| 24 | 24 | Thus, we divide the code, optimistically, into independent parts called superblocks that represent subsets of the execution graph. Each superblock doesn't overlap with other superblocks in accessed variables, and represents a long sequence of instructions, including branch statements, that commonly execute in this pattern. Since a branch instruction has taken and not taken paths, the superblock may contain one or both of the two paths according to the frequency of using those paths. For example, in biased branches, one of the paths is often considered; so it is included in the superblock, leaving the other path outside the superblock. On the other hand, in unbiased branches, both paths may be included in the superblock. Therefore, a superblock has multiple exits, according to the program control flow during its execution. A superblock also has multiple entries, since a jump or a branch instruction may target one of the basic blocks that constructs it. The parallelizer module orchestrates the construction of superblocks and distributes them over parallel threads. However, this may potentially lead to out-of-order execution of the code, which we address through STM concurrency control. I/O instructions are excluded from superblocks, as changing their execution order affects the program semantics. |
