he has a point. as the test showed, the issue is not blocksize per se, it's overall scaling. which involves paralellizing the software (irrespective of moore).
Off point.
Again.
Without a debrief of that experiment how could you possibly know what the issues were?
you may be right, you are here more and have keener sense. anyways, it is quite clear to dev teams where the present issues/bottlenecks are.
