This question is from Intro to parallel programming course.

We explained the Fox Multiplication algorithm by using 4x4 matrices on 16 processors. So in this case each processor was allocated one element of A and the corresponding element of B. Compute the time for 16 processors to multiply two 4x4 matrices using the following assumptions: 1. Time to broadcast one data item Time to send-receive an item Time to perform one multiply-add operation (Note: multiply-add operation is: cc Operations performed in parallel are completed simultaneously. For example, if k multiply-add operations are performed in parallel on k machines, we will count that m units instead of k*m units. a. b. c. b units s units m units a*b) d. Also, compute the time to multiply two 4x4 matrices on a single processor. Usually the broadcast or send-receive time is much larger than the computation time. So, b and s>> m. What is the bound on b and s so that time with 16 processors to multiply two 4x4 matrices is smaller than time to compute the product on a single processor. For simplicity, in this part of the problem, you may assume b-s We explained the Fox Multiplication algorithm by using 4x4 matrices on 16 processors. So in this case each processor was allocated one element of A and the corresponding element of B. Compute the time for 16 processors to multiply two 4x4 matrices using the following assumptions: 1. Time to broadcast one data item Time to send-receive an item Time to perform one multiply-add operation (Note: multiply-add operation is: cc Operations performed in parallel are completed simultaneously. For example, if k multiply-add operations are performed in parallel on k machines, we will count that m units instead of k*m units. a. b. c. b units s units m units a*b) d. Also, compute the time to multiply two 4x4 matrices on a single processor. Usually the broadcast or send-receive time is much larger than the computation time. So, b and s>> m. What is the bound on b and s so that time with 16 processors to multiply two 4x4 matrices is smaller than time to compute the product on a single processor. For simplicity, in this part of the problem, you may assume b-s