Let's say that the program given in question 1 above is slightly changed. It is very important and is constantly running. The only difference is that the total calculation from the main thread It is not expected to ensure that the worker threads are running. Just print the total. Some legal issues that are outside of your scope and that you cannot change You must run only one instance of this program at a time due to requirements. aforementioned. Another run of the app, but after finishing the previous run can be started. In other words, your application is running parallel to two or more separate terminals. You are not allowed to run. You can always assume there is some matrix data to be processed, There is no problem of matrix. There are enough matrices to keep your schedule busy and pending on disk. The firm only uses a dedicated physical (or virtual) server to run this program. wants to buy. The number of cores required in the server to be purchased and above thread mapping from the user to the kernel to be used in the operating system (user to kernel) You are requested to make a suggestion on the thread mapping strategy. Your solution is possible they want it to be optimized as much as possible, all CPU cores purchased must be kept busy. Convince that each CPU will be busy doing useful work throughout the day If they do, they are willing to increase the number of CPU cores. Reading matrices from disk stage, they already have an SSD-type ultra-high-speed device that they will connect to the server and use. Since it is a disk, it will not take much time compared to the calculation phase. Matrices are too big and doing the calculations will take much more time than reading them from the SSD unit. (The calculation for this question is not simple addition, but much more extensive statistical operations.) You might think it is).

a. (6.25 points) How many cores CPU on a server that will be used to run this application would you recommend it? Why? b. (6.25 points) What kind of user to kernel in the operating system to be installed on this server? would you recommend the thread mapping strategy? Why?