IBM developerWorks has a new article up describing the complexity of profiling on the POWER5 processor. “At any clock cycle, you have to handle a typical situation of five instructions/group, 20 groups past dispatch, 32 outstanding loads, 16 outstanding misses, two independent threads, and the decoupled nest/core.” The author discusses oprofile on Linux, curiously providing a ksh script to run it… but then all is made clear as they spend most of the article discussing AIX tools.

Read more…

Sorry, the comment form is closed at this time.

© 2000 - 2011 Suffusion theme by Sayontan Sinha