Is the head always guaranteed to be the largest piece of work? In QuickSort this is the case, however, not in general. In that case, do we want to steal the oldest job or the biggest job?

It seems like the PARALLEL_CUTOFF is 25 so I would have expected Thread 1 to have also added 126-150 to the stack and worked on 101-125. Am I missing something?

Sorry, ignore the above question, the next slide clarified it for me

^^^ I have the same question as @Martingale -- this looks like it works great for divide-and-conquer algorithms like quicksort, but what about other types of algorithms?

Would there be some way to implement a multiple queue system similar to implementation in modern operating systems where multiple queues exist depending on priority/workload? I feel like otherwise there would be unnecessary delay for shorter tasks