Pros and cons
The DataScalar model is primarily a memory system optimization intended for codes that are performance limited by the memory system and difficult to parallelize.
Simulation results : the DataScalar model of execution works best with codes for which traditional parallelization techniques fail.
Six unmodied SPEC95 binaries ran from 7 % slower to 50 % faster on two nodes, and from 9 % to 100 % faster on four nodes, than on a system with comparable, more traditional memory system.
Current technological parameters do not make DataScalar systems a cost-effective alternative to today's microprocessors.