Upgrade to CUDA 9.0 and cuDNN 7.0 (deeplearning4j#4129)

* Upgrade to CUDA 9.0 and cuDNN 7.0 * Change javacpp-presets back to release version for CUDA
shaobai · Oct 4, 2017 · 907d414 · 907d414
1 parent 69251fa
commit 907d414
Show file tree

Hide file tree

Showing 6 changed files with 33 additions and 23 deletions.
diff --git a/change-cuda-versions.sh b/change-cuda-versions.sh
@@ -21,9 +21,11 @@
 
 set -e
 
-VALID_VERSIONS=( 7.5 8.0 )
-CUDA_75_VERSION="7\.5"
+VALID_VERSIONS=( 8.0 9.0 )
 CUDA_80_VERSION="8\.0"
+CUDA_90_VERSION="9\.0"
+CUDNN_60_VERSION="6\.0"
+CUDNN_70_VERSION="7\.0"
 
 usage() {
   echo "Usage: $(basename $0) [-h|--help] <cuda version to be used>
@@ -49,16 +51,20 @@ check_cuda_version() {
 
 check_cuda_version "$TO_VERSION"
 
-if [ $TO_VERSION = "8.0" ]; then
-  FROM_BINARY="-7\.5"
-  TO_BINARY="-8\.0"
-  FROM_VERSION=$CUDA_75_VERSION
-  TO_VERSION=$CUDA_80_VERSION
-else
+if [ $TO_VERSION = "9.0" ]; then
   FROM_BINARY="-8\.0"
-  TO_BINARY="-7\.5"
+  TO_BINARY="-9\.0"
   FROM_VERSION=$CUDA_80_VERSION
-  TO_VERSION=$CUDA_75_VERSION
+  TO_VERSION=$CUDA_90_VERSION
+  FROM_VERSION2=$CUDNN_60_VERSION
+  TO_VERSION2=$CUDNN_70_VERSION
+else
+  FROM_BINARY="-9\.0"
+  TO_BINARY="-8\.0"
+  FROM_VERSION=$CUDA_90_VERSION
+  TO_VERSION=$CUDA_80_VERSION
+  FROM_VERSION2=$CUDNN_70_VERSION
+  TO_VERSION2=$CUDNN_60_VERSION
 fi
 
 sed_i() {
@@ -71,16 +77,20 @@ echo "Updating CUDA versions in pom.xml files to CUDA $1";
 
 BASEDIR=$(dirname $0)
 
-#Artifact ids, ending with "-7.5" or "-8.0". nd4j-cuda, deeplearning4j-cuda, etc.
+#Artifact ids, ending with "-8.0" or "-9.0". nd4j-cuda, deeplearning4j-cuda, etc.
 find "$BASEDIR" -name 'pom.xml' -not -path '*target*' \
   -exec bash -c "sed_i 's/\(artifactId>.*\)'$FROM_BINARY'<\/artifactId>/\1'$TO_BINARY'<\/artifactId>/g' {}" \;
 
-#Artifact ids, ending with "-7.5-platform" or "-8.0-platform". nd4j-cuda-platform, etc.
+#Artifact ids, ending with "-8.0-platform" or "-9.0-platform". nd4j-cuda-platform, etc.
 find "$BASEDIR" -name 'pom.xml' -not -path '*target*' \
   -exec bash -c "sed_i 's/\(artifactId>.*\)'$FROM_BINARY'-platform<\/artifactId>/\1'$TO_BINARY'-platform<\/artifactId>/g' {}" \;
 
-#CUDA versions, like <cuda.version>7.5</cuda.version>
+#CUDA versions, like <cuda.version>9.0</cuda.version>
 find "$BASEDIR" -name 'pom.xml' -not -path '*target*' \
   -exec bash -c "sed_i 's/\(cuda.version>\)'$FROM_VERSION'<\/cuda.version>/\1'$TO_VERSION'<\/cuda.version>/g' {}" \;
 
+#cuDNN versions, like <cudnn.version>7.0</cudnn.version>
+find "$BASEDIR" -name 'pom.xml' -not -path '*target*' \
+  -exec bash -c "sed_i 's/\(cudnn.version>\)'$FROM_VERSION2'<\/cudnn.version>/\1'$TO_VERSION2'<\/cudnn.version>/g' {}" \;
+
 echo "Done updating CUDA versions.";
diff --git a/deeplearning4j-core/pom.xml b/deeplearning4j-core/pom.xml
@@ -57,7 +57,7 @@
             </dependency>
             <dependency>
                 <groupId>org.nd4j</groupId>
-                <artifactId>nd4j-cuda-8.0</artifactId>
+                <artifactId>nd4j-cuda-9.0</artifactId>
                 <version>${nd4j.version}</version>
                 <scope>test</scope>
             </dependency>

diff --git a/deeplearning4j-cuda/pom.xml b/deeplearning4j-cuda/pom.xml
@@ -18,7 +18,7 @@
 
 <project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
     <modelVersion>4.0.0</modelVersion>
-    <artifactId>deeplearning4j-cuda-8.0</artifactId>
+    <artifactId>deeplearning4j-cuda-9.0</artifactId>
     <name>deeplearning4j-cuda</name>
     <parent>
         <groupId>org.deeplearning4j</groupId>
@@ -27,8 +27,8 @@
     </parent>
 
     <properties>
-        <cuda.version>8.0</cuda.version>
-        <cudnn.version>6.0</cudnn.version>
+        <cuda.version>9.0</cuda.version>
+        <cudnn.version>7.0</cudnn.version>
     </properties>
 
     <dependencyManagement>

diff --git a/...j-cuda/src/main/java/org/deeplearning4j/nn/layers/convolution/CudnnConvolutionHelper.java b/...j-cuda/src/main/java/org/deeplearning4j/nn/layers/convolution/CudnnConvolutionHelper.java
@@ -164,7 +164,7 @@ public Pair<Gradient, INDArray> backpropGradient(INDArray input, INDArray weight
                         srcStride[0], srcStride[1], srcStride[2], srcStride[3]));
         checkCudnn(cudnnSetTensor4dDescriptorEx(cudnnContext.deltaTensorDesc, dataType, miniBatch, outDepth, outH, outW,
                         deltaStride[0], deltaStride[1], deltaStride[2], deltaStride[3]));
-        checkCudnn(cudnnSetConvolution2dDescriptor_v5(cudnnContext.convDesc, pad[0], pad[1], strides[0], strides[1], dilation[0],
+        checkCudnn(cudnnSetConvolution2dDescriptor(cudnnContext.convDesc, pad[0], pad[1], strides[0], strides[1], dilation[0],
                         dilation[1], CUDNN_CROSS_CORRELATION, dataType));
         checkCudnn(cudnnSetFilter4dDescriptor(cudnnContext.filterDesc, dataType, tensorFormat, outDepth, inDepth, kH,
                         kW));
@@ -341,7 +341,7 @@ public INDArray preOutput(INDArray input, INDArray weights, INDArray bias, int[]
                         srcStride[0], srcStride[1], srcStride[2], srcStride[3]));
         checkCudnn(cudnnSetFilter4dDescriptor(cudnnContext.filterDesc, dataType, tensorFormat, outDepth, inDepth, kH,
                         kW));
-        checkCudnn(cudnnSetConvolution2dDescriptor_v5(cudnnContext.convDesc, pad[0], pad[1], strides[0], strides[1], dilation[0],
+        checkCudnn(cudnnSetConvolution2dDescriptor(cudnnContext.convDesc, pad[0], pad[1], strides[0], strides[1], dilation[0],
                         dilation[1], CUDNN_CROSS_CORRELATION, dataType));
 
         // find dimension of convolution output

diff --git a/...learning4j-cuda/src/main/java/org/deeplearning4j/nn/layers/recurrent/CudnnLSTMHelper.java b/...learning4j-cuda/src/main/java/org/deeplearning4j/nn/layers/recurrent/CudnnLSTMHelper.java
@@ -450,9 +450,9 @@ public FwdPassReturn activate(final Layer layer, final NeuralNetConfiguration co
         checkCudnn(cudnnSetDropoutDescriptor(cudnnContext.dropoutDesc, cudnnContext, dropout, stateSpace, stateSize,
                         Nd4j.getRandom().getSeed()));
 
-        checkCudnn(cudnnSetRNNDescriptor(cudnnContext.rnnDesc, hiddenLayerSize, numLayers, cudnnContext.dropoutDesc,
-                        CUDNN_LINEAR_INPUT, bidirectional ? CUDNN_BIDIRECTIONAL : CUDNN_UNIDIRECTIONAL, RNNMode,
-                        dataType));
+        checkCudnn(cudnnSetRNNDescriptor_v6(cudnnContext, cudnnContext.rnnDesc, hiddenLayerSize, numLayers, cudnnContext.dropoutDesc,
+                         CUDNN_LINEAR_INPUT, bidirectional ? CUDNN_BIDIRECTIONAL : CUDNN_UNIDIRECTIONAL, RNNMode,
+                        CUDNN_RNN_ALGO_STANDARD, dataType));
 
         cudnnTensorStruct xDesc0 = xDesc.get(cudnnTensorStruct.class, 0);
         checkCudnn(cudnnGetRNNParamsSize(cudnnContext, cudnnContext.rnnDesc, xDesc0, sizeInBytes, dataType));

diff --git a/pom.xml b/pom.xml
@@ -362,7 +362,7 @@
                 </dependency>
                 <dependency>
                     <groupId>org.nd4j</groupId>
-                    <artifactId>nd4j-cuda-8.0</artifactId>
+                    <artifactId>nd4j-cuda-9.0</artifactId>
                     <version>${nd4j.version}</version>
                     <scope>test</scope>
                 </dependency>