xingyousong / parallelized-particle-simulation-cs267-hw2 Goto Github PK

In order to use cuda you must first load the cuda module by running:
'module load cuda'

See https://portal.xsede.org/psc-bridges "Using Bridges GPU Nodes" for additional details.

'autocorrect' is the program that checks whether your particles are interacting properly. If it prints an error, then your particles are not interacting properly.

If you are using the autograder to check your performance, you will need to submit 'auto-bridges-gpu' and then run 'python gen_gpusum.py' to general summary statistics and grade your code.

Note that 'auto-bridges-gpu' tests your GPU code against 'serial.cu'. Without any modifications, it tests the O(n^2) GPU code against the O(n^2) serial code that are already provided. 

You will need to modify either 'auto-bridges-gpu' or replace 'gpu.cu' and 'serial.cu' with their O(n) versions. You should already have a working O(n) serial code from Part 1 of this assignment.

We will be testing your code against an optimized, O(n), 'serial.cu' code. If you would like this optimized code to test against your GPU code, then please ask your instructor.