High-performance multicores providing strong consistency guarantees speculatively reorder memory operations. If a memory reordering is seen by other cores, speculative operations are squashed and re-executed. This talk presents the concept of non-speculative and invisible reordering of memory operations, the foundation on which the ECHO project (an ERC Consolidator Grant) is based on. The talk offers a background about memory-level speculation in current multicores. Then, it shows that, for the case of the load-load reordering, is not necessary to squash and re-execute reordered loads to guarantee the load-load order. It also presents a case for store-store ordering, which allows for the first time cost-effective store coalescing with strong consistency. Finally, it discusses solutions for secure speculative execution.