for exploitation and dissemination. There is an absolute requirement on the project partners to show that the result of the project in the form of products or new software or hardware technology is being put to commercial use by one or more of the project partners. There is attention paid to the fact that the information about the project should be accessible freely by the industry.If there are IPR issues that might be an inhibiting factor, OMI addresses those with the consortium from the early phase of the project and is trying to find a solution with the consortium to have as much as possible of the inhibiting factors removed during the project.
Building an industry glue. The process to build an infrastructure of relationships and beachheads of knowledge through the OMI sponsored projects is also a focus of the program. Many of the projects end up with consortium partners that extend their working relationship far beyond the actual duration of the projects. The activities of the user networks, the OMIMO office, the OMI yearly conference and many formal and informal mechanisms, all contribute to the building of informal and formal relationships and the exchange of information. The objective of this glue is to make sure that private enterprises can stretch beyond their current level of competence and utilize technology that they might otherwise not have been ready for yet.
The European challenge in microdevices has been unfolding and the achievements of OMI so far have indicated that Europes ambition in the design of RISC processors, software and tools and the utilization of all those ingredients into the design of application specific ASICs with embedded software are realistic. A long road has still to be traversed but as can be seen in the document in the section analysing the microcomponents market for Europe and the role that European companies are playing in this market, improvement of the market position is taking place. Since a long period of a sliding percentage of marketshare held by European companies in their home market, the year 1995 was the first year that an improvement in the marketshare took place. It would be presumptuous to claim that this would be only thanks to OMI but I thinkone can say that OMI has delivered its contribution to this result. Other contributing factors have been the catch-up that took place in the technological level of semiconductor manufacturing capability of the European industry (where also ESPRIT projects have contributed), as well as many innovative efforts by the industry itself.