The result of the read from src is not affected by

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: strates one of the challenges of evaluating program performance. Measurements can be strongly affected by characteristics of the data and operating conditions that initially seem insignificant. 5.12. BRANCH PREDICTION AND MISPREDICTION PENALTIES Function combine1 combine1 combine2 combine3 combine4 combine5 combine6 Page 211 211 212 217 219 234 241 Method Abstract unoptimized Abstract -O2 Move vec length Direct data access Accumulate in temporary Unroll ¢ Unroll ¢½ Unroll ¢ , Parallelism ¢¾ Unroll ¢ , Parallelism ¢ Unroll ¢ , Parallelism ¢ Integer + * 40.14 47.14 25.08 36.05 19.19 32.18 6.26 12.52 1.76 9.01 1.51 9.01 1.25 9.01 1.19 4.69 1.15 4.12 1.11 4.24 36.2 11.4 Floating Point + * 52.07 53.71 37.37 32.02 28.73 32.73 13.26 13.01 8.01 8.01 6.32 6.32 6.33 6.22 4.44 4.45 2.34 2.01 2.36 2.08 22.3 26.7 249 Worst:Best Figure 5.28: Comparative Result for All Combining Routines Running on a Compaq Alpha 21164 Processor. The same general optimization techniques are useful on this machine as well. 5.11.2 Changing Platforms Although we presented our optimization strategies in the context of a specific machine and comp...
View Full Document

This note was uploaded on 09/02/2010 for the course ELECTRICAL 360 taught by Professor Schultz during the Spring '10 term at BYU.

Ask a homework question - tutors are online