An example of a read-write conflict would be one transaction performing a read to a node where another transaction is writing to, right?
So seeing this is the benefit of transactional memory that instead of completely locking some (in this case) node for all accesses you just determine if there are critical reads and writes? In the previous slide it showed two transactions as going down from path 1 -> 2 -> 3 for one transaction or 1 -> 2 -> 4 for the other, but although they have conflicting paths the actual reads/writes are non-conflicting.
An example of a read-write conflict would be one transaction performing a read to a node where another transaction is writing to, right?
So seeing this is the benefit of transactional memory that instead of completely locking some (in this case) node for all accesses you just determine if there are critical reads and writes? In the previous slide it showed two transactions as going down from path 1 -> 2 -> 3 for one transaction or 1 -> 2 -> 4 for the other, but although they have conflicting paths the actual reads/writes are non-conflicting.