Speculative execution is anoptimization technique where acomputer system performs some task that may not be needed. Work is done before it is known whether it is actually needed, so as to prevent a delay that would have to be incurred by doing the work after it is known that it is needed. If it turns out the work was not needed after all, most changes made by the work are reverted and the results are ignored.
Modernpipelinedmicroprocessors use speculative execution to reduce the cost ofconditional branch instructions using schemes that predict the execution path of a program based on the history of branch executions.[2] In order to improve performance and utilization of computer resources, instructions can be scheduled at a time when it has not yet been determined that the instructions will need to be executed, ahead of abranch.[4]
Eager execution is a form of speculative execution where both sides of the conditional branch are executed; however, the results are committed only if the predicate is true. With unlimited resources, eager execution (also known asoracle execution) would in theory provide the same performance as perfectbranch prediction. With limited resources, eager execution should be employed carefully, since the number of resources needed growsexponentially with each level of branch executed eagerly.[6]
Predictive execution is a form of speculative execution where some outcome is predicted and execution proceeds along the predicted path until the actual result is known. If the prediction is true, the predicted execution is allowed to commit; however, if there is a misprediction, execution has to be unrolled and re-executed. Common forms of this includebranch predictors andmemory dependence prediction. A generalized form is sometimes referred to as value prediction.[7]
Runahead is a technique that allows acomputer processor to speculatively pre-processinstructions duringcache miss cycles. The pre-processed instructions are used to generate instruction anddata streamprefetches by executing instructions leading tocache misses (typically calledlong latency loads) before they would normally occur, effectively hidingmemory latency. In runahead, the processor uses the idle execution resources to calculate instruction and data stream addresses using the available information that is independent of a cache miss. Once the processor has resolved the initial cache miss, all runahead results are discarded, and the processor resumes execution as normal. The primary use case of the technique is to mitigate the effects of thememory wall. The technique may also be used for other purposes, such as pre-computing branch outcomes to achieve highly accuratebranch prediction.[8]
Lazy execution is the opposite of eager execution, and does not involve speculation. The incorporation of speculative execution into implementations of theHaskell programming language, a lazy language, is a current research topic.Eager Haskell, a variant of the language, is designed around the idea of speculative execution. A 2003 PhD thesis madeGHC support a kind of speculative execution with an abortion mechanism to back out in case of a bad choice calledoptimistic execution.[9] It was deemed too complicated.[10]
Starting in 2017, a series ofsecurity vulnerabilities were found in the implementations of speculative execution on common processor architectures which effectively enabled an elevation ofprivileges.
