Energy efficient code generation for processors with exposed datapath

In a modern processor architecture the register ??le (RF) consumes considerable amount of power. Therefore it is important to reduce the RF accesses when designing an energy e??cient architecture. It is well-known that with datapath exposed to software, the transport-triggered architectures (TTAs) can substantially reduce the RF tra??c. In this paper, we analyze the potential of using MOVE-Pro, a TTAbased processor architecture. And we propose the compiler back-end for MOVE-Pro which can generate code that saves energy consumption by performing energy aware instruction scheduling to reduce RF accesses. The proposed architecture and compiler design is exible. In the experiments we compare the proposed architecture with a RISC processor with the same resource, and achieve a reduction of RF accesses by up to 80%, which results in up to 11% saving in total core power. Meanwhile the dynamic cycle count remains almost the same as the reference processor, which means energy is saved without compromising performance.

[1]  H. Nakayama,et al.  A 2.5 GFLOPS 6.5 million polygons per second 4-way VLIW geometry processor with SIMD instructions and a software bypass mechanism , 1999, 1999 IEEE International Solid-State Circuits Conference. Digest of Technical Papers. ISSCC. First Edition (Cat. No.99CH36278).

[2]  Scott A. Mahlke,et al.  AnySP: Anytime Anywhere Anyway Signal Processing , 2010, IEEE Micro.

[3]  Jarmo Takala,et al.  Reducing processor energy consumption by compiler optimization , 2009, 2009 IEEE Workshop on Signal Processing Systems.

[4]  Jarmo Takala,et al.  Impact of Software Bypassing on Instruction Level Parallelism and Register File Traffic , 2008, SAMOS.

[5]  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).

[6]  Victor V. Zyuban,et al.  The energy complexity of register files , 1998, Proceedings. 1998 International Symposium on Low Power Electronics and Design (IEEE Cat. No.98TH8379).

[7]  Yifan He,et al.  Xetal-Pro: An ultra-low energy and high throughput SIMD processor , 2010, Design Automation Conference.

[8]  Stamatis Vassiliadis,et al.  The TM3270 media-processor , 2005, 38th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO'05).

[9]  Guang R. Gao,et al.  An investigation of the performance of various instruction-issue buffer topologies , 1995, Proceedings of the 28th Annual International Symposium on Microarchitecture.

[10]  Henk Corporaal Microprocessor architectures - from VLIW to TTA , 1997 .

[11]  Henk Corporaal,et al.  Code generation for transport triggered architectures , 1994, Code Generation for Embedded Processors.

[12]  William J. Dally,et al.  Operand Registers and Explicit Operand Forwarding , 2009, IEEE Computer Architecture Letters.

[13]  Johan Janssen,et al.  Compiler Strategies for Transport Triggered Architectures , 2001 .

[14]  Jarmo Takala,et al.  Customized Exposed Datapath Soft-Core Design Flow with Compiler Support , 2010, 2010 International Conference on Field Programmable Logic and Applications.