Question how can i get the RWByteAddressBuffer back to cpu correctly？

i'm trying to get this buffer back to cpu for debug rendering, but after i use GetData to get back it to cpu in a list, this looks different to what i'm expect, for example if i render every pixel in camera, the every value in RWByteAddressBuffer should be filled, but after i get this to cpu, i found that only part of value at front of list has been filled. ( by read value in compute shader, i'm sure that in gpu side is read and write correctly)

Code (CSharp):

1.  
2. private GraphicsBuffer startOffsetBuffer;
3. private int startOffsetBufferId;
4.  
5. //....
6. int bufferSizeHead = Dim.x * Dim.y;
7. int bufferStrideHead = sizeof(uint);
8. startOffsetBuffer = new GraphicsBuffer(GraphicsBuffer.Target.Raw, bufferSizeHead, bufferStrideHead);
9. startOffsetBufferId = Shader.PropertyToID("StartOffsetBuffer");
10.  
11. cmd = new CommandBuffer();
12. cmd.SetRenderTarget(targetTexture);
13. cmd.ClearRenderTarget(true, true,Color.black);
14. cmd.SetRandomWriteTarget(1, fragmentLinkBuffer);
15. cmd.SetRandomWriteTarget(2, startOffsetBuffer);
16. cmd.DrawRenderer(targetMeshRenderer, linkedListMaterial);
17. targetCamera.AddCommandBuffer(CameraEvent.AfterEverything, cmd);
18. //....
19. linkedListMaterial.SetBuffer(startOffsetBufferId, startOffsetBuffer);
20. //....
21.  
22. // try to get back to cpu
23. uint[] startOffsetTable = new uint[Dim.x * Dim.y];
24. startOffsetBuffer.GetData(startOffsetTable);
25.  

RWByteAddressBuffer StartOffsetBuffer : register(u2);

//...



uint uStartOffsetAddress = idx.y*(_Dim.x +1)+ idx.x;

uint uOldStartOffset;

StartOffsetBuffer.InterlockedExchange(uStartOffsetAddress, uPixelCount, uOldStartOffset);

 

emmm, i found that i can get close value by divide idx by 4, the problem seems like the date stride not right so it jumped 3 values...
but i still don't known how to fix it...