Round 2- parallel divide & conquer reduction function

gistfile1.m

- (id)reduceWithBlock:(id (^)(id memo, id obj))block andAccumulator:(id)accumulator withBaseLength:(NSUInteger)baseLength
{
    __block id acc = [[accumulator copy] autorelease];
    NSUInteger base_job_length = baseLength;
    if (!baseLength) {
        //  Calculate the base job length
        NSUInteger num_processors = [[NSProcessInfo processInfo] processorCount];
        base_job_length = (NSUInteger)([self count] / sqrt(num_processors)); // (L = n / p squared)
    }
    
    if ([self count] <= base_job_length) {
        //  If the size of the array is <= base case, reduce it serially
        for (id obj in self) { acc = block(acc, obj); }
    } else {
        //  If the array length is > base case, divide & conquer
        NSArray* sub_jobs = [self splitIntoSubArraysOfLength:base_job_length];
        
        dispatch_queue_t result_queue = dispatch_queue_create(NULL, DISPATCH_QUEUE_CONCURRENT);
        
        dispatch_apply([sub_jobs count], result_queue, ^(size_t i) {
            acc = block(acc, [[sub_jobs objectAtIndex:i] reduceWithBlock:block andAccumulator:accumulator withBaseLength:base_job_length]);
        });
        dispatch_release(result_queue);
    }
    
    return acc;
    
}