English: Nowadays, most modern high performance processors employ out-of-order (O3) execution. In these processors, instructions are executed as soon as possible increasing in this way the instruction level parallelism (ILP) and, in consequence, the processor performance. However, not all instructions could be executed in O3 way. Memory access instructions sharing the same memory address must be executed in order to keep the original program semantic. For this reason, O3 processors use memory dependence predictors. These are specialized units in charge of reducing, as much as possible, the number of loads and stores executed in-order. Good predictors aid to release all the ILP potential in O3 processors. This project studies current used (in commercial hardware) and proposed (in academic papers) methods for predicting memory dependencies in an O3 processor. New opportunities to exploit instructions locality and improve predictor¿s accuracy are proposed and tested. In particular, the concept of extreme locality is introduced and applied in a new method, named MiniCAM. The results using this method are presented and discussed.
All rights reserved. This work is protected by the corresponding intellectual and industrial property rights. Without prejudice to any existing legal exemptions, reproduction, distribution, public communication or transformation of this work are prohibited without permission of the copyright holder. If you wish to make any use of the work not provided for in the law, please contact: email@example.com