Direct addressed caches for reduced power consumption

ABSTRACT

A direct addressed cache is a hardware-software design for an energy-efficient microprocessor data cache. Direct addressing allows software to access cache data without a hardware cache tag check. These tag-unchecked loads and stores save the energy of a tag check when the compiler can guarantee an access will be to the same line as an earlier access. We have added support for tag-unchecked loads and stores to C and Java compilers. For Mediabench C programs, the compiler eliminates 16-76% of data cache tag accesses, with half of the benchmarks avoiding over 40% of the data tag checks. For SPECjvm98 Java programs, the compiler eliminates 18-63% of data cache tag checks. These tag check reductions translate into data cache energy savings of 9-40%, and overall processor and cache energy savings of 2-8%.

REFERENCES

• 1: T. Burd. Energy-Efficient Processor System Design. PhD thesis, University of California at Berkeley, 2001.
• 2: C. A. Moritz et al. Hot pages: Software caching for RAW microprocessors. MIT-LCS Technical Memo LCS-TM-599, August 1999.
• 3: J. Montanaro et al. A 160-MHz, 32-b, 0.5-W CMOS RISC microprocessor. IEEE JSSC, 31(11):1703-1714, November 1996.
• 4: M. Rinard et al. The FLEX compiler infrastructure. 1999-2001. http://www.flex-compiler.lcs.mit.edu.
• 5: M. S. Lam et al. The SUIF compiler system. 1992-2001. http://www-suif.stanford.edu.
• 6: Osman S. Unsal , Raksit Ashok , Israel Koren , C. Mani Krishna , Csaba Andras Moritz, Cool-cache for hot multimedia, Proceedings of the 34th annual ACM/IEEE international symposium on Microarchitecture, December 01-05, 2001, Austin, Texas
• 7: S. B. Furber et al. ARM3 - 32b RISC processor with 4kbyte on-chip cache. In G. Musgrave and U. Lauther, editors, Proceedings IFIP TC 1O/WG 10.5 Int. Conf. on VLSI (VLSI'89), pages 35-44. Elsevier (North Holland), 1989.
• 8: Walter Lee , Rajeev Barua , Matthew Frank , Devabhaktuni Srikrishna , Jonathan Babb , Vivek Sarkar , Saman Amarasinghe, Space-time scheduling of instruction-level parallelism on a raw machine, ACM SIGPLAN Notices, v.33 n.11, p.46-57, Nov. 1998
• 9: Intel Corp. Intel Xscale core developers manual, order no. 273473-001 edition, December 2000.
• 10: Johnson Kin , Munish Gupta , William H. Mangione-Smith, The filter cache: an energy efficient memory structure, Proceedings of the 30th annual ACM/IEEE international symposium on Microarchitecture, p.184-193, December 01-03, 1997, Research Triangle Park, North Carolina, United States
• 11: R. Krashinsky. Microprocessor energy characterization and optimization through fast, accurate, and flexible simulation. Master's thesis, Massachusetts Institute of Technology, May 2001.
• 12: A. Ma, M. Zhang, and K. Asanović. Way memoization to reduce fetch energy in instruction caches. ISCA Workshop on Complexity Effective Design, July 2001.
• 13: C. A. Moritz, M. Frank, and S. Amarasinghe. Flex-cache: A framework for flexible compiler generated data caching. Lecture Notes of Computer Science, Springer-Verlag, 2001.
• 14: M. Muller. Power efficiency & low cost: The ARM6 family. In Hot Chips IV, August 1992.
• 15: Ramesh Panwar , David Rennels, Reducing the frequency of tag compares for low power I-cache design, Proceedings of the 1995 international symposium on Low power design, p.57-62, April 23-26, 1995, Dana Point, California, United States [doi>10.1145/224081.224092]
• 16: M. Zhang and K. Asanović. Highly-associative caches for low-power processors. Kool Chips Workshop, MICRO-33, 2000.