Add cuda reshape, xcorr and unittest.

src/Function1D.cu

@@ -1382,3 +1382,54 @@ CudaMatrix Aurora::intersect(const CudaMatrix& aMatrix1, const CudaMatrix& aMatr
     aIa = CudaMatrix::fromRawData(iaResult,size);
     return result;
 }
+
+CudaMatrix Aurora::reshape(const CudaMatrix& aMatrix, int aRows, int aColumns, int aSlices)
+{
+    if(aMatrix.isNull() || (aMatrix.getDataSize() != aRows * aColumns * aSlices))
+    {
+        std::cerr<<"reshape diffirent size with cudamatrix"<<std::endl;
+        return CudaMatrix();
+    }
+    return CudaMatrix::copyFromRawData(aMatrix.getData(),aRows,aColumns,aSlices);
+}
+
+__global__ void xcorrKernel(float* aInputData1, float* aInputData2,unsigned int aInputSize, float* aOutput, unsigned int aOutputSize)
+{
+    unsigned int idx = blockIdx.x * blockDim.x + threadIdx.x;
+    if (idx < aOutputSize/2 + 1)
+    {
+        for(unsigned int i=0; i<=idx; ++i)
+        {
+            aOutput[idx] += aInputData1[i] * aInputData2[aInputSize - idx - 1 + i];
+        }
+        return;
+    }
+
+    if (idx < aOutputSize)
+    {
+        for(int i=0; i<idx-aOutputSize/2; ++i)
+        {
+            aOutput[aOutputSize - idx + aOutputSize/2] += aInputData1[aInputSize + i - idx + aOutputSize/2] * aInputData2[i];
+        }
+        return;
+    }
+}
+
+CudaMatrix Aurora::xcorr(const CudaMatrix& aMatrix1, const CudaMatrix& aMatrix2)
+{
+    if (aMatrix1.isNull() || aMatrix2.isNull() || aMatrix1.getDataSize() != aMatrix2.getDataSize() || aMatrix1.isComplex() || aMatrix2.isComplex())
+    {
+        std::cerr<<"xcorr not surpport with diffirent input size or complex cudamatrix"<<std::endl;
+        return CudaMatrix();
+    }
+
+    size_t size = aMatrix1.getDataSize() * 2 - 1;
+    float* data = nullptr;
+    cudaMalloc((void**)&data, sizeof(float) * size);
+    cudaMemset(data, 0.0, size);
+    int blocksPerGrid = (size + THREADS_PER_BLOCK - 1) / THREADS_PER_BLOCK;
+    xcorrKernel<<<blocksPerGrid, THREADS_PER_BLOCK>>>(aMatrix1.getData(), aMatrix2.getData(), aMatrix1.getDataSize(), data, size);
+    cudaDeviceSynchronize();
+
+    return CudaMatrix::fromRawData(data, size);
+}

src/Function1D.cuh

@@ -77,7 +77,13 @@ namespace Aurora
 
     CudaMatrix intersect(const CudaMatrix& aMatrix1, const CudaMatrix& aMatrix2);
 
-    CudaMatrix intersect(const CudaMatrix& aMatrix1, const CudaMatrix& aMatrix2, CudaMatrix& aIa);    /**
+    CudaMatrix intersect(const CudaMatrix& aMatrix1, const CudaMatrix& aMatrix2, CudaMatrix& aIa);
+
+    CudaMatrix reshape(const CudaMatrix& aMatrix, int aRows, int aColumns, int aSlices);
+
+    CudaMatrix xcorr(const CudaMatrix& aMatrix1, const CudaMatrix& aMatrix2);
+
+     /**
      * 将所有nan值设置为特定值
      * @attention 直接在原数据上进行修改！
      * @param aMatrix 向量