Code: Select all
1 public void Update()
2 {
3 foreach (var bucket in occupiedBucket.Select(x => bucketList[x]))
4 {
5 int bucketListCount = bucket.Count();
6 for (int j = 0; j < bucketListCount - 1; j++)
7 for (int k = j + 1; k < bucketListCount; k++)
8 bucket[j].handleCollide(bucket[k]);
9 }
10 } the function it calls inside of itself (handlecollide) isnt spending much time itself, 30% is all just in the for+foreach statements. There has to be a better way to code this. Bucketlist count is usually around 3 or 4 and count is around 200-400.
this is how I am declaring the variables
Code: Select all
1 public static int min = 0;
2 public static int max = Game1.map.levelHeight;
3 public static int cellSize = 64;
4 public static int width = (max-min)/cellSize;
5 public static int numberOfBuckets = width*width;
6 public List<GameEntity>[] bucketList = new List<GameEntity>[numberOfBuckets];
7 public List<int> occupiedBucket = new List<int>();
8 Vector2 normal;
9
10
11 public Grid()
12 {
13 for (int i = 0; i < numberOfBuckets; i++)
14 {
15 bucketList[i] = new List<GameEntity>();
16 }
17 } thanks guys,
lmao, I literally spent 2 weeks optimizing a quad tree for my program needs to its peak and in the course became fairly familiar with them, haha. The maximum efficiency from a quadTree is O(NlogN) for N objects while a hashing technique is O(N) with a cell query of O(1) and an object query of O(N), as you can see this can be extremely more efficient. Especially with hundreds of dynamic objects restructuring the tree each frame, the quadtree is not ideal in this situation.
Without anymore code or a more worthwhile profile it'd be difficult to say. However, more times than not if you're spending ridiculous amounts of time in Present it means that your CPU is at least several frames ahead of your GPU and you're stalling it. This isn't to say that your draw calls are expensive, this may or may not be the case. If the complexity of all the tasks within your main loop aren't too terribly deep, then it can simply mean that you're sitting in a "busy-wait" loop. Eg: You're not handling the case where attempting to present when the GPU is excessively busy. 
