Could anyone give an example for the last point on this slide? - "Computations where the correspondence between tasks and data is less clear"
lament
Maybe this is an example - maybe:
Say you write a task that processes one piece of data, then depending on the result, is suppose to access or modify another piece. In general this might not be good parallelizeable code, but if somehow it is, then the programmer would have to reason about long, complex chains of conditionals as to what data a task would access.
Could anyone give an example for the last point on this slide? - "Computations where the correspondence between tasks and data is less clear"
Maybe this is an example - maybe: Say you write a task that processes one piece of data, then depending on the result, is suppose to access or modify another piece. In general this might not be good parallelizeable code, but if somehow it is, then the programmer would have to reason about long, complex chains of conditionals as to what data a task would access.