GCCS97 # 56 Proceedings of the GCC Developers Summit 2007

By ols.108.redhat.com. that the bits of the scheduler should keep state be localized in this context. 4.1 Initial performance tuning at the end of 2006, we have begun to work on performance tuning of the designers. This work done is based on the Itanium architecture. As a result, we have made a number of enhancements for both the IA-64 backend and the scheduler itself. The back-end updates include changes direction, a better placement of stop bits, and memory dependence tweaks. The standard function approachfor IA-64 is now set to 64 bytes, and the standard loop alignment is 32 bytes. This is consistent with the values of the Intel compiler uses consistent. The function corresponds to the orientation Icache line size, orientation, and the loop will ensure that the processor front-end



http://www.youtube.com/watch?v=2jRFL372Bj0&hl=en