Part 1: Modify the parallel Sieve of Eratosthenes program in class so that the program does NOT set aside memory for even integers.
Part 2: Modify the parallel Sieve of Eratosthenes program in Part 1 so that each process of the program finds its own sieving primes via local computations instead of broadcasts.
Part 3: Modify the parallel Sieve of Eratosthenes program in Part2 so that the program can have effective uses of caches.
Use your program to find all prime numbers within 1010. Output the total number of prime numbers within 1010 and the program execution time (i.e., maximum time of all processes used in the MPI program). Benchmark your program on TARDIS with 32 (1 node), 64(2 nodes), 128(4 nodes), and 256 (8 node) cores to see whether your execution time is reduced by half or not when double the number of computing cores.
Compare the execution time of each version of your program to see how different designs affect the execution time of your program. Note that, in syllabus, we emphasize for ALL homework assignments: “Please make sure that your programs are properly documented and indented. Provide instructions on how to run your programs, give example runs, and analyze your results.”
Additional Instructions: Like previous projects, you’ll have a framework to complete and test the codes. Please check the framework manual for the usage. For submission,please just submit your PDF report, and sieve1/2/3.c. Do not zip them together.
本网站支持淘宝 支付宝 微信支付 paypal等等交易。如果不放心可以用淘宝交易！
E-mail: email@example.com 微信:itcsdx