Profiling and Optimizing Micro-Architecture Bottlenecks on the Hardware Level

Most mechanisms in current superscalar processors use instruction granularity information for speculation, such as branch predictors or prefetchers. However, many of these characteristics can be obtained at the basic block level, increasing the amount of code that can be covered while requiring less space to store the data. Furthermore, the code can be profiled more accurately and provide a higher variety of information by analyzing different instruction types inside a block. Because of these advantages, block-level analysis can provide more opportunities for mechanisms that use this information. For example, it is possible to integrate information of branch prediction and memory accesses to provide precise information for speculative mechanisms, increasing accuracy and performance. We propose BLAP, an online mechanism that profiles bottlenecks on the micro-architectural level, such as delinquent memory loads, hard-to-predict branches and contention for functional units. BLAP works on the basic block level, providing information that can be used to optimize these bottlenecks. A prefetch dropping mechanism and a memory controller policy were created to use the profiled information provided by BLAP. Together, these mechanisms are able to improve performance by up to 17.39% (3.9% on average). Our technique showed average gains of 13.14% when evaluated with higher memory pressure due to higher prefetch aggressivity.