Resource Assignment in a Compiler for Transport Triggered Architectures
نویسندگان
چکیده
The paper describes how resource assignment is performed in a compiler for transport triggered architectures, and in particular, how busses are assigned to data transports. This becomes an important issue when function units and register files are only partially interconnected via an irregular network of busses. Reducing this connectivity is important for processors exploiting instruction level parallelism; it may not only reduce the required chip area, but may also shorten the processor cycle time and therefore increase the performance. The bus assignment problem is solved by transforming it to a bipartite matching problem for which efficient algorithms exists. Experiments have been performed to measure how the cycle count depends on the connectivity between function units and register files. Results show that a significant amount of connectivity from a fully connectedconfiguration can be removed before the cycle count starts to increase.
منابع مشابه
Evaluating transport triggered architectures for scalar applications
In transport triggered architectures (TTAs) the programming and operational model is mirrored when compared with regular RISC and VLIW architectures; instead of programming operations which cause data transports as side eeects, in TTAs the transports are programmed, where a transport may trigger an operation if necessary. Transports are therefore visible at the architecture level, and are compl...
متن کاملEnergy Efficient Code Generation for Processors with Exposed Datapath
In a modern processor architecture the register file (RF) consumes considerable amount of power. Therefore it is important to reduce the RF accesses when designing an energy efficient architecture. It is well-known that with datapath exposed to software, the transport-triggered architectures (TTAs) can substantially reduce the RF traffic. In this paper, we analyze the potential of using MOVE-Pr...
متن کاملOn Efficiency of Transport Triggered Architectures in DSP Applications
The trend in programmable architectures for digital signal processing (DSP) is to move towards high-level language programming, which sets high requirements for compilers to efficiently exploit the instruction level parallelism in modern processors. In this paper, efficiency of transport triggered architectures (TTA) in DSP applications is discussed. The efficiency of a high-level compiler on a...
متن کاملRegister Bank Assignment for Spatially Partitioned Processors
Demand for instruction level parallelism calls for increasing register bandwidth without increasing the number of register ports. Emerging architectures address this need by partitioning registers into multiple distributed banks, which offers a technology scalable substrate but a challenging compilation target. This paper introduces a register allocator for spatially partitioned architectures. ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 1996