remove dependency on cuda sample repo

cuda/cuda_stream_manager.cpp

@@ -5,7 +5,6 @@
 #include <iostream>
 
 #include "cuda_stream_manager.h"
-#include <helper_cuda.h> 
 
 #define SMGR_N_STREAMS 16
 

cuda/cuda_stream_manager.h

 #ifndef CUDA_STREAM_MANAGER_H
 #define CUDA_STREAM_MANAGER_H
 
-#include <cuda_runtime.h>
-#include <cublas_v2.h>
+#include "helper_cuda.h"
 
 #ifdef MOE_USE_NCCL
 #include <nccl.h>

cuda/helper_cuda.h

+/* Copyright (c) 2019, NVIDIA CORPORATION. All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *  * Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ *  * Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *  * Neither the name of NVIDIA CORPORATION nor the names of its
+ *    contributors may be used to endorse or promote products derived
+ *    from this software without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY
+ * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
+ * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
+ * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
+ * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
+ * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+ * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
+ * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ */
+
+////////////////////////////////////////////////////////////////////////////////
+// These are CUDA Helper functions for initialization and error checking
+// This file is clipped from the original header file by laekov
+
+#include <cuda_runtime.h>
+#include <cublas_v2.h>
+
+#ifndef HELPER_CUDA_H
+#define HELPER_CUDA_H
+
+static const char *_cudaGetErrorEnum(cudaError_t error) {
+  return cudaGetErrorName(error);
+}
+
+#ifdef CUDA_DRIVER_API
+// CUDA Driver API errors
+static const char *_cudaGetErrorEnum(CUresult error) {
+  static char unknown[] = "<unknown>";
+  const char *ret = NULL;
+  cuGetErrorName(error, &ret);
+  return ret ? ret : unknown;
+}
+#endif
+
+// cuBLAS API errors
+static const char *_cudaGetErrorEnum(cublasStatus_t error) {
+  switch (error) {
+    case CUBLAS_STATUS_SUCCESS:
+      return "CUBLAS_STATUS_SUCCESS";
+
+    case CUBLAS_STATUS_NOT_INITIALIZED:
+      return "CUBLAS_STATUS_NOT_INITIALIZED";
+
+    case CUBLAS_STATUS_ALLOC_FAILED:
+      return "CUBLAS_STATUS_ALLOC_FAILED";
+
+    case CUBLAS_STATUS_INVALID_VALUE:
+      return "CUBLAS_STATUS_INVALID_VALUE";
+
+    case CUBLAS_STATUS_ARCH_MISMATCH:
+      return "CUBLAS_STATUS_ARCH_MISMATCH";
+
+    case CUBLAS_STATUS_MAPPING_ERROR:
+      return "CUBLAS_STATUS_MAPPING_ERROR";
+
+    case CUBLAS_STATUS_EXECUTION_FAILED:
+      return "CUBLAS_STATUS_EXECUTION_FAILED";
+
+    case CUBLAS_STATUS_INTERNAL_ERROR:
+      return "CUBLAS_STATUS_INTERNAL_ERROR";
+
+    case CUBLAS_STATUS_NOT_SUPPORTED:
+      return "CUBLAS_STATUS_NOT_SUPPORTED";
+
+    case CUBLAS_STATUS_LICENSE_ERROR:
+      return "CUBLAS_STATUS_LICENSE_ERROR";
+  }
+
+  return "<unknown>";
+}
+
+#ifdef _CUFFT_H_
+// cuFFT API errors
+static const char *_cudaGetErrorEnum(cufftResult error) {
+  switch (error) {
+    case CUFFT_SUCCESS:
+      return "CUFFT_SUCCESS";
+
+    case CUFFT_INVALID_PLAN:
+      return "CUFFT_INVALID_PLAN";
+
+    case CUFFT_ALLOC_FAILED:
+      return "CUFFT_ALLOC_FAILED";
+
+    case CUFFT_INVALID_TYPE:
+      return "CUFFT_INVALID_TYPE";
+
+    case CUFFT_INVALID_VALUE:
+      return "CUFFT_INVALID_VALUE";
+
+    case CUFFT_INTERNAL_ERROR:
+      return "CUFFT_INTERNAL_ERROR";
+
+    case CUFFT_EXEC_FAILED:
+      return "CUFFT_EXEC_FAILED";
+
+    case CUFFT_SETUP_FAILED:
+      return "CUFFT_SETUP_FAILED";
+
+    case CUFFT_INVALID_SIZE:
+      return "CUFFT_INVALID_SIZE";
+
+    case CUFFT_UNALIGNED_DATA:
+      return "CUFFT_UNALIGNED_DATA";
+
+    case CUFFT_INCOMPLETE_PARAMETER_LIST:
+      return "CUFFT_INCOMPLETE_PARAMETER_LIST";
+
+    case CUFFT_INVALID_DEVICE:
+      return "CUFFT_INVALID_DEVICE";
+
+    case CUFFT_PARSE_ERROR:
+      return "CUFFT_PARSE_ERROR";
+
+    case CUFFT_NO_WORKSPACE:
+      return "CUFFT_NO_WORKSPACE";
+
+    case CUFFT_NOT_IMPLEMENTED:
+      return "CUFFT_NOT_IMPLEMENTED";
+
+    case CUFFT_LICENSE_ERROR:
+      return "CUFFT_LICENSE_ERROR";
+
+    case CUFFT_NOT_SUPPORTED:
+      return "CUFFT_NOT_SUPPORTED";
+  }
+
+  return "<unknown>";
+}
+#endif
+
+#ifdef CUSPARSEAPI
+// cuSPARSE API errors
+static const char *_cudaGetErrorEnum(cusparseStatus_t error) {
+  switch (error) {
+    case CUSPARSE_STATUS_SUCCESS:
+      return "CUSPARSE_STATUS_SUCCESS";
+
+    case CUSPARSE_STATUS_NOT_INITIALIZED:
+      return "CUSPARSE_STATUS_NOT_INITIALIZED";
+
+    case CUSPARSE_STATUS_ALLOC_FAILED:
+      return "CUSPARSE_STATUS_ALLOC_FAILED";
+
+    case CUSPARSE_STATUS_INVALID_VALUE:
+      return "CUSPARSE_STATUS_INVALID_VALUE";
+
+    case CUSPARSE_STATUS_ARCH_MISMATCH:
+      return "CUSPARSE_STATUS_ARCH_MISMATCH";
+
+    case CUSPARSE_STATUS_MAPPING_ERROR:
+      return "CUSPARSE_STATUS_MAPPING_ERROR";
+
+    case CUSPARSE_STATUS_EXECUTION_FAILED:
+      return "CUSPARSE_STATUS_EXECUTION_FAILED";
+
+    case CUSPARSE_STATUS_INTERNAL_ERROR:
+      return "CUSPARSE_STATUS_INTERNAL_ERROR";
+
+    case CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED:
+      return "CUSPARSE_STATUS_MATRIX_TYPE_NOT_SUPPORTED";
+  }
+
+  return "<unknown>";
+}
+#endif
+
+#ifdef CUSOLVER_COMMON_H_
+// cuSOLVER API errors
+static const char *_cudaGetErrorEnum(cusolverStatus_t error) {
+  switch (error) {
+    case CUSOLVER_STATUS_SUCCESS:
+      return "CUSOLVER_STATUS_SUCCESS";
+    case CUSOLVER_STATUS_NOT_INITIALIZED:
+      return "CUSOLVER_STATUS_NOT_INITIALIZED";
+    case CUSOLVER_STATUS_ALLOC_FAILED:
+      return "CUSOLVER_STATUS_ALLOC_FAILED";
+    case CUSOLVER_STATUS_INVALID_VALUE:
+      return "CUSOLVER_STATUS_INVALID_VALUE";
+    case CUSOLVER_STATUS_ARCH_MISMATCH:
+      return "CUSOLVER_STATUS_ARCH_MISMATCH";
+    case CUSOLVER_STATUS_MAPPING_ERROR:
+      return "CUSOLVER_STATUS_MAPPING_ERROR";
+    case CUSOLVER_STATUS_EXECUTION_FAILED:
+      return "CUSOLVER_STATUS_EXECUTION_FAILED";
+    case CUSOLVER_STATUS_INTERNAL_ERROR:
+      return "CUSOLVER_STATUS_INTERNAL_ERROR";
+    case CUSOLVER_STATUS_MATRIX_TYPE_NOT_SUPPORTED:
+      return "CUSOLVER_STATUS_MATRIX_TYPE_NOT_SUPPORTED";
+    case CUSOLVER_STATUS_NOT_SUPPORTED:
+      return "CUSOLVER_STATUS_NOT_SUPPORTED ";
+    case CUSOLVER_STATUS_ZERO_PIVOT:
+      return "CUSOLVER_STATUS_ZERO_PIVOT";
+    case CUSOLVER_STATUS_INVALID_LICENSE:
+      return "CUSOLVER_STATUS_INVALID_LICENSE";
+  }
+
+  return "<unknown>";
+}
+#endif
+
+#ifdef CURAND_H_
+// cuRAND API errors
+static const char *_cudaGetErrorEnum(curandStatus_t error) {
+  switch (error) {
+    case CURAND_STATUS_SUCCESS:
+      return "CURAND_STATUS_SUCCESS";
+
+    case CURAND_STATUS_VERSION_MISMATCH:
+      return "CURAND_STATUS_VERSION_MISMATCH";
+
+    case CURAND_STATUS_NOT_INITIALIZED:
+      return "CURAND_STATUS_NOT_INITIALIZED";
+
+    case CURAND_STATUS_ALLOCATION_FAILED:
+      return "CURAND_STATUS_ALLOCATION_FAILED";
+
+    case CURAND_STATUS_TYPE_ERROR:
+      return "CURAND_STATUS_TYPE_ERROR";
+
+    case CURAND_STATUS_OUT_OF_RANGE:
+      return "CURAND_STATUS_OUT_OF_RANGE";
+
+    case CURAND_STATUS_LENGTH_NOT_MULTIPLE:
+      return "CURAND_STATUS_LENGTH_NOT_MULTIPLE";
+
+    case CURAND_STATUS_DOUBLE_PRECISION_REQUIRED:
+      return "CURAND_STATUS_DOUBLE_PRECISION_REQUIRED";
+
+    case CURAND_STATUS_LAUNCH_FAILURE:
+      return "CURAND_STATUS_LAUNCH_FAILURE";
+
+    case CURAND_STATUS_PREEXISTING_FAILURE:
+      return "CURAND_STATUS_PREEXISTING_FAILURE";
+
+    case CURAND_STATUS_INITIALIZATION_FAILED:
+      return "CURAND_STATUS_INITIALIZATION_FAILED";
+
+    case CURAND_STATUS_ARCH_MISMATCH:
+      return "CURAND_STATUS_ARCH_MISMATCH";
+
+    case CURAND_STATUS_INTERNAL_ERROR:
+      return "CURAND_STATUS_INTERNAL_ERROR";
+  }
+
+  return "<unknown>";
+}
+#endif
+
+#ifdef NVJPEGAPI
+// nvJPEG API errors
+static const char *_cudaGetErrorEnum(nvjpegStatus_t error) {
+  switch (error) {
+    case NVJPEG_STATUS_SUCCESS:
+      return "NVJPEG_STATUS_SUCCESS";
+
+    case NVJPEG_STATUS_NOT_INITIALIZED:
+      return "NVJPEG_STATUS_NOT_INITIALIZED";
+
+    case NVJPEG_STATUS_INVALID_PARAMETER:
+      return "NVJPEG_STATUS_INVALID_PARAMETER";
+
+    case NVJPEG_STATUS_BAD_JPEG:
+      return "NVJPEG_STATUS_BAD_JPEG";
+
+    case NVJPEG_STATUS_JPEG_NOT_SUPPORTED:
+      return "NVJPEG_STATUS_JPEG_NOT_SUPPORTED";
+
+    case NVJPEG_STATUS_ALLOCATOR_FAILURE:
+      return "NVJPEG_STATUS_ALLOCATOR_FAILURE";
+
+    case NVJPEG_STATUS_EXECUTION_FAILED:
+      return "NVJPEG_STATUS_EXECUTION_FAILED";
+
+    case NVJPEG_STATUS_ARCH_MISMATCH:
+      return "NVJPEG_STATUS_ARCH_MISMATCH";
+
+    case NVJPEG_STATUS_INTERNAL_ERROR:
+      return "NVJPEG_STATUS_INTERNAL_ERROR";
+  }
+
+  return "<unknown>";
+}
+#endif
+
+#ifdef NV_NPPIDEFS_H
+// NPP API errors
+static const char *_cudaGetErrorEnum(NppStatus error) {
+  switch (error) {
+    case NPP_NOT_SUPPORTED_MODE_ERROR:
+      return "NPP_NOT_SUPPORTED_MODE_ERROR";
+
+    case NPP_ROUND_MODE_NOT_SUPPORTED_ERROR:
+      return "NPP_ROUND_MODE_NOT_SUPPORTED_ERROR";
+
+    case NPP_RESIZE_NO_OPERATION_ERROR:
+      return "NPP_RESIZE_NO_OPERATION_ERROR";
+
+    case NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY:
+      return "NPP_NOT_SUFFICIENT_COMPUTE_CAPABILITY";
+
+#if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) <= 0x5000
+
+    case NPP_BAD_ARG_ERROR:
+      return "NPP_BAD_ARGUMENT_ERROR";
+
+    case NPP_COEFF_ERROR:
+      return "NPP_COEFFICIENT_ERROR";
+
+    case NPP_RECT_ERROR:
+      return "NPP_RECTANGLE_ERROR";
+
+    case NPP_QUAD_ERROR:
+      return "NPP_QUADRANGLE_ERROR";
+
+    case NPP_MEM_ALLOC_ERR:
+      return "NPP_MEMORY_ALLOCATION_ERROR";
+
+    case NPP_HISTO_NUMBER_OF_LEVELS_ERROR:
+      return "NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR";
+
+    case NPP_INVALID_INPUT:
+      return "NPP_INVALID_INPUT";
+
+    case NPP_POINTER_ERROR:
+      return "NPP_POINTER_ERROR";
+
+    case NPP_WARNING:
+      return "NPP_WARNING";
+
+    case NPP_ODD_ROI_WARNING:
+      return "NPP_ODD_ROI_WARNING";
+#else
+
+    // These are for CUDA 5.5 or higher
+    case NPP_BAD_ARGUMENT_ERROR:
+      return "NPP_BAD_ARGUMENT_ERROR";
+
+    case NPP_COEFFICIENT_ERROR:
+      return "NPP_COEFFICIENT_ERROR";
+
+    case NPP_RECTANGLE_ERROR:
+      return "NPP_RECTANGLE_ERROR";
+
+    case NPP_QUADRANGLE_ERROR:
+      return "NPP_QUADRANGLE_ERROR";
+
+    case NPP_MEMORY_ALLOCATION_ERR:
+      return "NPP_MEMORY_ALLOCATION_ERROR";
+
+    case NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR:
+      return "NPP_HISTOGRAM_NUMBER_OF_LEVELS_ERROR";
+
+    case NPP_INVALID_HOST_POINTER_ERROR:
+      return "NPP_INVALID_HOST_POINTER_ERROR";
+
+    case NPP_INVALID_DEVICE_POINTER_ERROR:
+      return "NPP_INVALID_DEVICE_POINTER_ERROR";
+#endif
+
+    case NPP_LUT_NUMBER_OF_LEVELS_ERROR:
+      return "NPP_LUT_NUMBER_OF_LEVELS_ERROR";
+
+    case NPP_TEXTURE_BIND_ERROR:
+      return "NPP_TEXTURE_BIND_ERROR";
+
+    case NPP_WRONG_INTERSECTION_ROI_ERROR:
+      return "NPP_WRONG_INTERSECTION_ROI_ERROR";
+
+    case NPP_NOT_EVEN_STEP_ERROR:
+      return "NPP_NOT_EVEN_STEP_ERROR";
+
+    case NPP_INTERPOLATION_ERROR:
+      return "NPP_INTERPOLATION_ERROR";
+
+    case NPP_RESIZE_FACTOR_ERROR:
+      return "NPP_RESIZE_FACTOR_ERROR";
+
+    case NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR:
+      return "NPP_HAAR_CLASSIFIER_PIXEL_MATCH_ERROR";
+
+#if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) <= 0x5000
+
+    case NPP_MEMFREE_ERR:
+      return "NPP_MEMFREE_ERR";
+
+    case NPP_MEMSET_ERR:
+      return "NPP_MEMSET_ERR";
+
+    case NPP_MEMCPY_ERR:
+      return "NPP_MEMCPY_ERROR";
+
+    case NPP_MIRROR_FLIP_ERR:
+      return "NPP_MIRROR_FLIP_ERR";
+#else
+
+    case NPP_MEMFREE_ERROR:
+      return "NPP_MEMFREE_ERROR";
+
+    case NPP_MEMSET_ERROR:
+      return "NPP_MEMSET_ERROR";
+
+    case NPP_MEMCPY_ERROR:
+      return "NPP_MEMCPY_ERROR";
+
+    case NPP_MIRROR_FLIP_ERROR:
+      return "NPP_MIRROR_FLIP_ERROR";
+#endif
+
+    case NPP_ALIGNMENT_ERROR:
+      return "NPP_ALIGNMENT_ERROR";
+
+    case NPP_STEP_ERROR:
+      return "NPP_STEP_ERROR";
+
+    case NPP_SIZE_ERROR:
+      return "NPP_SIZE_ERROR";
+
+    case NPP_NULL_POINTER_ERROR:
+      return "NPP_NULL_POINTER_ERROR";
+
+    case NPP_CUDA_KERNEL_EXECUTION_ERROR:
+      return "NPP_CUDA_KERNEL_EXECUTION_ERROR";
+
+    case NPP_NOT_IMPLEMENTED_ERROR:
+      return "NPP_NOT_IMPLEMENTED_ERROR";
+
+    case NPP_ERROR:
+      return "NPP_ERROR";
+
+    case NPP_SUCCESS:
+      return "NPP_SUCCESS";
+
+    case NPP_WRONG_INTERSECTION_QUAD_WARNING:
+      return "NPP_WRONG_INTERSECTION_QUAD_WARNING";
+
+    case NPP_MISALIGNED_DST_ROI_WARNING:
+      return "NPP_MISALIGNED_DST_ROI_WARNING";
+
+    case NPP_AFFINE_QUAD_INCORRECT_WARNING:
+      return "NPP_AFFINE_QUAD_INCORRECT_WARNING";
+
+    case NPP_DOUBLE_SIZE_WARNING:
+      return "NPP_DOUBLE_SIZE_WARNING";
+
+    case NPP_WRONG_INTERSECTION_ROI_WARNING:
+      return "NPP_WRONG_INTERSECTION_ROI_WARNING";
+
+#if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) >= 0x6000
+    /* These are 6.0 or higher */
+    case NPP_LUT_PALETTE_BITSIZE_ERROR:
+      return "NPP_LUT_PALETTE_BITSIZE_ERROR";
+
+    case NPP_ZC_MODE_NOT_SUPPORTED_ERROR:
+      return "NPP_ZC_MODE_NOT_SUPPORTED_ERROR";
+
+    case NPP_QUALITY_INDEX_ERROR:
+      return "NPP_QUALITY_INDEX_ERROR";
+
+    case NPP_CHANNEL_ORDER_ERROR:
+      return "NPP_CHANNEL_ORDER_ERROR";
+
+    case NPP_ZERO_MASK_VALUE_ERROR:
+      return "NPP_ZERO_MASK_VALUE_ERROR";
+
+    case NPP_NUMBER_OF_CHANNELS_ERROR:
+      return "NPP_NUMBER_OF_CHANNELS_ERROR";
+
+    case NPP_COI_ERROR:
+      return "NPP_COI_ERROR";
+
+    case NPP_DIVISOR_ERROR:
+      return "NPP_DIVISOR_ERROR";
+
+    case NPP_CHANNEL_ERROR:
+      return "NPP_CHANNEL_ERROR";
+
+    case NPP_STRIDE_ERROR:
+      return "NPP_STRIDE_ERROR";
+
+    case NPP_ANCHOR_ERROR:
+      return "NPP_ANCHOR_ERROR";
+
+    case NPP_MASK_SIZE_ERROR:
+      return "NPP_MASK_SIZE_ERROR";
+
+    case NPP_MOMENT_00_ZERO_ERROR:
+      return "NPP_MOMENT_00_ZERO_ERROR";
+
+    case NPP_THRESHOLD_NEGATIVE_LEVEL_ERROR:
+      return "NPP_THRESHOLD_NEGATIVE_LEVEL_ERROR";
+
+    case NPP_THRESHOLD_ERROR:
+      return "NPP_THRESHOLD_ERROR";
+
+    case NPP_CONTEXT_MATCH_ERROR:
+      return "NPP_CONTEXT_MATCH_ERROR";
+
+    case NPP_FFT_FLAG_ERROR:
+      return "NPP_FFT_FLAG_ERROR";
+
+    case NPP_FFT_ORDER_ERROR:
+      return "NPP_FFT_ORDER_ERROR";
+
+    case NPP_SCALE_RANGE_ERROR:
+      return "NPP_SCALE_RANGE_ERROR";
+
+    case NPP_DATA_TYPE_ERROR:
+      return "NPP_DATA_TYPE_ERROR";
+
+    case NPP_OUT_OFF_RANGE_ERROR:
+      return "NPP_OUT_OFF_RANGE_ERROR";
+
+    case NPP_DIVIDE_BY_ZERO_ERROR:
+      return "NPP_DIVIDE_BY_ZERO_ERROR";
+
+    case NPP_RANGE_ERROR:
+      return "NPP_RANGE_ERROR";
+
+    case NPP_NO_MEMORY_ERROR:
+      return "NPP_NO_MEMORY_ERROR";
+
+    case NPP_ERROR_RESERVED:
+      return "NPP_ERROR_RESERVED";
+
+    case NPP_NO_OPERATION_WARNING:
+      return "NPP_NO_OPERATION_WARNING";
+
+    case NPP_DIVIDE_BY_ZERO_WARNING:
+      return "NPP_DIVIDE_BY_ZERO_WARNING";
+#endif
+
+#if ((NPP_VERSION_MAJOR << 12) + (NPP_VERSION_MINOR << 4)) >= 0x7000
+    /* These are 7.0 or higher */
+    case NPP_OVERFLOW_ERROR:
+      return "NPP_OVERFLOW_ERROR";
+
+    case NPP_CORRUPTED_DATA_ERROR:
+      return "NPP_CORRUPTED_DATA_ERROR";
+#endif
+  }
+
+  return "<unknown>";
+}
+#endif
+
+
+template <typename T>
+void check(T result, char const *const func, const char *const file,
+           int const line) {
+  if (result) {
+    fprintf(stderr, "CUDA error at %s:%d code=%d(%s) \"%s\" \n", file, line,
+            static_cast<unsigned int>(result), _cudaGetErrorEnum(result), func);
+    exit(EXIT_FAILURE);
+  }
+}
+
+// This will output the proper CUDA error strings in the event
+// that a CUDA host call returns an error
+#define checkCudaErrors(val) check((val), #val, __FILE__, __LINE__)
+
+#endif  // HELPER_CUDA_H

cuda/moe_comm_kernel.cu

@@ -7,7 +7,6 @@
 #include <cuda.h>
 #include <cuda_runtime.h>
 #include <cublas_v2.h>
-#include <helper_cuda.h> 
 
 #include "cuda_stream_manager.h"
 

cuda/moe_compute_kernel.cu

@@ -7,7 +7,6 @@
 #include <cuda.h>
 #include <cuda_runtime.h>
 #include <cublas_v2.h>
-#include <helper_cuda.h> 
 #include <c10/cuda/CUDAGuard.h>
 
 #include "timer.hh"

cuda/moe_cuda_kernel.h

@@ -4,6 +4,7 @@
 #include <vector>
 #include <torch/extension.h>
 #include <torch/torch.h>
+#include "helper_cuda.h"
 
 std::vector<torch::Tensor> moe_cuda_expert_count(
     torch::Tensor gate, size_t num_expert);

cuda/moe_fused_kernel.cu

@@ -7,7 +7,6 @@
 #include <cuda.h>
 #include <cuda_runtime.h>
 #include <cublas_v2.h>
-#include <helper_cuda.h> 
 #include <c10/cuda/CUDAGuard.h>
 
 #include "cuda_stream_manager.h"

requirements.txt

 torch
+numpy

setup.py

@@ -3,11 +3,9 @@ from torch.utils.cpp_extension import BuildExtension, CUDAExtension
 import os
 
 
-CUDA_HELPER = os.environ.get('CUDA_HELPER', '/usr/local/cuda/samples/common/inc')
-cxx_flags = [
-        '-I{}'.format(CUDA_HELPER)
-        ]
+cxx_flags = []
 ext_libs = []
+
 if os.environ.get('USE_NCCL', '0') == '1':
     cxx_flags.append('-DMOE_USE_NCCL')
     ext_libs.append('nccl')