The video recording of the lecture ends on this slide while Kayvon is in the middle of a sentence, so I wonder if the remaining of the lecture covers much after this point.
If each core supported two hyper-threads, then, would there be any benefit of having the two threads on separate cores?
Discussion for 5 threads (from Piazza): OS needs to schedule what threads run on available hardware execution contexts. When one thread is blocked by memory request, OS can do context switching to save its execution contexts to memory and then swap in execution context for another available thread to run into the CPU hardware execution context. OS needs to decide when to swap threads and what threads to swap.
https://piazza.com/class/kts8r2wwk6p5gs?cid=179
Please log in to leave a comment.
Also, putting two threads on separate cores might lead to fault tolerance as at least 1 thread will be running when 1 core fails.