Thread Level Speculation, also known as Speculative Multithreading, or Speculative Parallelization, is a technique to speculatively execute a section of computer code that is anticipated to be executed later in parallel with the normal execution on a separate independent thread. Wikipedia