Optimize d3d12 mutable shader object implementation. (#2138)

* Optimize d3d12 mutable shader object implementation.

* Disable mismatched clear value warning message from d3d sdk.

* Fix.

* Fix.

* gfx: Avoid redundant d3d12 QueryInterface call.

Co-authored-by: Yong He <yhe@nvidia.com>

1 files changed, 18 insertions, 1 deletions

diff --git a/tools/gfx/cuda/render-cuda.cpp b/tools/gfx/cuda/render-cuda.cpp
index b98582ded..db5661f20 100644
--- a/tools/gfx/cuda/render-cuda.cpp
+++ b/tools/gfx/cuda/render-cuda.cpp
@@ -953,8 +953,14 @@ public:
 
     public:
         CUDADevice* m_device;
+        TransientResourceHeapBase* m_transientHeap;
+
+        void init(CUDADevice* device, TransientResourceHeapBase* transientHeap)
+        {
+            m_device = device;
+            m_transientHeap = transientHeap;
+        }
 
-        void init(CUDADevice* device) { m_device = device; }
         virtual SLANG_NO_THROW void SLANG_MCALL encodeRenderCommands(
             IRenderPassLayout* renderPass,
             IFramebuffer* framebuffer,
@@ -1171,6 +1177,17 @@ public:
                 return SLANG_OK;
             }
 
+            virtual SLANG_NO_THROW Result SLANG_MCALL
+                bindPipelineWithRootObject(IPipelineState* state, IShaderObject* rootObject) override
+            {
+                m_writer->setPipelineState(state);
+                PipelineStateBase* pipelineImpl = static_cast<PipelineStateBase*>(state);
+                SLANG_RETURN_ON_FAIL(m_commandBuffer->m_device->createRootShaderObject(
+                    pipelineImpl->m_program, m_rootObject.writeRef()));
+                m_rootObject->copyFrom(rootObject, m_commandBuffer->m_transientHeap);
+                return SLANG_OK;
+            }
+
             virtual SLANG_NO_THROW void SLANG_MCALL dispatchCompute(int x, int y, int z) override
             {
                 m_writer->bindRootShaderObject(m_rootObject);