Consider the two matrix multiplication schemes in Example 3.23 of the book. Assuming a fixed efficiency of E = 0.25, system B is more scalable than system A by the isoefficiency metric, for all values of b, c >0. Assuming b = 4c, we have W(A) = 1.33^3 n^1.5 and W(B) = ^ 1.5. How much more scalable is B over A under the above conditions? See the parallel execution times in Table 3.25. Derive an expression for the ratio Tn(B) /Tn(A), when. Can system B be slower than system A? Explain why or why not.
