Proposal for Uniied System Design Meta Ow in Task-level and Instruction-level Design Technology Research for Multi-media Applications
چکیده
This paper describes an attempt to bring together the many di erent system design ows existing in architecture and system design technology research, into a more abstract but unifying meta ow. Many existing system and architecture design ows have a strong resemblance and unnecessary overlap. Mainly due to a lack of a common and consistent terminology coupled to a common reference basis, it is now nearly impossible to compare and reuse (sub)steps. In addition, there is a too strong separation between research in di erent communities. To alleviate this problem, we introduce a more abstract but unifying meta ow which attemps to bridge the gap between the existing ows. From this meta ow, a particular design ow can be instantiated for a given application (domain) by leaving out the non-required stages/steps, by selecting a (sub)step sequence which is compatible with the partial metaow order, and by selecting the appropriate technique for all remaining (sub)steps (e.g. the type of scheduler). This paper focuses on the principles at the taskand instruction-level abstractions. It also provides an illustration of the power of the metaow principles for a realistic multi-media compression demonstrator from the MPEG4 context. 1 Motivation and objectives Many di erent design ows (or partial ows) exist in the system and architecture design literature (see e.g. [2, 3, 8]), in many cases intended also for di erent application domains. Many of these ows have resemblances but it is usually di cult to pin-point where the corresponding parts are and where the di erent (sub)steps are situated. This is for a large part due to incompatible terminology and unnecessary details or lack of unambiguous information which obscure the global picture. In order to allow more \reuse" of knowledge and information, it is vital to overcome this problem. For this purpose, a common terminology basis is a rst step. The basic Professor at K.U.Leuven; Work partly sponsored by the MEDEA-IWT project SMT. high-level synthesis terminology proposed in [11, 12] was useful but not su cient to cover the rapidly advancing research. Moreover, in order to really use a common terminology there is also a need to have a shared view of an underlying system design meta ow. Recently, the main trust in the design technology community has moved into the \system" and \software" areas and the underlying system design ow basis is largely lacking still. At IMEC, an attempt has been made to remedy this situation. The resulting proposal is (to our knowledge) the rst attempt to obtain uni cation over such a broad scope. Part of the results are summarized in this paper, focused on the task-level and instruction-level abstractions and illustrated by applications from the multimedia domain. We believe that the scope of validity of the full document contains all digital electronic systems. The targeted application domains are very broad but the main focus is on end-user telecom, multimedia processing, telecom protocols and telecom network components. Heterogeneous system architectures (\systems-on-a-chip") are necessary to meet the system requirements for these application domains, encompassing instruction-set (programmable) and customized processors but also storage, control and interface components (see gure 1).