Add auto-diff support for `IRSwizzleStore` (#3102)

* Add auto-diff support for `IRSwizzleStore`

 - Lower IRSwizzleStore to multiple stores during AD preprocess.
 - Fix typo in `transcribeNonDiffInst`

* Remove unnecessary file & add more robust check for 'local' addresses

* Fix.

* Update slang-ir-autodiff-fwd.cpp

---------

Co-authored-by: Yong He <yonghe@outlook.com>

2 files changed, 44 insertions, 0 deletions

diff --git a/tests/autodiff/swizzled-store.slang b/tests/autodiff/swizzled-store.slang
new file mode 100644
index 000000000..58980616b
--- /dev/null
+++ b/tests/autodiff/swizzled-store.slang
@@ -0,0 +1,39 @@
+//TEST(compute):COMPARE_COMPUTE_EX:-slang -compute -shaderobj -output-using-type
+//TEST(compute, vulkan):COMPARE_COMPUTE_EX:-vk -compute -shaderobj -output-using-type
+
+//TEST_INPUT:ubuffer(data=[0 0 0 0], stride=4):out,name=outputBuffer
+RWStructuredBuffer<float> outputBuffer;
+
+typedef DifferentialPair<float2> dpfloat2;
+typedef DifferentialPair<float3> dpfloat3;
+typedef DifferentialPair<float4> dpfloat4;
+
+[Differentiable]
+float2 f(float3 x)
+{
+    float3 u;
+    u.zy = x.yx;
+    return u.zy;
+}
+
+[numthreads(1, 1, 1)]
+void computeMain(uint3 dispatchThreadID: SV_DispatchThreadID)
+{
+    {
+        float3 a = float3(2.0, 2.0, 2.0);
+        float3 da = float3(1.0, 0.5, 1.0);
+
+        outputBuffer[0] = fwd_diff(f)(dpfloat3(a, da)).d.x;
+    }
+
+    {
+        float3 a = float3(2.0, 2.0, 2.0);
+        var dpa = diffPair(a);
+
+        bwd_diff(f)(dpa, float2(0.5, 1.0));
+
+        outputBuffer[1] = dpa.d.x; // 1.0
+        outputBuffer[2] = dpa.d.y; // 0.5
+        outputBuffer[3] = dpa.d.z; // 0.0
+    }
+}
diff --git a/tests/autodiff/swizzled-store.slang.expected.txt b/tests/autodiff/swizzled-store.slang.expected.txt
new file mode 100644
index 000000000..8fe60c0db
--- /dev/null
+++ b/tests/autodiff/swizzled-store.slang.expected.txt
@@ -0,0 +1,5 @@
+type: float
+0.500000
+1.000000
+0.500000
+0.000000