Uses of Class
org.apache.sysds.runtime.instructions.gpu.context.GPUContext

Packages that use GPUContext

Package	Description
org.apache.sysds.runtime.controlprogram.caching
org.apache.sysds.runtime.controlprogram.context
org.apache.sysds.runtime.instructions.gpu.context
org.apache.sysds.runtime.matrix.data

Uses of GPUContext in org.apache.sysds.runtime.controlprogram.caching

Methods in org.apache.sysds.runtime.controlprogram.caching with parameters of type GPUContext

Modifier and Type	Method and Description
GPUObject	CacheableData.getGPUObject(GPUContext gCtx)
void	CacheableData.setGPUObject(GPUContext gCtx, GPUObject gObj)

Uses of GPUContext in org.apache.sysds.runtime.controlprogram.context

Methods in org.apache.sysds.runtime.controlprogram.context that return GPUContext

Modifier and Type	Method and Description
GPUContext	ExecutionContext.getGPUContext(int index) Get the i-th GPUContext

Methods in org.apache.sysds.runtime.controlprogram.context that return types with arguments of type GPUContext

Modifier and Type	Method and Description
List<GPUContext>	ExecutionContext.getGPUContexts() Gets the list of GPUContexts

Method parameters in org.apache.sysds.runtime.controlprogram.context with type arguments of type GPUContext

Modifier and Type	Method and Description
void	ExecutionContext.setGPUContexts(List<GPUContext> gpuContexts) Sets the list of GPUContexts

Uses of GPUContext in org.apache.sysds.runtime.instructions.gpu.context

Methods in org.apache.sysds.runtime.instructions.gpu.context that return types with arguments of type GPUContext

Modifier and Type	Method and Description
static List<GPUContext>	GPUContextPool.reserveAllGPUContexts() Reserves and gets an initialized list of GPUContexts

Methods in org.apache.sysds.runtime.instructions.gpu.context with parameters of type GPUContext

Modifier and Type	Method and Description
static CSRPointer	CSRPointer.allocateEmpty(GPUContext gCtx, long nnz2, long rows) Factory method to allocate an empty CSR Sparse matrix on the GPU
static CSRPointer	CSRPointer.allocateForDgeam(GPUContext gCtx, jcuda.jcusparse.cusparseHandle handle, CSRPointer A, CSRPointer B, int m, int n) Estimates the number of non zero elements from the results of a sparse cusparseDgeam operation C = a op(A) + b op(B)
static CSRPointer	CSRPointer.allocateForMatrixMultiply(GPUContext gCtx, jcuda.jcusparse.cusparseHandle handle, CSRPointer A, int transA, CSRPointer B, int transB, int m, int n, int k) Estimates the number of non-zero elements from the result of a sparse matrix multiplication C = A * B and returns the CSRPointer to C with the appropriate GPU memory.
static CSRPointer	GPUObject.columnMajorDenseToRowMajorSparse(GPUContext gCtx, jcuda.jcusparse.cusparseHandle cusparseHandle, jcuda.Pointer densePtr, int rows, int cols) Convenience method to convert a CSR matrix to a dense matrix on the GPU Since the allocated matrix is temporary, bookkeeping is not updated.
static void	CSRPointer.copyToDevice(GPUContext gCtx, CSRPointer dest, int rows, long nnz, int[] rowPtr, int[] colInd, double[] values) Static method to copy a CSR sparse matrix from Host to Device
static jcuda.Pointer	GPUObject.transpose(GPUContext gCtx, jcuda.Pointer densePtr, int m, int n, int lda, int ldc) Transposes a dense matrix on the GPU by calling the cublasDgeam operation

Constructors in org.apache.sysds.runtime.instructions.gpu.context with parameters of type GPUContext

Constructor and Description
GPUMemoryManager(GPUContext gpuCtx)

Uses of GPUContext in org.apache.sysds.runtime.matrix.data

Methods in org.apache.sysds.runtime.matrix.data with parameters of type GPUContext

Modifier and Type	Method and Description
static void	LibMatrixCUDA.abs(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "abs" operation on a matrix on the GPU
static void	LibMatrixCUDA.acos(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "acos" operation on a matrix on the GPU
static void	LibMatrixCUDA.asin(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "asin" operation on a matrix on the GPU
static void	LibMatrixCUDA.atan(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "atan" operation on a matrix on the GPU
static void	LibMatrixCUDA.axpy(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName, double constant) Performs daxpy operation
static void	LibMatrixCuDNN.batchNormalizationBackward(GPUContext gCtx, String instName, MatrixObject image, MatrixObject dout, MatrixObject scale, MatrixObject dX, MatrixObject dScale, MatrixObject dBias, double epsilon, MatrixObject resultSaveMean, MatrixObject resultSaveInvVariance) This method computes the backpropagation errors for image, scale and bias of batch normalization layer
static void	LibMatrixCuDNN.batchNormalizationForwardInference(GPUContext gCtx, String instName, MatrixObject image, MatrixObject scale, MatrixObject bias, MatrixObject runningMean, MatrixObject runningVar, MatrixObject ret, double epsilon) Performs the forward BatchNormalization layer computation for inference
static void	LibMatrixCuDNN.batchNormalizationForwardTraining(GPUContext gCtx, String instName, MatrixObject image, MatrixObject scale, MatrixObject bias, MatrixObject runningMean, MatrixObject runningVar, MatrixObject ret, MatrixObject retRunningMean, MatrixObject retRunningVar, double epsilon, double exponentialAverageFactor, MatrixObject resultSaveMean, MatrixObject resultSaveInvVariance) Performs the forward BatchNormalization layer computation for training
static void	LibMatrixCUDA.biasAdd(GPUContext gCtx, String instName, MatrixObject input, MatrixObject bias, MatrixObject outputBlock) Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input + matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_add(input, bias) built-in function
static void	LibMatrixCUDA.biasMultiply(GPUContext gCtx, String instName, MatrixObject input, MatrixObject bias, MatrixObject outputBlock) Performs the operation corresponding to the DML script: ones = matrix(1, rows=1, cols=Hout*Wout) output = input * matrix(bias %*% ones, rows=1, cols=F*Hout*Wout) This operation is often followed by conv2d and hence we have introduced bias_add(input, bias) built-in function
static void	LibMatrixCUDA.cbind(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName)
static void	LibMatrixCUDA.ceil(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "ceil" operation on a matrix on the GPU
static void	LibMatrixCUDA.channelSums(GPUContext gCtx, String instName, MatrixObject input, MatrixObject outputBlock, long C, long HW) Perform channel_sums operations: out = rowSums(matrix(colSums(A), rows=C, cols=HW))
static int	LibMatrixCUDA.computeNNZ(GPUContext gCtx, jcuda.Pointer densePtr, int length) Utility to compute number of non-zeroes on the GPU
static void	LibMatrixCuDNN.conv2d(GPUContext gCtx, String instName, MatrixObject image, MatrixObject filter, MatrixObject outputBlock, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, double intermediateMemoryBudget) Performs a 2D convolution
static void	LibMatrixCuDNN.conv2dBackwardData(GPUContext gCtx, String instName, MatrixObject filter, MatrixObject dout, MatrixObject output, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, double intermediateMemoryBudget) This method computes the backpropogation errors for previous layer of convolution operation
static void	LibMatrixCuDNN.conv2dBackwardFilter(GPUContext gCtx, String instName, MatrixObject image, MatrixObject dout, MatrixObject outputBlock, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, double intermediateMemoryBudget) This method computes the backpropogation errors for filter of convolution operation
static void	LibMatrixCuDNN.conv2dBiasAdd(GPUContext gCtx, String instName, MatrixObject image, MatrixObject bias, MatrixObject filter, MatrixObject output, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, double intermediateMemoryBudget) Does a 2D convolution followed by a bias_add
static void	LibMatrixCUDA.cos(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "cos" operation on a matrix on the GPU
static void	LibMatrixCUDA.cosh(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "cosh" operation on a matrix on the GPU
static LibMatrixCuDNNConvolutionAlgorithm	LibMatrixCuDNNConvolutionAlgorithm.cudnnGetConvolutionBackwardDataAlgorithm(GPUContext gCtx, String instName, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, long workspaceLimit) Factory method to get the algorithm wrapper for convolution backward data
static LibMatrixCuDNNConvolutionAlgorithm	LibMatrixCuDNNConvolutionAlgorithm.cudnnGetConvolutionBackwardFilterAlgorithm(GPUContext gCtx, String instName, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, long workspaceLimit) Factory method to get the algorithm wrapper for convolution backward filter
static LibMatrixCuDNNConvolutionAlgorithm	LibMatrixCuDNNConvolutionAlgorithm.cudnnGetConvolutionForwardAlgorithm(GPUContext gCtx, String instName, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, long workspaceLimit) Factory method to get the algorithm wrapper for convolution forward
static LibMatrixCuDNNPoolingDescriptors	LibMatrixCuDNNPoolingDescriptors.cudnnPoolingBackwardDescriptors(GPUContext gCtx, String instName, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, LibMatrixDNN.PoolingType poolingType) Get descriptors for maxpooling backward operation
static LibMatrixCuDNNPoolingDescriptors	LibMatrixCuDNNPoolingDescriptors.cudnnPoolingDescriptors(GPUContext gCtx, String instName, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, LibMatrixDNN.PoolingType poolingType) Get descriptors for maxpooling operation
static void	LibMatrixCUDA.cumulativeScan(ExecutionContext ec, GPUContext gCtx, String instName, String kernelFunction, MatrixObject in, String outputName) Cumulative scan
static void	LibMatrixCUDA.cumulativeSumProduct(ExecutionContext ec, GPUContext gCtx, String instName, String kernelFunction, MatrixObject in, String outputName) Cumulative sum-product kernel cascade invokation
static void	LibMatrixCUDA.denseTranspose(ExecutionContext ec, GPUContext gCtx, String instName, jcuda.Pointer A, jcuda.Pointer C, long numRowsA, long numColsA) Computes C = t(A)
void	CudaSupportFunctions.deviceToHost(GPUContext gCtx, jcuda.Pointer src, double[] dest, String instName, boolean isEviction)
void	DoublePrecisionCudaSupportFunctions.deviceToHost(GPUContext gCtx, jcuda.Pointer src, double[] dest, String instName, boolean isEviction)
void	SinglePrecisionCudaSupportFunctions.deviceToHost(GPUContext gCtx, jcuda.Pointer src, double[] dest, String instName, boolean isEviction)
static jcuda.Pointer	LibMatrixCUDA.double2float(GPUContext gCtx, jcuda.Pointer A, jcuda.Pointer ret, int numElems)
static void	LibMatrixCUDA.exp(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "exp" operation on a matrix on the GPU
static jcuda.Pointer	LibMatrixCUDA.float2double(GPUContext gCtx, jcuda.Pointer A, jcuda.Pointer ret, int numElems)
static void	LibMatrixCUDA.floor(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "floor" operation on a matrix on the GPU
static JCudaKernels	LibMatrixCUDA.getCudaKernels(GPUContext gCtx)
static jcuda.Pointer	LibMatrixCUDA.getDensePointer(GPUContext gCtx, MatrixObject input, String instName) Convenience method to get jcudaDenseMatrixPtr.
static jcuda.Pointer	LibMatrixCuDNN.getDensePointerForCuDNN(GPUContext gCtx, MatrixObject image, String instName, int numRows, int numCols) Convenience method to get jcudaDenseMatrixPtr.
static long	LibMatrixCUDA.getNnz(GPUContext gCtx, String instName, MatrixObject mo, boolean recomputeDenseNNZ) Note: if the matrix is in dense format, it explicitly re-computes the number of nonzeros.
void	CudaSupportFunctions.hostToDevice(GPUContext gCtx, double[] src, jcuda.Pointer dest, String instName)
void	DoublePrecisionCudaSupportFunctions.hostToDevice(GPUContext gCtx, double[] src, jcuda.Pointer dest, String instName)
void	SinglePrecisionCudaSupportFunctions.hostToDevice(GPUContext gCtx, double[] src, jcuda.Pointer dest, String instName)
static boolean	LibMatrixCUDA.isInSparseFormat(GPUContext gCtx, MatrixObject mo)
static void	LibMatrixCUDA.log(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "log" operation on a matrix on the GPU
static void	LibMatrixCuDNN.lstm(ExecutionContext ec, GPUContext gCtx, String instName, jcuda.Pointer X, jcuda.Pointer wPointer, jcuda.Pointer out0, jcuda.Pointer c0, boolean return_sequences, String outputName, String cyName, int N, int M, int D, int T) Computes the forward pass for an LSTM layer with M neurons.
static void	LibMatrixCuDNN.lstmBackward(ExecutionContext ec, GPUContext gCtx, String instName, jcuda.Pointer x, jcuda.Pointer hx, jcuda.Pointer cx, jcuda.Pointer wPointer, String doutName, String dcyName, String dxName, String dwName, String dbName, String dhxName, String dcxName, boolean return_sequences, int N, int M, int D, int T)
static MatrixObject	LibMatrixCuMatMult.matmult(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject left, MatrixObject right, String outputName, boolean isLeftTransposed, boolean isRightTransposed) Matrix multiply on GPU Examines sparsity and shapes and routes call to appropriate method from cuBLAS or cuSparse C = op(A) x op(B) The user is expected to call ec.releaseMatrixOutputForGPUInstruction(outputName);
static void	LibMatrixCUDA.matmultTSMM(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject left, String outputName, boolean isLeftTransposed) Performs tsmm, A %*% A' or A' %*% A, on GPU by exploiting cublasDsyrk(...)
static void	LibMatrixCUDA.matrixMatrixArithmetic(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName, boolean isLeftTransposed, boolean isRightTransposed, BinaryOperator op) Performs elementwise arithmetic operation specified by op of two input matrices in1 and in2
static void	LibMatrixCUDA.matrixMatrixRelational(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName, BinaryOperator op) Performs elementwise operation relational specified by op of two input matrices in1 and in2
static void	LibMatrixCUDA.matrixScalarArithmetic(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in, String outputName, boolean isInputTransposed, ScalarOperator op) Entry point to perform elementwise matrix-scalar arithmetic operation specified by op
static void	LibMatrixCUDA.matrixScalarOp(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in, String outputName, boolean isInputTransposed, ScalarOperator op) Utility to do matrix-scalar operation kernel
static void	LibMatrixCUDA.matrixScalarRelational(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in, String outputName, ScalarOperator op) Entry point to perform elementwise matrix-scalar relational operation specified by op
static void	LibMatrixCuDNN.pooling(GPUContext gCtx, String instName, MatrixObject image, MatrixObject outputBlock, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, LibMatrixDNN.PoolingType poolingType, double intermediateMemoryBudget) performs maxpooling on GPU by exploiting cudnnPoolingForward(...)
static void	LibMatrixCuDNN.poolingBackward(GPUContext gCtx, String instName, MatrixObject image, MatrixObject dout, MatrixObject maxpoolOutput, MatrixObject outputBlock, int N, int C, int H, int W, int K, int R, int S, int pad_h, int pad_w, int stride_h, int stride_w, int P, int Q, LibMatrixDNN.PoolingType poolingType, double intermediateMemoryBudget) Performs maxpoolingBackward on GPU by exploiting cudnnPoolingBackward(...) This method computes the backpropogation errors for previous layer of maxpooling operation
static void	LibMatrixCUDA.rbind(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName)
static void	LibMatrixCuDNN.relu(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in, String outputName) Performs the relu operation on the GPU.
static void	LibMatrixCUDA.reluBackward(GPUContext gCtx, String instName, MatrixObject input, MatrixObject dout, MatrixObject outputBlock) This method computes the backpropagation errors for previous layer of relu operation
static void	LibMatrixCUDA.round(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "round" operation on a matrix on the GPU
static void	LibMatrixCUDA.sigmoid(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "sigmoid" operation on a matrix on the GPU
static void	LibMatrixCUDA.sign(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "sign" operation on a matrix on the GPU
static void	LibMatrixCUDA.sin(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "sin" operation on a matrix on the GPU
static void	LibMatrixCUDA.sinh(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "sinh" operation on a matrix on the GPU
static void	LibMatrixCUDA.sliceOperations(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, IndexRange ixrange, String outputName) Method to perform rightIndex operation for a given lower and upper bounds in row and column dimensions.
static void	LibMatrixCuDNN.softmax(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "softmax" operation on a matrix on the GPU
static void	LibMatrixCUDA.solve(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, MatrixObject in2, String outputName) Implements the "solve" function for systemds Ax = B (A is of size m*n, B is of size m*1, x is of size n*1)
static void	LibMatrixCUDA.sqrt(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "sqrt" operation on a matrix on the GPU
static void	LibMatrixCUDA.tan(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "tan" operation on a matrix on the GPU
static void	LibMatrixCUDA.tanh(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String outputName) Performs an "tanh" operation on a matrix on the GPU
static void	LibMatrixCUDA.transpose(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in, String outputName) Transposes the input matrix using cublasDgeam
static void	LibMatrixCUDA.unaryAggregate(ExecutionContext ec, GPUContext gCtx, String instName, MatrixObject in1, String output, AggregateUnaryOperator op) Entry point to perform Unary aggregate operations on the GPU.

Constructors in org.apache.sysds.runtime.matrix.data with parameters of type GPUContext

Constructor and Description
LibMatrixCuDNNInputRowFetcher(GPUContext gCtx, String instName, MatrixObject image) Initialize the input fetcher
LibMatrixCuDNNRnnAlgorithm(ExecutionContext ec, GPUContext gCtx, String instName, String rnnMode, int N, int T, int M, int D, boolean isTraining, jcuda.Pointer w)