00:01
Hello students to multiply this two square matrices of size n cross n using the p processor and achieving the fastest parallel execution time of the big o log n and as the premium model.
00:17
So for that here is the how with the help of an algorithm i'll explain it over here.
00:24
So first of all we'll divide the input matrix into the p sub matrix so that each of the size of n by n by p into n by p and then the second step is to distribute these sub matrix to the across the p processor in a way that ensures that each processor get the portion of the both input matrix.
00:47
So the third is to perform the multiplication of the sub matrices locally on the each of the processor then later aggregate the local result to compute the final result.
00:59
So as of here you can see here is the pseudo code written with it.
01:04
So let's go for the expression for the performance matrix that is assuming as the p is less than or equal to n to the power of 3 processor.
01:19
So now let's define the each of the notation over here.
01:23
So with n is the size of the matrix size of the square matrix with the size of square matrix of the square matrix of matrix n cross n and p is the number of the processor number of the processors that is with the p is less than or equal to the n to the power of 3.
01:52
Then the next comes is the ip that is the number of the processor used in a p number of number of processor processor used number of processor used that is the p and the sp as the number of steps of the algorithm or the time complexity of the algorithm we can say the time complexity of algorithm of the algorithm.
02:28
Then next comes the ep as the number of the operation performed by the each processor.
02:35
So the number of operations operations performed performed by each processor by each processor...