Question How to avoid race condition when accessing NativeReference in parallel?

Hi all. I want to figure out how to correctly change NativeReference<int> simultaneously from several threads.

Code (CSharp):

1. using UnityEngine;
2. using Unity.Jobs;
3. using Unity.Collections;
4.  
5. public class ParallelJobsExample : MonoBehaviour
6. {
7.     private void Start()
8.     {
9.         NativeReference<int> result = new NativeReference<int>(Allocator.TempJob);
10.  
11.         JobHandle jobHandle = new MyJob()
12.         {
13.             result = result,
14.         }.Schedule(128, 64);
15.  
16.         jobHandle.Complete();
17.  
18.         Debug.Log(result.Value);
19.  
20.         result.Dispose();
21.     }
22.  
23.     private struct MyJob : IJobParallelFor
24.     {
25.         [NativeDisableParallelForRestriction]
26.         public NativeReference<int> result;
27.  
28.         public void Execute(int index)
29.         {
30.             result.Value++;
31.         }
32.     }
33. }

This option generally works, but causes a race condition in which different threads write to the same place. As a result, the final value of result is less than expected. How can this be avoided? It's a pity without using unsafe code.

 

You'll need unsafe code and Interlocked.Add.

 

I have seen such options, but I would like an option without unsafe code. Are there really none?

 

There is an AtomicCounter in the Unity.Collections

 

I mean this is not a NativeReference problem, it's the multithreading problem 101. AtomicCounter has a cool solution by using per thread accumulators which is faster than Interlocked.Add.