Shown below is a very simple parallel solution to the problem, for Mathematica 7. Upon running it for 30 minutes, the approximation to pi is still not that good (only a few decimal places) using a single multicore processor. In fact, it is roughly identical to this MPI example.

Dynamic[Clock[1800]];
generations = 0;
hits = 0;

TimeConstrained
[

While
[True,

tableValues = ParallelTable[(RandomReal[])^2 + (RandomReal[])^2, {1024}];

Do
[
If[tableValues[[i]] <= 1, hits++];
generations++;,
{i, 1024}
]

]

, 1800]

Print["Approximation of Pi: " , N[(4*hits)/generations, 100]];

However, upon running the above MPI Implementation on Abe at NCSA using 32 cores for close to the given walltime, the approximation improved by many decimal places, giving 3.1415926536. Use this as a benchmark against which to base your results.

An example MPI C program is also included below.