Enhancing Microkernel Performance on VLIW DSP Processors via Multiset Context Switch

High-performance and low-power VLIW DSP processors are increasingly being deployed in mobile devices to process video and multimedia applications. The diverse applications of such systems has led to recent research efforts focusing on their resource management and kernel scheduling. In this paper, we address the enhancing the performance of the microkernel for a VLIW DSP processor, called PAC architectures. In order to reduce the number of read and write ports in register files of VLIW architectures, so as to reduce both the power consumption and implementation costs, a distributed register file and multibank register architectures are being adopted in PAC architectures. These methods present challenges for microkernel designs in terms of reducing context switch overhead. In our work, we propose a multiset descriptor mechanism with compiler support to reduce the context switch overheads associated with the use of registers. The experiments were done with the microkernel system called pCore which has an efficient and tiny design that prunes its code size down under 11 Kbytes. Experimental results show that our multiset context-switching mechanism may reduce the context switch overhead up to 30%.

[1]  Y. Danieli Guide , 2005 .

[2]  Anant Agarwal,et al.  APRIL: a processor architecture for multiprocessing , 1990, ISCA '90.

[3]  William J. Dally,et al.  A mechanism for efficient context switching , 1991, [1991 Proceedings] IEEE International Conference on Computer Design: VLSI in Computers and Processors.

[4]  Volker Barthelmann Inter-task register-allocation for static operating systems , 2002, LCTES/SCOPES '02.

[5]  Jenq Kuen Lee,et al.  Compiler Supports and Optimizations for PAC VLIW DSP Processors , 2005, LCPC.

[6]  Burton J. Smith Architecture And Applications Of The HEP Multiprocessor Computer System , 1982, Optics & Photonics.

[7]  Jenq-Kuen Lee,et al.  0RC2DSP: compiler infrastructure supports for VLIW DSP processors , 2005, 2005 IEEE VLSI-TSA International Symposium on VLSI Design, Automation and Test, 2005. (VLSI-TSA-DAT)..

[8]  Chein-Wei Jen,et al.  A novel register organization for VLIW digital signal processors , 2005, 2005 IEEE VLSI-TSA International Symposium on VLSI Design, Automation and Test, 2005. (VLSI-TSA-DAT)..

[9]  Robert A. Iannucci Toward a dataflow/von Neumann hybrid architecture , 1988, ISCA '88.

[10]  Jenq Kuen Lee,et al.  Compiler optimization on VLIW instruction scheduling for low power , 2003, TODE.

[11]  Jenq Kuen Lee,et al.  Copy Propagation Optimizations for VLIW DSP Processors with Distributed Register Files , 2006, LCPC.

[12]  Anoop Gupta,et al.  Exploring The Benefits Of Multiple Hardware Contexts In A Multiprocessor Architecture: Preliminary Results , 1989, The 16th Annual International Symposium on Computer Architecture.

[13]  LeeJenq Kuen,et al.  Compiler optimization on VLIW instruction scheduling for low power , 2003 .

[14]  Marios C. Papaefthymiou,et al.  Efficient Block Scheduling to Minimize Context Switching Time for Programmable Embedded Processors , 1999, Des. Autom. Embed. Syst..

[15]  Jenq Kuen Lee,et al.  Compilers for leakage power reduction , 2006, TODE.

[16]  William J. Dally,et al.  Register organization for media processing , 2000, Proceedings Sixth International Symposium on High-Performance Computer Architecture. HPCA-6 (Cat. No.PR00550).

[17]  David B. Whalley,et al.  Fast context switches: compiler and architectural support for preemptive scheduling , 1995, Microprocess. Microsystems.

[18]  Jenq Kuen Lee,et al.  Interprocedural probabilistic pointer analysis , 2004, IEEE Transactions on Parallel and Distributed Systems.

[19]  H. J. Chairman-Siegel Proceedings of the 15th Annual International Symposium on Computer architecture , 1988 .

[20]  Shau-Yin Tseng,et al.  PAC DSP Core and Application Processors , 2006, 2006 IEEE International Conference on Multimedia and Expo.

[21]  B. Zolfaghari A dynamic scheduling algorithm with minimum context switches for spacecraft avionics systems , 2004, 2004 IEEE Aerospace Conference Proceedings (IEEE Cat. No.04TH8720).

[22]  A. Gupta,et al.  Exploring the benefits of multiple hardware contexts in a multiprocessor architecture: preliminary results , 1989, ISCA '89.

[23]  Susan J. Eggers,et al.  The effectiveness of multiple hardware contexts , 1994, ASPLOS VI.