Revolutionary computers on the way. It remains to learn how to program them.

This is the name of the post on gigoma, which tells us that soon everything can change a lot.

In short:
A team of scientists at Stanford has created a new board called the NeuroGrid, which contains 16 cores that can emulate over a million neurons and a billion synapses. They suggest that mass production of such devices is possible at costs of about $ 400 for each unit, which makes it very attractive to use them in all sorts of devices, starting with robots, ending with artificial limbs, in order to significantly increase their computing power, while reducing power consumption.

However, even if we assume that it is already today, there is one important point - to program a NeuroGrid, a neurologist is needed (with appropriate training).

Now the same team of scientists is working on special compilers that would allow people without appropriate scientific baggage to write programs for this system.
')


An interesting point: without creating high-quality compilers for such systems, which will be able to reduce the required amount of knowledge to start working with them, this development can simply "not take off." Because there simply will not be programmers capable of at least something to perplex these systems.
Approximately the same problems are now experiencing the industry of supercomputers. Technically, it is possible to create machines of almost any computing power, but there are very few specialists in the world who are able to write software for them to puzzle them at full capacity, and new ones learn too slowly.