Application-specific clustered VLIW datapaths: early exploration on a parameterized design space

V.S. Lapinskii1, M.F. Jacome1, G.A. de Veciana1
1Department of Electrical and Computer Engineering, University of Texas, Austin, Austin, TX, USA

Tóm tắt

Specialized clustered very large instruction word (VLIW) processors combined with effective compilation techniques enable aggressive exploitation of the high instruction-level parallelism inherent in many embedded media applications, while unlocking a variety of possible performance/cost tradeoffs. In this work, the authors propose a methodology to support early design space exploration of clustered VLIW datapaths, in the context of a specific target application. They argue that, due to the large size and complexity of the design space, the early design space exploration phase should consider only design space parameters that have a first-order impact on two key physical figures of merit: clock rate and power dissipation. These parameters were found to be: maximum cluster capacity, number of clusters, and bus (interconnect) capacity. Experimental validation of their design space exploration algorithm shows that a thorough exploration of the complex design space can be performed very efficiently in this abstract parameterized design space.

Từ khóa

#VLIW #Space exploration #Clocks #Algorithm design and analysis #Application specific processors #Registers #Computer architecture #Parallel processing #Time factors #Radio frequency

Tài liệu tham khảo

10.1109/DAC.1992.227830 10.1109/ISSS.1995.520616 dixit, 2001, performance speculations—benchmarks, friend or foe, Seventh Int Symp High Performance Computer Architecture 10.1109/MICRO.1997.645830 10.1023/A:1019799515784 10.1145/375977.375978 10.1109/MICRO.1996.566472 10.1109/MICRO.1992.697033 10.1109/ISSS.1998.730607 10.1109/ICCD.1999.808603 wolfe, 2000, High Performance Compilers for Parallel Computing 10.1007/978-1-4757-2849-1 10.1007/978-1-4615-5107-2 agarwal, 2000, Clock rate versus IPC: the end of the road for conventional microarchitectures, Proceedings of 27th International Symposium on Computer Architecture (IEEE Cat No RS00201) ISCA, 248 10.1109/HPCA.2000.824366 10.1109/MICRO.1994.717418 10.1109/EDAC.1992.205918 10.1109/MICRO.1995.476825 10.1145/378239.379051 10.1109/40.149733 10.1007/978-1-4615-2307-9 lapinskii, 2001, Algorithms for compiler-assisted design space exploration of clustered VLIW ASIP Datapaths faraboschi, 1998, Clustered instruction-level parallel processors 10.1109/43.240076 10.1109/ICPP.1999.797422 basoglu, 2000, The MAP-CA VLIW-Based Media Processor 2000, TMS320C6000 CPU and Instruction Set Reference Guide 0, ADSP-TS001M TigerSHARC DSP Product Description 10.1109/ICCAD.2000.896523 burger, 1997, guest editors' introduction: billion-transistor architectures, IEEE Trans Computes, 30, 46 marwedel, 1995, code generation for embedded processors, Kluwer International Series in Engineering and Computer Science 10.1007/978-1-4757-6422-2 10.1023/A:1018743102645 10.1007/978-1-4419-8720-4 10.1109/MICRO.1997.645806 10.1109/ISHLS.1994.302348 10.1145/157485.164920 10.1109/SUPERC.1990.130118 10.1109/ICCAD.1994.629828 10.1109/54.844333 faraboschi, 2000, Lx: a technology platform for customizable VLIW embedded processing, Proceedings of 27th International Symposium on Computer Architecture (IEEE Cat No RS00201) ISCA, 203 kathail, 2000, HPL-PD Architecture Specification Version 1 1 abraham, 2000, Fast design space exploration through validity and quality filtering of subsystem designs 0, NOVA Project ASIP s and retargetable compilers CAD for embedded systems Dept ECE Univ Texas at Austin 10.1109/92.766750 10.1109/ASAP.2000.862383 ramakrishna rau, 2000, Embedded computing New directions in architecture and automation 10.1109/ISSS.1999.814268 aditya, 2000, Automatic architecture synthesis and compiler retargeting for VLIW and EPIC processors