Add xxhash64 support

src/main/cpp/CMakeLists.txt

@@ -161,6 +161,7 @@ add_library(
   src/map_utils.cu
   src/murmur_hash.cu
   src/row_conversion.cu
+  src/xxhash64.cu
   src/zorder.cu
 )
 

src/main/cpp/src/HashJni.cpp

@@ -36,4 +36,21 @@ JNIEXPORT jlong JNICALL Java_com_nvidia_spark_rapids_jni_Hash_murmurHash32(
   }
   CATCH_STD(env, 0);
 }
+
+JNIEXPORT jlong JNICALL Java_com_nvidia_spark_rapids_jni_Hash_xxhash64(JNIEnv* env,
+                                                                       jclass,
+                                                                       jlong seed,
+                                                                       jlongArray column_handles)
+{
+  JNI_NULL_CHECK(env, column_handles, "array of column handles is null", 0);
+
+  try {
+    cudf::jni::auto_set_device(env);
+    auto column_views =
+      cudf::jni::native_jpointerArray<cudf::column_view>{env, column_handles}.get_dereferenced();
+    return cudf::jni::release_as_jlong(
+      spark_rapids_jni::xxhash64(cudf::table_view{column_views}, seed));
+  }
+  CATCH_STD(env, 0);
+}
 }

src/main/cpp/src/hash.cuh

@@ -25,6 +25,8 @@
 
 namespace spark_rapids_jni {
 
+constexpr int64_t DEFAULT_XXHASH64_SEED = 42;
+
 /**
  * @brief Converts a cudf decimal128 value to a java bigdecimal value.
  *
@@ -91,4 +93,20 @@ std::unique_ptr<cudf::column> murmur_hash3_32(
   rmm::cuda_stream_view stream        = cudf::get_default_stream(),
   rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
 
+/**
+ * @brief Computes the xxhash64 hash value of each row in the input set of columns.
+ *
+ * @param input The table of columns to hash
+ * @param seed Optional seed value to use for the hash function
+ * @param stream CUDA stream used for device memory operations and kernel launches
+ * @param mr Device memory resource used to allocate the returned column's device memory
+ *
+ * @returns A column where each row is the hash of a column from the input.
+ */
+std::unique_ptr<cudf::column> xxhash64(
+  cudf::table_view const& input,
+  int64_t seed                        = DEFAULT_XXHASH64_SEED,
+  rmm::cuda_stream_view stream        = cudf::get_default_stream(),
+  rmm::mr::device_memory_resource* mr = rmm::mr::get_current_device_resource());
+
 }  // namespace spark_rapids_jni