Efficient and highly parallel computation

There are many variations of the basic parallel random access machine (PRAM), which includes an arbitrary number of processors sharing an arbitrary amount of random access memory. An exclusive-read, exclusive-write PRAM (EREW PRAM) does not allow any concurrent memory accesses (so all concurrent memory accesses must be explicitly serialized in the program). An EREW PRAM is useful because any algorithm running on such a machine experiences no contention due to concurrent memory accesses. A concurrent-read, concurrentwrite PRAM (CRCW PRAM), on the other hand, has a unit cost for any number of processors concurrently accessing any single memory location. In this model, contention due to concurrent memory accesses is not necessarily included in the running time. Can an EREW PRAM simulate a CRCW PRAM (with the intention of reducing or eliminating time spent queuing memory concurrent accesses in physical hardware implementations)? Due to a refinement of a simulation originally by Vishkin [28], a CRCW PRAM algorithm running in time t(n) on p(n) processors can be simulated by an EREW PRAM algorithm running in time O(t(n) · lgn) on p(n) processors.