What exactly is fine-grain sharing in terms of caching?
pdp
@blah329: If multiple processors are working on different threads of the same program and they have the same working set, its easier to have a single cache and have a fine-grained sharing of the data between threads in the same cache.
srb
Are implementations common that are a combination of the two extremes of each processor having its own cache, and one global cache existing for all processors? Maybe one cache per every four processors, or something like this?
themj
Shared caches can have some benefits especially if all of the processors are accessing the same data. Then, one processor can pre-fetch some lines and then all of the processors can benefit.
What exactly is fine-grain sharing in terms of caching?
@blah329: If multiple processors are working on different threads of the same program and they have the same working set, its easier to have a single cache and have a fine-grained sharing of the data between threads in the same cache.
Are implementations common that are a combination of the two extremes of each processor having its own cache, and one global cache existing for all processors? Maybe one cache per every four processors, or something like this?
Shared caches can have some benefits especially if all of the processors are accessing the same data. Then, one processor can pre-fetch some lines and then all of the processors can benefit.