tested on P100

driver/src/driver.cpp

@@ -597,7 +597,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 8x8 image
-    // cuDNN 68%, ck:nvidia: 72.6%, ck:amd 34%
+    // cudnn@V100 68%, ck@V100 72%, ck@P100 52%, ck@MI50 34%
     constexpr index_t N  = 64;
     constexpr index_t C  = 1536;
     constexpr index_t HI = 8;
@@ -613,7 +613,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 8x8 image
-    // cuDNN 77%, ck:nvidia 76.4%, ck:amd 47%
+    // cudnn@V100 77%, ck@V100 76%, ck@P100 79%, ck@MI50 47%
     constexpr index_t N  = 128;
     constexpr index_t C  = 2048;
     constexpr index_t HI = 8;
@@ -629,7 +629,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 7x7 image
-    // cuDNN 82%, ck:nvidia 76.6%, ck:amd 54%
+    // cudnn@V100 82%, ck@V100 76%, ck@P100 67%, ck@MI50 54%
     constexpr index_t N  = 128;
     constexpr index_t C  = 832;
     constexpr index_t HI = 7;
@@ -645,7 +645,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 8x8 image
-    // cuDNN 83%, ck:nvidia 75.4%, ck:amd 58%
+    // cudnn@V100 83%, ck@V100 75%, ck@P100 78%, ck@MI50 58%
     constexpr index_t N  = 128;
     constexpr index_t C  = 1280;
     constexpr index_t HI = 8;
@@ -659,9 +659,9 @@ int main(int argc, char* argv[])
 
     constexpr index_t HPad = 0;
     constexpr index_t WPad = 0;
-#elif 0
+#elif 1
     // 1x1 filter, 14x14 image
-    // cuDNN 62%, ck:nvidia 68.4%, ck:amd 44%
+    // cudnn@V100 62%, ck@V100 68%, ck@P100 70%, ck@MI50 44%
     constexpr index_t N  = 128;
     constexpr index_t C  = 512;
     constexpr index_t HI = 14;
@@ -677,7 +677,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 8x8 image
-    // cuDNN 74%, ck:nvidia 57.1%, ck:amd 52%
+    // cudnn@V100 74%, ck@V100 57%, ck@P100 78%, ck@MI50 52%
     constexpr index_t N  = 64;
     constexpr index_t C  = 1536;
     constexpr index_t HI = 8;
@@ -693,7 +693,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 28x28 image
-    // cuDNN 86%, ck:nvidia 84.6%, ck:amd 64%
+    // cudnn@V100 86%, ck@V100 84%, ck@P100 80%, ck@MI50 64%
     constexpr index_t N  = 128;
     constexpr index_t C  = 256;
     constexpr index_t HI = 28;
@@ -709,7 +709,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 7x7 image
-    // cuDNN 71%, ck:55.9%, ck:amd 54%
+    // cudnn@V100 71%, ck@V100 55%, ck@P100 70%, ck@MI50 54%
     constexpr index_t N  = 128;
     constexpr index_t C  = 832;
     constexpr index_t HI = 7;
@@ -725,7 +725,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 3x3 filter, 2x2 stride, 35x35 input, 17x17 output
-    // cuDNN 90%, ck:nvidia 93%, ck:amd 73%
+    // cudnn@V100 90%, ck@V100 93%, ck@P100 83%, ck@MI50 73%
     constexpr index_t N  = 128;
     constexpr index_t C  = 288;
     constexpr index_t HI = 35;
@@ -741,7 +741,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 17x17 input
-    // cuDNN 81%, ck:nvidia 76.8%, ck:amd 66%
+    // cudnn@V100 81%, ck@V100 76%, ck@P100 70%, ck@MI50 66%
     constexpr index_t N  = 128;
     constexpr index_t C  = 768;
     constexpr index_t HI = 17;
@@ -755,9 +755,9 @@ int main(int argc, char* argv[])
 
     constexpr index_t HPad = 0;
     constexpr index_t WPad = 0;
-#elif 1
+#elif 0
     // 1x1 filter, 14x14 image
-    // cuDNN 73%, ck:nvidia 72.7%, ck:amd 65%
+    // cudnn@V100 73%, ck@V100 71%, ck@P100 70%, ck@MI50 65%
     constexpr index_t N  = 128;
     constexpr index_t C  = 528;
     constexpr index_t HI = 14;
@@ -773,7 +773,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 14x14 image
-    // cuDNN 73%, ck:nvidia 72.7%, ck:amd 65%
+    // cudnn@V100 73%, ck@V100 72%, ck@P100 79%, ck@MI50 65%
     constexpr index_t N  = 128;
     constexpr index_t C  = 528;
     constexpr index_t HI = 14;
@@ -789,7 +789,7 @@ int main(int argc, char* argv[])
     constexpr index_t WPad = 0;
 #elif 0
     // 1x1 filter, 7x7 image
-    // cuDNN 49%, ck:nvidia 52.8%, ck:amd 45%
+    // cudnn@V100 49%, ck@V100 50%, ck@P100 61%, ck@MI50 45%
     constexpr index_t N  = 128;
     constexpr index_t C  = 832;
     constexpr index_t HI = 7;

script/cmake-cuda_docker.sh

+#!/bin/bash
+
+rm -f CMakeCache.txt
+rm -f *.cmake
+rm -rf CMakeFiles
+
+MY_PROJECT_SOURCE=../../../
+MY_PROJECT_INSTALL=../install.dir
+
+export CUDA_ROOT=/usr/local/cuda
+export CPATH=$CPATH:$CUDA_ROOT/include
+export LIBRARY_PATH=$LIBRARY_PATH:$CUDA_ROOT/lib64
+export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CUDA_ROOT/lib64
+
+cmake                                                                                       \
+-D CMAKE_INSTALL_PREFIX=${MY_PROJECT_INSTALL}                                               \
+-D CMAKE_CXX_COMPILER=clang++-6.0                                                           \
+-D CMAKE_BUILD_TYPE=Release                                                                 \
+-D CMAKE_VERBOSE_MAKEFILE:BOOL=ON                                                           \
+-D DEVICE_BACKEND=NVIDIA                                                                    \
+-D CUDA_COMMON_INCLUDE_DIR="/root/workspace/NVIDIA_CUDA-10.1_Samples/common/inc"            \
+-D CMAKE_CUDA_FLAGS="-ccbin clang++-6.0 -m64 -Xcompiler -fopenmp -lineinfo --source-in-ptx -keep  -Xptxas -v -gencode=arch=compute_60,code=sm_60 -Xptxas -v -gencode=arch=compute_70,code=sm_70" \
+${MY_PROJECT_SOURCE}
+
+
+#-D CMAKE_CUDA_COMPILER="/package/install/cuda_10.0/bin/nvcc"                                \
+#-D CMAKE_CUDA_FLAGS="-ccbin clang++ -m64 -Xcompiler -fopenmp -lineinfo --source-in-ptx -keep -Xptxas -v -gencode=arch=compute_61,code=sm_61" \
+#-D CMAKE_CUDA_FLAGS="-ccbin clang++ -m64 -Xcompiler -fopenmp -lineinfo --source-in-ptx -keep -Xptxas -v -gencode=arch=compute_61,code=sm_61 -Xptxas -v -Xptxas -v -maxrregcount=128" \
+#-D CMAKE_CUDA_FLAGS="-ccbin clang++ -m64 -Xcompiler -fopenmp -lineinfo --source-in-ptx -keep -Xptxas -v -gencode=arch=compute_61,code=sm_61 -Xptxas -v -gencode=arch=compute_70,code=sm_70" \
+#-D CMAKE_CUDA_FLAGS="-ccbin clang++ -m64 -Xcompiler -fopenmp -lineinfo --source-in-ptx -keep -Xptxas -v -gencode=arch=compute_61,code=sm_61 -Xptxas -v -gencode=arch=compute_70,code=sm_70 -Xptxas -v -maxrregcount=128" \

script/extract_asm-cuda.sh

-cuobjdump -xelf all ./driver/driver && nvdisasm --print-code -g driver.sm_61.cubin > driver.sm_61.asm && nvdisasm --print-code -g driver.sm_70.cubin > driver.sm_70.asm
+cuobjdump -xelf sm_60 ./driver/driver && nvdisasm --print-code -g driver.sm_60.cubin > driver.sm_60.asm
+cuobjdump -xelf sm_61 ./driver/driver && nvdisasm --print-code -g driver.sm_61.cubin > driver.sm_61.asm
+cuobjdump -xelf sm_70 ./driver/driver && nvdisasm --print-code -g driver.sm_70.cubin > driver.sm_70.asm