Why does four large contexts have low latency hiding ability compared to many smaller contexts?
Thanks!
unique_name
I believe it is because we can run fewer threads at a time (only four here, compare to the previous slide with 16), and this means that we can't interleave as much while we wait for data to come.
I think I missed this point during lecture.
Why does four large contexts have low latency hiding ability compared to many smaller contexts?
Thanks!
I believe it is because we can run fewer threads at a time (only four here, compare to the previous slide with 16), and this means that we can't interleave as much while we wait for data to come.