kernel calls

torch_cluster/kernel/generic/serial.cu

+#ifndef THC_GENERIC_FILE
+#define THC_GENERIC_FILE "generic/serial.cu"
+#else
+
+void cluster_(serial)(THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCTensor *weight, THCudaLongTensor *degree) {
+}
+
+#endif
+
+

torch_cluster/kernel/serial.cu

+#include <THC.h>
+
+#include "serial.h"
+
+#define cluster_(NAME) TH_CONCAT_4(cluster_, NAME, _kernel_, Real)
+#define thc_(NAME) TH_CONCAT_4(thc_, NAME, _, Real)
+
+void cluster_serial_kernel(THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCudaLongTensor *degree) {
+}
+
+#include "generic/serial.cu"
+#include "THCGenerateFloatType.h"
+#include "generic/serial.cu"
+#include "THCGenerateDoubleType.h"
+#include "generic/serial.cu"
+#include "THCGenerateByteType.h"
+#include "generic/serial.cu"
+#include "THCGenerateCharType.h"
+#include "generic/serial.cu"
+#include "THCGenerateShortType.h"
+#include "generic/serial.cu"
+#include "THCGenerateIntType.h"
+#include "generic/serial.cu"
+#include "THCGenerateLongType.h"

torch_cluster/kernel/serial.h

+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void cluster_serial_kernel(THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCudaLongTensor *degree);
+
+void cluster_serial_kernel_Float (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,       THCudaTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Double(THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCudaDoubleTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Byte  (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,   THCudaByteTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Char  (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,   THCudaCharTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Short (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,  THCudaShortTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Int   (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,    THCudaIntTensor *weight, THCudaLongTensor *degree);
+void cluster_serial_kernel_Long  (THCState *state, THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col,   THCudaLongTensor *weight, THCudaLongTensor *degree);
+
+#ifdef __cplusplus
+}
+#endif

torch_cluster/src/generic/serial_cuda.c

@@ -3,6 +3,7 @@
 #else
 
 void cluster_(serial)(THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCTensor *weight, THCudaLongTensor *degree) {
+  cluster_kernel_(serial)(state, output, row, col, weight, degree);
 }
 
 #endif

torch_cluster/src/serial_cuda.c

@@ -8,6 +8,7 @@
 extern THCState *state;
 
 void cluster_serial_cuda(THCudaLongTensor *output, THCudaLongTensor *row, THCudaLongTensor *col, THCudaLongTensor *degree) {
+  cluster_serial_kernel(state, output, row, col, degree);
 }
 
 #include "generic/serial_cuda.c"