Parallel Programing HW4 (Pthread)

compile

g++ -g -Wall -o hw4 hw4.cpp -lpthread -lrt

OUTPUT

no difference betweeen non_pthread_output and pthread_output

What have you done

1.seperate rows of bmp_img to each threads :
    assign remainder to the last thread(due to hardware limitaions,# of threads won't be too large)

2. pthread_barrier_wait(&barrier);
    synchronize each threads operation

3. use strtol(argv[1], NULL, 10); 
    to get # of threads at runtime ,  convert string to int (in base 10)

4. sequentially create and join threads by for loop

5. get_time() to calculate runtime
    reference: (stack overflow) https://stackoverflow.com/questions/15976790/how-can-i-calculate-the-running-time-of-a-pthread-matrix-multiplication-program

Analysis on your result

note : I've tested three times each cases, and take the smallest run time as example

case1 : 1000 smoothing time

discuss on the results

There is hardware limitation ; take (16_threads 2000 loops) as an extreme example.The runtime is even slightly longer than the 8 threads one.
runtime of thread6 is longer than threads4 in both cases ->the reason is that bmpinfo.biheight can't be devided by 6