An analysis of the amount of global level redundant computation in the SPEC 95 and SPEC 2000 benchmarks

Authors:
J. J. Yi;D. J. Lija
Affiliations:
Dept. of Electr. & Comput. Eng., Minnesota Univ., Minneapolis, MN, USA;Dept. of Electr. & Comput. Eng., Minnesota Univ., Minneapolis, MN, USA
Venue:
WWC '01 Proceedings of the Workload Characterization, 2001. WWC-4. 2001 IEEE International Workshop
Year:
2001

Citing 0
Cited 2

Partial resolution for redundant operation table

Microprocessors & Microsystems
“Look it up" or "do the math": an energy, area, and timing analysis of instruction reuse and memoization

PACS'03 Proceedings of the Third international conference on Power - Aware Computer Systems

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper analyzes the amount of global level redundant computation within selected benchmarks of the SPEC 95 and SPEC 2000 benchmark suites. Local level redundant computations are redundant computations that are the result of a single static instruction (i.e. PC dependent) while global level redundant computations are redundant computations that are the result of multiple static instructions (i.e. PC independent). The results show that for all benchmarks more than 90% of the unique computations account for only 1.2% to 31.5% of the total number of instructions. In fact, less than 1000 (0.14%) of the most frequently occurring unique computations accounted for 19.4%-95.5% of the dynamic instructions. Furthermore, more redundant computation exists at the global level as compared to the traditional local level. For an equal number of unique computations-approximately 100 for each benchmark at both the global and local levels, the global level unique computations accounted for an additional 1.5% to 12.6% of the total number of dynamic instructions as compared to the local level unique computations. As a result, exploiting redundant computations through value reuse at the global level should yield a significant performance improvement as compared to exploiting redundant computations only at the local level. 7.