Merge pull request #390 from denghuilu/master

fix bug of cuda-11 compilation

Author: amcadmus

source/op/cuda/CMakeLists.txt

@@ -19,7 +19,17 @@ include_directories(cub)
 
 message(STATUS "CUDA major version is " ${CUDA_VERSION_MAJOR})
 
-if (${CUDA_VERSION_MAJOR} GREATER "10")
+if (${CUDA_VERSION_MAJOR} GREATER "11")
+    # nvcc flags
+    set(CUDA_NVCC_FLAGS -gencode arch=compute_60,code=sm_60; # Pascal – GP100/Tesla P100 – DGX-1 (Generic Pascal)
+                        -gencode arch=compute_61,code=sm_61; # Pascal - GTX 1080, GTX 1070, GTX 1060, GTX 1050, GTX 1030, Titan Xp, Tesla P40, Tesla P4, Discrete GPU on the NVIDIA Drive PX2
+                        -gencode arch=compute_70,code=sm_70; # Volta  - GV100/Tesla V100, GTX 1180 (GV104)
+                        -gencode arch=compute_75,code=sm_75; # Turing - RTX 2080, Titan RTX, Quadro R8000
+                        -gencode arch=compute_80,code=sm_80; # Anpere - A100
+                        -gencode arch=compute_86,code=sm_86; # Anpere - RTX 3090
+                        -O3; -Xcompiler -fPIC;
+        )
+elseif (${CUDA_VERSION_MAJOR} STREQUAL "11" AND ${CUDA_VERSION_MINOR} GREATER "0")
     # nvcc flags
     set(CUDA_NVCC_FLAGS -gencode arch=compute_50,code=sm_50;
                         -gencode arch=compute_52,code=sm_52; # Tesla M40, Tesla M40, Quadro M6000...
@@ -32,6 +42,18 @@ if (${CUDA_VERSION_MAJOR} GREATER "10")
                         -gencode arch=compute_86,code=sm_86; # Anpere - RTX 3090
                         -O3; -Xcompiler -fPIC;
         )
+elseif (${CUDA_VERSION_MAJOR} STREQUAL "11" AND ${CUDA_VERSION_MINOR} STREQUAL "0")
+    # nvcc flags
+    set(CUDA_NVCC_FLAGS -gencode arch=compute_50,code=sm_50;
+                        -gencode arch=compute_52,code=sm_52; # Tesla M40, Tesla M40, Quadro M6000...
+                        -gencode arch=compute_53,code=sm_53; 
+                        -gencode arch=compute_60,code=sm_60; # Pascal – GP100/Tesla P100 – DGX-1 (Generic Pascal)
+                        -gencode arch=compute_61,code=sm_61; # Pascal - GTX 1080, GTX 1070, GTX 1060, GTX 1050, GTX 1030, Titan Xp, Tesla P40, Tesla P4, Discrete GPU on the NVIDIA Drive PX2
+                        -gencode arch=compute_70,code=sm_70; # Volta  - GV100/Tesla V100, GTX 1180 (GV104)
+                        -gencode arch=compute_75,code=sm_75; # Turing - RTX 2080, Titan RTX, Quadro R8000
+                        -gencode arch=compute_80,code=sm_80; # Anpere - A100
+                        -O3; -Xcompiler -fPIC;
+        )
 elseif (${CUDA_VERSION_MAJOR} STREQUAL "10")
     set(CUDA_NVCC_FLAGS -gencode arch=compute_30,code=sm_30; # Tesla K10, Quadro K600 K420 K410,
                         -gencode arch=compute_35,code=sm_35; # Tesla K20 K40, TITAN Z Black, GTX 780Ti 780