ALFA Fine Grain Data
نویسنده
چکیده
This paper presents a new version of ALFA VER92], a ne grain dataaow machine. This machine uses the static dataaow execution model and is constituted by 128 clusters; each cluster groups 128 identical Functional Units (FU) with homogeneous I/O conditions. The peculiarity of this machine is the possibility to map directly part of dataaow graph programs in hardware for the simple FU design VER94b]. Since ALFA does not support the conventional processor cycle , at cluster level its behavior is asynchronous and determinate when part of a dataaow graph is mapped and executed on it. After compile time, in ALFA, instructions and data are no longer related, and no control token is generated during the computation, but only data tokens. Consequently, during the execution phase no memory is required to contain the partial results exchanged among FUs of a same cluster.Furthermore, in this paper we explain the reasons and the architectural choices of this static dataaow machine and we also introduce the architecture of the new communication system.
منابع مشابه
Experience with Fine-Grain Communication in EM-X Multiprocessor for Parallel Sparse Matrix Computation
Sparse matrix problems require a communication paradigm different from those used in conventional distributed-memory multiprocessors. We present in this paper how fine-grain communication can help obtain high performance in the experimental distributed-memory multiprocessor, EM-X, developed at ETL, which can handle fine-grain communication very efficiently. The sparse matrix kernel, Conjugate G...
متن کاملIntroduction and discussion about three types of fine grain products and their harvesters
Fine grain products are harvested in two manual and mechanized forms. Nowadays, common combine harvesters are used, but fine grain products such as Rapeseed, Sesame and Sorghum cannot be harvested with conventional heads of combine harvesters due to the excessive grain loss in the head of the combine. Therefore, the necessity of design and fabricating machines and heads of combine harvesters be...
متن کاملIntroduction and discussion about three types of fine grain products and their harvesters
Fine grain products are harvested in two manual and mechanized forms. Nowadays, common combine harvesters are used, but fine grain products such as Rapeseed, Sesame and Sorghum cannot be harvested with conventional heads of combine harvesters due to the excessive grain loss in the head of the combine. Therefore, the necessity of design and fabricating machines and heads of combine harvesters be...
متن کاملEfficient Fine-Grain Synchronization on a Multi-Core Chip Architecture: A Fresh Look
Multi-core chip architectures are becoming mainstream, permitting increasing on-chip parallelism through hardware support for multithreading. Fine-grain synchronization is essential to the effective utilization of the capacity provided by future high-performance multi-core architectures. However, there are also new challenges realizing such fine-grain synchronization in large-scale multi-core c...
متن کاملPerformance Analysis of Enhanced Fine–grain Multithreaded Distributed–memory Systems
In fine–grain multithreading, the thread changes in each processor cycle, consecutive instructions are thus issued from different threads, and no data dependencies stall the pipeline. Enhanced fine–grain multithreading maintains a number of additional threads which are used to replace an active thread when it initiates a long–latency operation. Performance improvements due to enhanced multithre...
متن کامل