Fix D3D12 Compute Derivative Tests

Use groupId for the output buffer index to match the python
Only output results for workgroup 1,0,0

Do not reset test failing to false in check_compute_derivative_tests()

Author: Zorro666

util/test/demos/d3d12/d3d12_shader_debug_zoo.cpp

@@ -1758,7 +1758,7 @@ void SetOutput(float4 val)
 
 void Init(float4 val)
 {
-  flatId = tid.z * GROUP_SIZE_X * GROUP_SIZE_Y + tid.y * GROUP_SIZE_X + tid.x;
+  flatId = gid.z * GROUP_SIZE_X * GROUP_SIZE_Y + gid.y * GROUP_SIZE_X + gid.x;
   SetOutput(val);
 }
 
@@ -1771,6 +1771,10 @@ groupshared uint3 gsmUint3[1024];
 [numthreads(GROUP_SIZE_X, GROUP_SIZE_Y, 1)]
 void main(uint3 inDTID : SV_DispatchThreadID, uint3 inGID : SV_GroupThreadID, uint3 inGroup : SV_GroupID)
 {
+  // Only want the workgroup (1,0,0) to output results
+  if ((inGroup.x != 1) || (inGroup.y != 0) || (inGroup.z != 0))
+    return;
+
   float4 testResult = float4(0,0,0,0);
   tid = inDTID;
   gid = inGID;

util/test/tests/D3D12/D3D12_Shader_Debug_Zoo.py

@@ -27,7 +27,6 @@ def check_compute_derivative_tests(self):
                 a for a in comp_dim.children if a.flags & rd.ActionFlags.Dispatch]
 
             for test, action in enumerate(compute_tests):
-                failed = False
                 self.controller.SetFrameEvent(action.eventId, False)
                 pipe = self.controller.GetPipelineState()
                 csrefl = pipe.GetShaderReflection(rd.ShaderStage.Compute)