Just like AI, nothing worth a real prize. Look at what we expect, no compromise, machine learning is NOT real AI, similarly, what we have now is not real parallel. Can we build the parallel machine? Program it easily?
Why are we not satisfied with MPI? Usability...
Yes, we are looking forward to parallelism, we are looking forward to easy programming and lightning fast running.
But, a wish is a wish.
We did the same to Artificial Intelligence, look at it now, how it looks? Still, a wish.
IMHO, What really MATTERS is,
We do NOT know how to program efficiently with multicores, which in current phase, stands for most of the parallel platforms.
We do have parallel applications, BUT, they are painful to design and implement, and they are running on mainframes or clusters. They are written once, by specialists.
If we want a world of parallel machines and programs, MAKE THE PROGRAMMING MODEL SUITABLE. Make programming easy, with fine grained parallel, not just dispatching processes to cores, but making one process running on many cores WITH EASY PROGRAMMING.
If we have hardware security support, this is not that easy..
Just like AI, nothing worth a real prize. Look at what we expect, no compromise, machine learning is NOT real AI, similarly, what we have now is not real parallel. Can we build the parallel machine? Program it easily? Why are we not satisfied with MPI? Usability...
Yes, we are looking forward to parallelism, we are looking forward to easy programming and lightning fast running. But, a wish is a wish. We did the same to Artificial Intelligence, look at it now, how it looks? Still, a wish.
IMHO, What really MATTERS is, We do NOT know how to program efficiently with multicores, which in current phase, stands for most of the parallel platforms. We do have parallel applications, BUT, they are painful to design and implement, and they are running on mainframes or clusters. They are written once, by specialists. If we want a world of parallel machines and programs, MAKE THE PROGRAMMING MODEL SUITABLE. Make programming easy, with fine grained parallel, not just dispatching processes to cores, but making one process running on many cores WITH EASY PROGRAMMING.