cavis/libnd4j/include/helpers/cuda_off/MmulHelper.cu

416 lines
19 KiB
Plaintext
Raw Normal View History

2019-06-06 14:21:15 +02:00
/*******************************************************************************
* Copyright (c) 2015-2018 Skymind, Inc.
*
* This program and the accompanying materials are made available under the
* terms of the Apache License, Version 2.0 which is available at
* https://www.apache.org/licenses/LICENSE-2.0.
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
* SPDX-License-Identifier: Apache-2.0
******************************************************************************/
//
// @author raver119@gmail.com
// @author Yurii Shyrma (iuriish@yahoo.com)
//
#include <exceptions/cuda_exception.h>
#include <cublas_v2.h>
#include "../MmulHelper.h"
#include <specials_cuda.h>
namespace nd4j {
//////////////////////////////////////////////////////////////////////////////
// MXK x KxN = MxN
// C array must be in f order
template <typename T1, typename T2, typename T3>
static __global__ void usualCudaGemm(const bool transA, const bool transB, const int M, const int N, const int K, const double alpha, const void* vA, const int lda, const void* vB, const int ldb, const double beta, void* vC, const int ldc) {
T1* A = reinterpret_cast<T1*>(const_cast<void*>(vA));
T2* B = reinterpret_cast<T2*>(const_cast<void*>(vB));
T3* C = reinterpret_cast<T3*>(vC);
__shared__ T3 alphaZ, betaZ;
__shared__ Nd4jLong strideArow, strideAcol, strideBrow, strideBcol;
const int row = blockIdx.y * blockDim.y + threadIdx.y;
const int col = blockIdx.x * blockDim.x + threadIdx.x;
if(row == 0 && col == 0) {
alphaZ = alpha;
betaZ = beta;
if(transA) { strideArow = lda; strideAcol = 1; } else { strideArow = 1; strideAcol = lda; }
if(transB) { strideBrow = ldb; strideBcol = 1; } else { strideBrow = 1; strideBcol = ldb; }
}
__syncthreads();
T3 val = 0;
if (row < M && col < N)
for (int i = 0; i < K; i++)
val = val + A[row * strideArow + i * strideAcol] * B[i * strideBrow + col * strideBcol];
C[row + col * ldc] = alphaZ * val + betaZ * C[row + col * ldc];
}
////////////////////////////////////////////////////////////////////////
template <typename T1, typename T2, typename T3>
__host__ static void usualGemm(const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const bool transA, const bool transB, const int M, const int N, const int K, const double alpha, const void* vA, const int lda, const void* vB, const int ldb, const double beta, void* vC, const int ldc) {
usualCudaGemm<T1,T2,T3><<<blocksPerGrid, threadsPerBlock, 1024, *stream>>>(transA, transB, M, N, K, alpha, vA, lda, vB, ldb, beta, vC, ldc);
}
//////////////////////////////////////////////////////////////////////////////
// MXN x N = M
template <typename T1, typename T2, typename T3>
static __global__ void usualCudaGemv(const bool transA, const int M, const int N, const double alpha, const void* vA, const int lda, const void* vX, const int incx, const double beta, void* vY, const int incy) {
T1* A = reinterpret_cast<T1*>(const_cast<void*>(vA));
T2* X = reinterpret_cast<T2*>(const_cast<void*>(vX));
T3* Y = reinterpret_cast<T3*>(vY);
__shared__ T3 alphaZ, betaZ;
__shared__ Nd4jLong strideArow, strideAcol;
const int row = blockIdx.x * blockDim.x + threadIdx.x;
if(row == 0) {
alphaZ = alpha;
betaZ = beta;
if(transA) { strideArow = lda; strideAcol = 1; } else { strideArow = 1; strideAcol = lda; }
}
__syncthreads();
T3 val = 0;
if (row < M)
[WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io>
2019-08-02 19:01:03 +02:00
for (int i = 0; i < N; i++)
2019-06-06 14:21:15 +02:00
val = val + A[row * strideArow + i * strideAcol] * X[i * incx];
Y[row * incy] = alphaZ * val + betaZ * Y[row * incy];
}
////////////////////////////////////////////////////////////////////////
template <typename T1, typename T2, typename T3>
__host__ static void usualGemv(const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const bool transA, const int M, const int N, const double alpha, const void* vA, const int lda, const void* vX, const int incx, const double beta, void* vY, const int incy) {
usualCudaGemv<T1,T2,T3><<<blocksPerGrid, threadsPerBlock, 1024, *stream>>>(transA, M, N, alpha, vA, lda, vX, incx, beta, vY, incy);
}
//////////////////////////////////////////////////////////////////////////////
template <typename T1, typename T2, typename T3>
static __global__ void usualCudaDot(const Nd4jLong length, const double alpha, const void* vX, const Nd4jLong incx, const void* vY, const Nd4jLong incy, const double beta, void* vZ) {
T1* X = reinterpret_cast<T1*>(const_cast<void*>(vX));
T2* Y = reinterpret_cast<T2*>(const_cast<void*>(vY));
T3* Z = reinterpret_cast<T3*>(vZ);
extern __shared__ char shmem[];
auto pairwiseMul = reinterpret_cast<T3*>(shmem);
const int tid = blockIdx.x * blockDim.x + threadIdx.x;
if(tid < length)
pairwiseMul[tid] = X[tid * incx] * Y[tid * incy];
__syncthreads();
if(tid == 0) {
T3 sum = 0;
for(Nd4jLong i = 0; i < length; ++i)
sum = sum + pairwiseMul[i];
*Z = (T3)alpha * sum + (T3)beta * *Z;
}
}
////////////////////////////////////////////////////////////////////////
template <typename T1, typename T2, typename T3>
__host__ static void usualDot(const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const Nd4jLong length, const double alpha, const void* vX, const Nd4jLong incx, const void* vY, const Nd4jLong incy, const double beta, void* vZ) {
usualCudaDot<T1,T2,T3><<<blocksPerGrid, threadsPerBlock, length*sizeof(T3) + 128, *stream>>>(length, alpha, vX, incx, vY, incy, beta, vZ);
}
//////////////////////////////////////////////////////////////////////////////
// MXK x KxN = MxN
NDArray* MmulHelper::mmulMxM(const NDArray* A, const NDArray* B, NDArray* C, double alpha, double beta, const char outOrder) {
if(A->rankOf() != 2)
throw std::runtime_error("MmulHelper::mmulMxM cuda: rank of A array is not equal 2 !");
if(B->rankOf() != 2)
throw std::runtime_error("MmulHelper::mmulMxM cuda: rank of B array is not equal 2 !");
auto M = A->sizeAt(0);
auto K = A->sizeAt(1);
auto N = B->sizeAt(1);
if(C != nullptr && C->rankOf() != 2)
throw std::runtime_error("MmulHelper::mmulMxM cuda: rank of C array is not equal 2 !");
if(B->sizeAt(0) != K)
throw std::runtime_error("MmulHelper::mmulMxM cuda: B array has wrong number of rows !");
if(C != nullptr && C->sizeAt(0) != M)
throw std::runtime_error("MmulHelper::mmulMxM cuda: C array has wrong number of rows !");
if(C != nullptr && C->sizeAt(1) != N)
throw std::runtime_error("MmulHelper::mmulMxM cuda: C array has wrong number of columns !");
if(C == nullptr)
C = new NDArray(outOrder, {M,N}, DataTypeUtils::pickPairwiseResultType(A->dataType(), B->dataType()), A->getContext());
NDArray *pA(const_cast<NDArray*>(A)), *pB(const_cast<NDArray*>(B)), *pC(const_cast<NDArray*>(C));
std::vector<NDArray*> toDelete;
if(A->ews() != 1) {
pA = pA->dup('f');
toDelete.push_back(pA);
}
if(B->ews() != 1) {
pB = pB->dup('f');
toDelete.push_back(pB);
}
if(C->ews() != 1) {
pC = pC->dup('f');
toDelete.push_back(pC);
}
if(pC->ordering() != 'f') {
auto temp = pA;
Merge master to upstream (#7945) * Shugeo strided slice zeros (#14) * Modified strided_slice op to properly work with empty-like shapes. * Fixed test for reduce_mean with empty-like input. * [WIP] Last merge (#15) * correct logsoftmax looss (#2) * Small SameDiff listener fix (#4) * Various fixes (#6) * #7839 Fix for asXMatrix and tests * #7866 EmbeddingSequenceLayer dtype fix + test * #7856 SameDiff save/load stream methods * #7859 RegressionEvaluation rank 4 fix + tests + axis configuration * EvaluationBinary 3d/4d * More evaluation 3d/4d tests * #7847 Evaluation empty checks * Small test ifx * #7848 Fix median edge case * Improve DL4J samediff layer tests * [WIP] FastText wrapper implemented (#8) * FastText implemented * Some fixes * Fix shapes for wordsNearest * Validation of input vectors * Fixes * Fixed test * Thread tagged * Some tweaks * setContextClassLoader for DeallocatorServiceThread * Numpy format tests (#1) * Various fixes (#11) * #7852 SameDiff gather fix * #7892 SameDiff placeholder to constant conversion * #7890 validate input rank for MLN/CG init methods * Fix broken permute shape calculation * Permute and gather fixes * Tests * #7850 LogSumExp fix + test * Handful of test fixes * Empty arrays with non-scalar shapes (#10) * minor rearrangements for lambdas * empty tensors with non-scalar shapes * numpy empty tensors with non-scalar shapes * few more empty tweaks * Small fixes * conv3d signature update * micro fix in batchnorm mkldnn * Import fixes * Fix * MKL-DNN update * Small fill fix * fill with empty input + test * Fixes * Small error improvement * Fix * one special test * couple of fixes for lstm * Rewrite TFGraphMapper.getNDArrayFromTensor to be maintainable and less error prone * Fixes * FP16 * Unsigned * BFloat16 * Fill op - empty tweaks * - couple of fixes for empty arrays construction - stack updated * strided slice fix * one transform test * provide method for reducing shapeInfo in case of input array is empty * Fixed reduceAlongDimensions to use empty input properly. * couple of broadcast tests * couple of tests broadcast tests + tweak to make them pass * add check of non-empty to methods producing sub-arrays * Fixed reshapeC with zeros in shape. * complete empty check in reduce_... legacy ops * Concat and cumsum/prod * Tweak to empty shape inference on import * add empty check to the rest of reduce legacy ops * one more test * correct typo in evalReduceShapeInfoEmpty * Added tests for reduce_* ops to tests with zero shapes. * few more tests for empty reductions * Fixed strided_slice op with empty case and tests. * one more empty reduction test * Fixed strided_slice test. * add empty check to NDArray::reshapei * infOrMax * empty min/max with infinity tests * made unstack working correctly with empty arrays * few IndexReduce tests + tweaks for empty shapes * add test for empty concat * few tests fixed * Validation fix for reductions on empty shapes * Reverse fix * Reduction shape calc fixes * SameDiff.generateOutputVariable: don't use shape function to determine number of outputs * Range fix * - NDArray constructor updated for scalars/empty arrays - few tests fixed * More fixes * Empty creator fixes * concat fix * concat fix * TF import tests: allow 'both all NaN' and 'both all inf' to pass * Slice, zero fraction, and reshape fixes * transpose, gather * Zero fraction * scalar cast fix * Empty reduction axis support * few more tests fixed * Fixed input checks conforming with TF for concat op and tests. * few tests fixed * matmul scalar shape fix * Fixed checkout for data type and scalarity with concat to allow non-empty scalars with vector concats. * broadcast bool fix * few more tests * few more tests * correct evalReduceShapeInfoEmpty * argmax/argmin + tests * one more empty edge case + one more test * argmax/argmin/realdiv_bp tweaks * empty reshape test + fix * Helper fixes * Small fixes * Gather test fix * Gather test fix * Small fixes * reduce scalar zero values * scalar mean workaround * Remove debug code * along dim mean workaround * one more test * - equalsTo() tweak for empty arrays - one more test * broadcast tweaks * [WIP] Fixing outstanding issues for NLP (#9) * Avoid using not-inited objects * Test fixed. * Redundant method avoided for models like FastText * KMeans++ implementation * KMeans++ implementation * Disable parallel execution * KMeans++ * Tests * Dev branch merge (#16) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Fix some issues on master (#17) * Fix DataVec test issue * Fix issue with dl4j SameDiff output layer * Dtype fix for lambda layers * #7912 BertIterator dtype fix (use float32 not global default) * [WIP] Next set of CUDA stuff (#7) New CUDA implementations and improvements * bad file * Dev branch master merge (#23) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * SameDiff ops, TF import and fixes (#24) * CheckNumerics tests + fixes + misc fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fake quant Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * FakeQuantWithMinMaxArgs Signed-off-by: AlexDBlack <blacka101@gmail.com> * CheckNumerics fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix libnd4j ALL_INTS and ALL_FLOATS declaration (uint and bfloat types) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Javadoc Signed-off-by: AlexDBlack <blacka101@gmail.com> * Exception tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix for out of scope stack allocated var use Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignores Signed-off-by: AlexDBlack <blacka101@gmail.com> * Ignore for known failing test (already logged issue) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Merge upstream to fork (#25) * Add thousand-separator commas to TotalParams (#7915) * Add thousand-separator commas to TotalParams The number of parameters can be quite large, and it would help the reading of the summary printout to have the TotalParams column & values at the bottom have thousand-separator-commas in them. * Add thousand-separator commas to MultiLayerNetwork Corresponding change to MultiLayerNetwork Signed-off-by: Jxtps Jxtps <jxtps435@gmail.com> * Update contributing and issue/PR templates (#7934) Signed-off-by: AlexDBlack <blacka101@gmail.com> * Fix link to AdaDelta paper (#7942) Fix link to AdaDelta paper hosted on matthewzeiler.com Signed-off-by: Jxtps * Fixes, and ignores for known/logged failing issues (#7943) Signed-off-by: AlexDBlack <blacka101@gmail.com> * SameDiff + DL4J/SameDiff: Multiple fixes (#28) * #7919 HDF5 attribute buffer length fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7909 Arbiter constructor exception ux improvements Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7925 RNN output layer length checks Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Add listener for validating inputs are not incorrectly modified Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7939 Integrate NonInplaceValidationListener into tests * #7844 DL4J SameDiff fixes for variable minibatch size * DL4J SameDiff fixes - ensure gradient for input placeholder is available Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tweaks to ExternalErrorsFunction - use placeholders, make more robust * Another fix * More fixes * More SameDiff/DL4J fixes * Scope out scalar array creation in BaseScalarOp * Remove debug code Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Final dev branch merge (#29) * SameDiff: convertDataType and gradient check util improvements (#12) * GradCheck util improvements * StopGradient constructor + test * SameDiff: Add datatype conversion * Javadoc and add DataType.isNumerical() * Small fix * Fix SameDiff TF import test cases intermediate naming (workaround for bad default) * TFGraphTestAllHelper: check intermediates in execution order * Add missing debug listener * [WIP] lstmBlock fix + other changes (#13) - fixes lstmBlock issue - changes NDArray method reshape(), permute(), transpose() by making them return instance instead of pointer - CheckNumerics op - fixes for ReduceBool IsInfOrNan & IsFinite * Small test fix * CheckNumerics op wrapper * Compatibility of deserialization (#18) Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * SameDiff: add activation gradient checking support for debugging (#19) * SameDiff gradient checker: first pass on activation gradient checks * Fixes + tests for activation gradient checking * Javadoc * [WIP] Some nd4j data type corrections (#20) * Adjust data type * Set correct Data type. * Size of proper data type. * fix averaged cpu load (#22) * [WIP] Multiple dataset iterators (#27) * Splitting dataset into arbitrary number * Fixes * Multiple split of iterator * Test * Test * Some fixes * signature change * one more tweak Signed-off-by: raver119 <raver119@gmail.com> * one more test for sequential use of DataSetIteratorSplitter Signed-off-by: raver119 <raver119@gmail.com> * Fixes * Fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * one more test for Alexander Signed-off-by: raver119 <raver119@gmail.com> * minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Some fixes * Some fixes * couple of assertions tweaked Signed-off-by: raver119 <raver119@gmail.com> * MDS splitter test :/ Signed-off-by: raver119 <raver119@gmail.com> * Minor refactoring * Multi dataset * Some fixes * More tests * Small number of test fixes/improvements (failures on CI) (#31) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] More CUDA stuff (#26) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * LRN BP CUDA Signed-off-by: raver119 <raver119@gmail.com> * less memory Signed-off-by: raver119 <raver119@gmail.com> * Fixed bug with crop_and_resize op helper. * get rid of unnecessary index-calculation dunction Signed-off-by: Yurii <yurii@skymind.io> * Fixed sort with nth_element cuda-based helper. * Refactored nth_element. * Refactored nth_element op and tests. * Modified usage of dim array with sortTad routine. * Refactored main routine of helper for non_max_image_suppression op. * non_max_image_suppression op helper with cuda kernel implementation. Initial revision. * fix vol2col cuda kernel * meh Signed-off-by: raver119 <raver119@gmail.com> * topK concept Signed-off-by: raver119 <raver119@gmail.com> * unsorted topK with scanWitdh of 1 Signed-off-by: raver119 <raver119@gmail.com> * correct vol2col tests * sorted/unsorted topK Signed-off-by: raver119 <raver119@gmail.com> * implementation and fixing col2im/col2vol * Corrected usage flags with input/output with reverse op. * dup is const now Signed-off-by: raver119 <raver119@gmail.com> * percentile op Signed-off-by: raver119 <raver119@gmail.com> * group tests for mapool2d Signed-off-by: Yurii <yurii@skymind.io> * special test for george Signed-off-by: raver119 <raver119@gmail.com> * less threads for sortTad Signed-off-by: raver119 <raver119@gmail.com> * provide conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * remove auther in sort tad kernel code Signed-off-by: Yurii <yurii@skymind.io> * provide depthwise_conv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * - max_pooling_with_argmax - null check for special use Signed-off-by: raver119 <raver119@gmail.com> * dts cuda Signed-off-by: raver119 <raver119@gmail.com> * provide sconv2d for cuda Signed-off-by: Yurii <yurii@skymind.io> * std cuda Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op to conform TF implementation. * Improved suppression helper. * provide pooling3d for cuda Signed-off-by: Yurii <yurii@skymind.io> * minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * more of minor lstm rearrangements Signed-off-by: raver119 <raver119@gmail.com> * (bi)dynamic_rnn Signed-off-by: raver119 <raver119@gmail.com> * templates init order Signed-off-by: raver119 <raver119@gmail.com> * Refactored non_max_suppression op. * Added cuda kernel for non_max_suppression. * CPU sort by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value Signed-off-by: raver119 <raver119@gmail.com> * CPU sort TAD by key/value tests Signed-off-by: raver119 <raver119@gmail.com> * Eliminate compiler error with cuda implementation. * - repaired gradCheck in cuda - provide conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * missed signature Signed-off-by: raver119 <raver119@gmail.com> * provide depthwise_conv2d_bp for cuda Signed-off-by: Yurii <yurii@skymind.io> * Implementation of lup helper with cuda kernel. Initial commit. * further work on backprops for convolutions Signed-off-by: Yurii <yurii@skymind.io> * CUDA linear sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * CUDA tad sort by key/val Signed-off-by: raver119 <raver119@gmail.com> * start providing of backprop for pooling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * Added atomicAdd for bool datatype. * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic partition scalar CUDA Signed-off-by: raver119 <raver119@gmail.com> * important comment Signed-off-by: raver119 <raver119@gmail.com> * fix pooling2d/3d backprop helpers Signed-off-by: Yurii <yurii@skymind.io> * Added non-linear test with dynamic_partition. * Improved test for dynamic_partition. * dynamic_partition TAD concept Signed-off-by: raver119 <raver119@gmail.com> * - dynamic_partition TAD CUDA impl - dynamic_partition TAD CPU fix Signed-off-by: raver119 <raver119@gmail.com> * - rewrite cpu code for usampling2d/3d - write cuda code for usampling2d/3d Signed-off-by: Yurii <yurii@skymind.io> * dynamic_stitch CUDA vector case Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case concept Signed-off-by: raver119 <raver119@gmail.com> * dynamic_stitch CUDA TAD case impl Signed-off-by: raver119 <raver119@gmail.com> * Added tests for dynamic_stitch 3D-4D cases. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * Fixed type check for dynamic stitch. * min/max bp Signed-off-by: raver119 <raver119@gmail.com> * rewrite code for upsampling2d/3d cpu Signed-off-by: Yurii <yurii@skymind.io> * reduce min/max/norm_max bp Signed-off-by: raver119 <raver119@gmail.com> * lup implementation. Additional enhancements. * provide code for upsamling2d/3d backprop Signed-off-by: Yurii <yurii@skymind.io> * weightedCrossEntropyWithLogits Signed-off-by: raver119 <raver119@gmail.com> * Fixed template math atomicMul for 64bit ints. * Refactored dynamic_partition_bp op. * inverseBroadcast fix Signed-off-by: raver119 <raver119@gmail.com> * DynamicPartitionBP test datatype fixed. * - nd4j_atomicMul Windows fix - cpu/NDArrayLambda.hpp excluded from CUDA Signed-off-by: raver119 <raver119@gmail.com>
2019-06-27 17:37:04 +02:00
pA = new NDArray(pB ->permute({1,0}));
pB = new NDArray(temp->permute({1,0}));
pC = new NDArray(pC ->permute({1,0}));
2019-06-06 14:21:15 +02:00
toDelete.push_back(pA);
toDelete.push_back(pB);
toDelete.push_back(pC);
M = pA->sizeAt(0);
K = pA->sizeAt(1);
N = pB->sizeAt(1);
}
const auto aOrder = pA->ordering();
const auto bOrder = pB->ordering();
const bool transA = aOrder != 'f';
const bool transB = bOrder != 'f';
const cublasOperation_t transAblas = transA ? CUBLAS_OP_T : CUBLAS_OP_N;
const cublasOperation_t transBblas = transB ? CUBLAS_OP_T : CUBLAS_OP_N;
const int lda = aOrder == 'f' ? M : K;
const int ldb = bOrder == 'f' ? K : N;
const int ldc = M; // cOrder == 'f' ? M : N;
const auto aType = pA->dataType();
const auto bType = pB->dataType();
const auto cType = pC->dataType();
auto handle = reinterpret_cast<cublasHandle_t *>(A->getContext()->getCublasHandle());
auto stream = A->getContext()->getCudaStream();
auto status = cublasSetStream_v2(*handle, *stream);
if (status != CUBLAS_STATUS_SUCCESS) throw cuda_exception::build("MmulHelper::mmulMxM cuda failed !", status);
const bool AB(aType == bType), AC(aType == cType), ABC(AB && AC);
NDArray::prepareSpecialUse({pC}, {pA, pB});
// choose appropriate cuda gemm api depending on data types
if(ABC && aType == DataType::DOUBLE) {
status = cublasDgemm(*handle, transAblas, transBblas, M, N, K, &alpha, (double*)pA->getSpecialBuffer(), lda, (double*)pB->getSpecialBuffer(), ldb, &beta, (double*)pC->getSpecialBuffer(), ldc);
}
else if(ABC && aType == DataType::FLOAT32) {
float alphaF(alpha), betaF(beta);
status = cublasSgemm(*handle, transAblas, transBblas, M, N, K, &alphaF, (float*)pA->getSpecialBuffer(), lda, (float*)pB->getSpecialBuffer(), ldb, &betaF, (float*)pC->getSpecialBuffer(), ldc);
}
#if __CUDA_ARCH__ >= 530
2019-06-06 14:21:15 +02:00
else if(ABC && aType == DataType::HALF) {
float16 alphaH(alpha), betaH(beta);
status = cublasHgemm(*handle, transAblas, transBblas, M, N, K, &alphaH.data, (__half*)pA->getSpecialBuffer(), lda, (__half*)pB->getSpecialBuffer(), ldb, &betaH.data, (__half*)pC->getSpecialBuffer(), ldc);
}
else if(AB && aType == DataType::INT8 && cType == DataType::FLOAT32) {
float alphaF(alpha), betaF(beta);
status = cublasSgemmEx(*handle, transAblas, transBblas, M, N, K, &alphaF, pA->getSpecialBuffer(), CUDA_R_8I, lda, pB->getSpecialBuffer(), CUDA_R_8I, ldb, &betaF, pC->getSpecialBuffer(), CUDA_R_32F, ldc);
}
else if(AB && aType == DataType::HALF && cType == DataType::FLOAT32) {
float alphaF(alpha), betaF(beta);
status = cublasSgemmEx(*handle, transAblas, transBblas, M, N, K, &alphaF, pA->getSpecialBuffer(), CUDA_R_16F, lda, pB->getSpecialBuffer(), CUDA_R_16F, ldb, &betaF, pC->getSpecialBuffer(), CUDA_R_32F, ldc);
}
#endif
2019-06-06 14:21:15 +02:00
else {
dim3 threadsPerBlock(N, M);
dim3 blocksPerGrid(1, 1);
if (M*N > 512){
threadsPerBlock.x = threadsPerBlock.y = 512;
blocksPerGrid.x = math::nd4j_ceil<double, int>(static_cast<double>(N) / threadsPerBlock.x); // cols
blocksPerGrid.y = math::nd4j_ceil<double, int>(static_cast<double>(M) / threadsPerBlock.y); // rows
}
[WIP] multi-device support (#80) * fix pad javadoc and @see links. (#72) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * [WIP] More fixes (#73) * special tests for ConstantTadHelper/ConstantShapeHelper Signed-off-by: raver119 <raver119@gmail.com> * release methods for data buffers Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary TadPack C++/Java side (#74) Signed-off-by: raver119 <raver119@gmail.com> * Zoo model TF import test updates (#75) * argLine fix, update compression_gru comment * updated comment for xception * undid but commented argLine change * updated xlnet comment * copyright headers * - new NDArray methods like()/ulike() (#77) - fix for depthwise_conv2d_bp + special test Signed-off-by: raver119 <raver119@gmail.com> * upsampling2d fix CUDA Signed-off-by: raver119 <raver119@gmail.com> * DL4J trace logging (#79) * MLN/CG trace logging for debugging Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tiny tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * strided_slice_bp shape fn leak fix Signed-off-by: raver119 <raver119@gmail.com> * SameDiff fixes and naming (#78) * remove SDVariable inplace methods * import methods * npe fix in OpVal * removed SameDiff inplace ops from tests * Naming updates, moved to centralized methods in SameDiff, should use op_#:# for everything * quick fixes * javadoc * SDVariable eval with placeholders * use regex match * better matching * initial commit Signed-off-by: raver119 <raver119@gmail.com> * initial commit Signed-off-by: raver119 <raver119@gmail.com> * fix javadoc. (#76) * fix javadoc. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace most @see with @link s. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * 4 additional tests Signed-off-by: raver119 <raver119@gmail.com> * launch context reorganization Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext reorganization Signed-off-by: raver119 <raver119@gmail.com> * per-device LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * Various DL4J/ND4J fixes (#81) * #7954 Force refresh of UI when switching tabs on overview page Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8017 Concurrent modification exception (synchronize) fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8033 Don't initialize updater in middle of writing memory crash dump Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8208 Fix shape checks for ND4J int[] creator methods Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6385 #7992 Keras import naming fixes + cleanup Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8016 Upsampling3D - add NDHWC format support Signed-off-by: AlexDBlack <blacka101@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Refactor NativeOps.h to export C functions * Actually export functions from NativeOps.h * Adapt the Java wrappers in ND4J generated with JavaCPP * Create C wrappers for some of the C++ classes currently used by ND4J * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * remove duplicate code in createBufferDetached. (#83) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Keras model import - updater lr fix (#84) * Keras model import - updater lr fix Signed-off-by: eraly <susan.eraly@gmail.com> * Keras model import - updater lr fix, cleanup Signed-off-by: eraly <susan.eraly@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Fix functions of OpaqueVariablesSet * thread-local buffers/affinity Signed-off-by: raver119 <raver119@gmail.com> * thread safety for LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * more of thread safety Signed-off-by: raver119 <raver119@gmail.com> * one more multi threaded test Signed-off-by: raver119 <raver119@gmail.com> * SameDiff Convolution Config validation, better output methods (#82) * Conv Config validation & tests Signed-off-by: Ryan Nett <rnett@skymind.io> * stackOutputs utility method Signed-off-by: Ryan Nett <rnett@skymind.io> * use constructor for validation, support negative kernel sizes (infered from weights) Signed-off-by: Ryan Nett <rnett@skymind.io> * better output methods Signed-off-by: Ryan Nett <rnett@skymind.io> * move output to be with fit and evaluate Signed-off-by: Ryan Nett <rnett@skymind.io> * fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * more fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * refactor duplicate code from pad methods. (#86) * refactor duplicate code from pad methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace switch with if. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes and improvements (#87) * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6488 ElementWiseVertex broadcast support Signed-off-by: AlexDBlack <blacka101@gmail.com> * Constructors and broadcast supported it Transforms.max/min Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8054 ElementWiseVertex now supports broadcast inputs Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8057 Nd4j.create overload dtype fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7551 ND4J Shape validation fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Numpy boolean import (#91) * numpy bool type Signed-off-by: raver119 <raver119@gmail.com> * numpy bool java side Signed-off-by: raver119 <raver119@gmail.com> * remove create method with unused parameter. (#89) * remove create method with unused parameter. * removed more unused methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * removing more unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * last removal of unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * remove createSparse methods. (#92) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes (#90) * Deprecate Old*Op instances Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8063 #8054 Broadcast exceptions + cleanup inplace ops Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Remove bad test condition Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7993 Fix shape function issue in crop_and_resize op Signed-off-by: AlexDBlack <blacka101@gmail.com> * DL4J SameDiff lambda layer fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8029 Fix for pnorm backprop math Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8038 Fix Op profiler NaN/Inf triggering + add tests (#93) Signed-off-by: AlexDBlack <blacka101@gmail.com> * createUninitializedDetached refactoring. (#94) * wip * update interface, add null implementations. * Breaking one test in a weird way. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * createUninitializedDetached refactored. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * cuda build fix for issues introduced by recent refactoring Signed-off-by: raver119 <raver119@gmail.com> * [WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io> * build fix Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI (#97) Signed-off-by: raver119 <raver119@gmail.com> * temporary stack fix Signed-off-by: raver119 <raver119@gmail.com> * round robin affinity test Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy CudaContext methods Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy ContextPool classes/methods Signed-off-by: raver119 <raver119@gmail.com> * one legacy test removed Signed-off-by: raver119 <raver119@gmail.com> * few more fields rearranged Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext++ Signed-off-by: raver119 <raver119@gmail.com> * more of OpaqueLaunchContext methods Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext -> CudaContext Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * cusolver handles Signed-off-by: raver119 <raver119@gmail.com> * typo Signed-off-by: raver119 <raver119@gmail.com> * cusolver method Signed-off-by: raver119 <raver119@gmail.com> * cusolver handle propagated Signed-off-by: raver119 <raver119@gmail.com> * blas/solver handles Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * legacy concat implementations replaced with new CustomOp Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * concat now uses way more blocks Signed-off-by: raver119 <raver119@gmail.com> * print Signed-off-by: raver119 <raver119@gmail.com> * no more triple template mmul Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bitonic sort reorganized Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * type conversions moved to generic impl Signed-off-by: raver119 <raver119@gmail.com> * cpu data types pass Signed-off-by: raver119 <raver119@gmail.com> * non_max_suppression Signed-off-by: raver119 <raver119@gmail.com> * sortByValue fix Signed-off-by: raver119 <raver119@gmail.com> * ignore all mixed datatype tests for mmul Signed-off-by: raver119 <raver119@gmail.com> * special handling of OpProfiler exceptions Signed-off-by: raver119 <raver119@gmail.com> * - one failing concat test in cpp - Nd4j.tile now uses op internally Signed-off-by: raver119 <raver119@gmail.com> * get back dtype exception for legacy arrays deserialization Signed-off-by: raver119 <raver119@gmail.com>
2019-08-14 15:52:34 +02:00
//BUILD_TRIPLE_SELECTOR(aType, bType, cType, usualGemm, (blocksPerGrid, threadsPerBlock, stream, transA, transB, M, N, K, alpha, pA->getSpecialBuffer(), lda, pB->getSpecialBuffer(), ldb, beta, pC->getSpecialBuffer(), ldc), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
BUILD_SINGLE_SELECTOR_THRICE(aType, usualGemm, (blocksPerGrid, threadsPerBlock, stream, transA, transB, M, N, K, alpha, pA->getSpecialBuffer(), lda, pB->getSpecialBuffer(), ldb, beta, pC->getSpecialBuffer(), ldc), NUMERIC_TYPES)
2019-06-06 14:21:15 +02:00
}
if (status != CUBLAS_STATUS_SUCCESS) throw cuda_exception::build("MmulHelper::mmulMxM cuda failed !", status);
auto cudaResult = cudaStreamSynchronize(*stream);
if (cudaResult != 0) throw cuda_exception::build("MmulHelper::mmulMxM cuda failed !", cudaResult);
NDArray::registerSpecialUse({pC}, {pA, pB});
if(C->ews() != 1)
C->assign(pC);
for(int i = toDelete.size() - 1; i >= 0; --i)
delete toDelete[i];
return C;
}
////////////////////////////////////////////////////////////////////////////
// MXN x N = M
NDArray* MmulHelper::mmulMxV(const NDArray* A, const NDArray* X, nd4j::NDArray* Y, const double alpha, const double beta, const char outOrder) {
int xLenDim, yLenDim(0);
if(A->rankOf() != 2)
throw std::runtime_error("MmulHelper::mmulMxV cuda: rank of A array is not equal 2 !");
if(!shape::isCommonVector(X->getShapeInfo(), xLenDim))
throw std::runtime_error("MmulHelper::mmulMxV cuda: X array must be vector !");
const auto M = A->sizeAt(0);
const auto N = A->sizeAt(1);
if(Y != nullptr && !shape::isCommonVector(Y->getShapeInfo(), yLenDim))
throw std::runtime_error("MmulHelper::mmulMxV cuda: Y array must be vector !");
if(X->lengthOf() != N)
throw std::runtime_error("MmulHelper::mmulMxV cuda: X vector has wrong length !");
if(Y != nullptr && Y->lengthOf() != M)
throw std::runtime_error("MmulHelper::mmulMxV cuda: Y array has wrong length !");
if(Y == nullptr)
Y = new NDArray(outOrder, {M}, DataTypeUtils::pickPairwiseResultType(A->dataType(), X->dataType()), A->getContext());
NDArray *pA(const_cast<NDArray*>(A));
if(A->ews() != 1)
pA = pA->dup('f');
const bool transA = pA->ordering() == 'c';
const cublasOperation_t transAblas = transA ? CUBLAS_OP_T : CUBLAS_OP_N;
int lda, lta;
if(transA) { lda = N; lta = M; }
else { lda = M; lta = N; }
const int incx = X->stridesOf()[xLenDim];
const int incy = Y->stridesOf()[yLenDim];
const auto aType = pA->dataType();
const auto xType = X->dataType();
const auto yType = Y->dataType();
auto handle = reinterpret_cast<cublasHandle_t *>(A->getContext()->getCublasHandle());
auto stream = A->getContext()->getCudaStream();
auto status = cublasSetStream_v2(*handle, *stream);
if (status != CUBLAS_STATUS_SUCCESS) throw cuda_exception::build("MmulHelper::mmulMxV cuda failed !", status);
const bool AX(aType == xType), AY(aType == yType), AXY(AX && AY);
NDArray::prepareSpecialUse({Y}, {pA, X});
// choose appropriate cuda gemm api depending on data types
if(AXY && aType == DataType::DOUBLE) {
status = cublasDgemv(*handle, transAblas, lda, lta, &alpha, (double*)pA->getSpecialBuffer(), lda, (double*)X->getSpecialBuffer(), incx, &beta, (double*)Y->getSpecialBuffer(), incy);
}
else if(AXY && aType == DataType::FLOAT32) {
float alphaF(alpha), betaF(beta);
status = cublasSgemv(*handle, transAblas, lda, lta, &alphaF, (float*)pA->getSpecialBuffer(), lda, (float*)X->getSpecialBuffer(), incx, &betaF, (float*)Y->getSpecialBuffer(), incy);
}
else {
dim3 threadsPerBlock(M);
dim3 blocksPerGrid(1);
if (M > 512){
threadsPerBlock.x = 512;
blocksPerGrid.x = math::nd4j_ceil<double, int>(static_cast<double>(M) / threadsPerBlock.x); // rows
}
[WIP] multi-device support (#80) * fix pad javadoc and @see links. (#72) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * [WIP] More fixes (#73) * special tests for ConstantTadHelper/ConstantShapeHelper Signed-off-by: raver119 <raver119@gmail.com> * release methods for data buffers Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary TadPack C++/Java side (#74) Signed-off-by: raver119 <raver119@gmail.com> * Zoo model TF import test updates (#75) * argLine fix, update compression_gru comment * updated comment for xception * undid but commented argLine change * updated xlnet comment * copyright headers * - new NDArray methods like()/ulike() (#77) - fix for depthwise_conv2d_bp + special test Signed-off-by: raver119 <raver119@gmail.com> * upsampling2d fix CUDA Signed-off-by: raver119 <raver119@gmail.com> * DL4J trace logging (#79) * MLN/CG trace logging for debugging Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tiny tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * strided_slice_bp shape fn leak fix Signed-off-by: raver119 <raver119@gmail.com> * SameDiff fixes and naming (#78) * remove SDVariable inplace methods * import methods * npe fix in OpVal * removed SameDiff inplace ops from tests * Naming updates, moved to centralized methods in SameDiff, should use op_#:# for everything * quick fixes * javadoc * SDVariable eval with placeholders * use regex match * better matching * initial commit Signed-off-by: raver119 <raver119@gmail.com> * initial commit Signed-off-by: raver119 <raver119@gmail.com> * fix javadoc. (#76) * fix javadoc. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace most @see with @link s. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * 4 additional tests Signed-off-by: raver119 <raver119@gmail.com> * launch context reorganization Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext reorganization Signed-off-by: raver119 <raver119@gmail.com> * per-device LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * Various DL4J/ND4J fixes (#81) * #7954 Force refresh of UI when switching tabs on overview page Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8017 Concurrent modification exception (synchronize) fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8033 Don't initialize updater in middle of writing memory crash dump Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8208 Fix shape checks for ND4J int[] creator methods Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6385 #7992 Keras import naming fixes + cleanup Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8016 Upsampling3D - add NDHWC format support Signed-off-by: AlexDBlack <blacka101@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Refactor NativeOps.h to export C functions * Actually export functions from NativeOps.h * Adapt the Java wrappers in ND4J generated with JavaCPP * Create C wrappers for some of the C++ classes currently used by ND4J * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * remove duplicate code in createBufferDetached. (#83) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Keras model import - updater lr fix (#84) * Keras model import - updater lr fix Signed-off-by: eraly <susan.eraly@gmail.com> * Keras model import - updater lr fix, cleanup Signed-off-by: eraly <susan.eraly@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Fix functions of OpaqueVariablesSet * thread-local buffers/affinity Signed-off-by: raver119 <raver119@gmail.com> * thread safety for LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * more of thread safety Signed-off-by: raver119 <raver119@gmail.com> * one more multi threaded test Signed-off-by: raver119 <raver119@gmail.com> * SameDiff Convolution Config validation, better output methods (#82) * Conv Config validation & tests Signed-off-by: Ryan Nett <rnett@skymind.io> * stackOutputs utility method Signed-off-by: Ryan Nett <rnett@skymind.io> * use constructor for validation, support negative kernel sizes (infered from weights) Signed-off-by: Ryan Nett <rnett@skymind.io> * better output methods Signed-off-by: Ryan Nett <rnett@skymind.io> * move output to be with fit and evaluate Signed-off-by: Ryan Nett <rnett@skymind.io> * fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * more fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * refactor duplicate code from pad methods. (#86) * refactor duplicate code from pad methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace switch with if. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes and improvements (#87) * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6488 ElementWiseVertex broadcast support Signed-off-by: AlexDBlack <blacka101@gmail.com> * Constructors and broadcast supported it Transforms.max/min Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8054 ElementWiseVertex now supports broadcast inputs Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8057 Nd4j.create overload dtype fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7551 ND4J Shape validation fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Numpy boolean import (#91) * numpy bool type Signed-off-by: raver119 <raver119@gmail.com> * numpy bool java side Signed-off-by: raver119 <raver119@gmail.com> * remove create method with unused parameter. (#89) * remove create method with unused parameter. * removed more unused methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * removing more unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * last removal of unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * remove createSparse methods. (#92) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes (#90) * Deprecate Old*Op instances Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8063 #8054 Broadcast exceptions + cleanup inplace ops Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Remove bad test condition Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7993 Fix shape function issue in crop_and_resize op Signed-off-by: AlexDBlack <blacka101@gmail.com> * DL4J SameDiff lambda layer fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8029 Fix for pnorm backprop math Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8038 Fix Op profiler NaN/Inf triggering + add tests (#93) Signed-off-by: AlexDBlack <blacka101@gmail.com> * createUninitializedDetached refactoring. (#94) * wip * update interface, add null implementations. * Breaking one test in a weird way. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * createUninitializedDetached refactored. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * cuda build fix for issues introduced by recent refactoring Signed-off-by: raver119 <raver119@gmail.com> * [WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io> * build fix Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI (#97) Signed-off-by: raver119 <raver119@gmail.com> * temporary stack fix Signed-off-by: raver119 <raver119@gmail.com> * round robin affinity test Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy CudaContext methods Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy ContextPool classes/methods Signed-off-by: raver119 <raver119@gmail.com> * one legacy test removed Signed-off-by: raver119 <raver119@gmail.com> * few more fields rearranged Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext++ Signed-off-by: raver119 <raver119@gmail.com> * more of OpaqueLaunchContext methods Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext -> CudaContext Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * cusolver handles Signed-off-by: raver119 <raver119@gmail.com> * typo Signed-off-by: raver119 <raver119@gmail.com> * cusolver method Signed-off-by: raver119 <raver119@gmail.com> * cusolver handle propagated Signed-off-by: raver119 <raver119@gmail.com> * blas/solver handles Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * legacy concat implementations replaced with new CustomOp Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * concat now uses way more blocks Signed-off-by: raver119 <raver119@gmail.com> * print Signed-off-by: raver119 <raver119@gmail.com> * no more triple template mmul Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bitonic sort reorganized Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * type conversions moved to generic impl Signed-off-by: raver119 <raver119@gmail.com> * cpu data types pass Signed-off-by: raver119 <raver119@gmail.com> * non_max_suppression Signed-off-by: raver119 <raver119@gmail.com> * sortByValue fix Signed-off-by: raver119 <raver119@gmail.com> * ignore all mixed datatype tests for mmul Signed-off-by: raver119 <raver119@gmail.com> * special handling of OpProfiler exceptions Signed-off-by: raver119 <raver119@gmail.com> * - one failing concat test in cpp - Nd4j.tile now uses op internally Signed-off-by: raver119 <raver119@gmail.com> * get back dtype exception for legacy arrays deserialization Signed-off-by: raver119 <raver119@gmail.com>
2019-08-14 15:52:34 +02:00
//BUILD_TRIPLE_SELECTOR(aType, xType, yType, usualGemv, (blocksPerGrid, threadsPerBlock, stream, transA, M, N, alpha, pA->getSpecialBuffer(), lda, X->getSpecialBuffer(), incx, beta, Y->getSpecialBuffer(), incy), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
BUILD_SINGLE_SELECTOR_THRICE(xType, usualGemv, (blocksPerGrid, threadsPerBlock, stream, transA, M, N, alpha, pA->getSpecialBuffer(), lda, X->getSpecialBuffer(), incx, beta, Y->getSpecialBuffer(), incy), NUMERIC_TYPES)
2019-06-06 14:21:15 +02:00
}
if (status != CUBLAS_STATUS_SUCCESS) throw cuda_exception::build("MmulHelper::mmulMxV cuda failed !", status);
auto cudaResult = cudaStreamSynchronize(*stream);
if (cudaResult != 0) throw cuda_exception::build("MmulHelper::mmulMxV cuda failed !", cudaResult);
NDArray::registerSpecialUse({Y}, {pA, X});
if(pA != A)
delete pA;
return Y;
}
////////////////////////////////////////////////////////////////////////////
// (X * Y) = Z[0]
NDArray* MmulHelper::dot(const NDArray* X, const NDArray* Y, nd4j::NDArray* Z, const double alpha, const double beta) {
int xLenDim(0), yLenDim(0);
if(!shape::isCommonVector(X->getShapeInfo(), xLenDim))
throw std::runtime_error("MmulHelper::dot cuda: X array must be vector !");
if(!shape::isCommonVector(Y->getShapeInfo(), yLenDim))
throw std::runtime_error("MmulHelper::dot cuda: Y array must be vector !");
if(Z != nullptr && !Z->isScalar())
throw std::runtime_error("MmulHelper::dot cuda: Z array must be scalar !");
const auto length = X->lengthOf();
if(Y->lengthOf() != length)
throw std::runtime_error("MmulHelper::dot cuda: lengths of input vectors are different !");
if(Z == nullptr)
Z = new NDArray(DataTypeUtils::pickPairwiseResultType(X->dataType(), Y->dataType()), X->getContext());
const Nd4jLong incx = X->stridesOf()[xLenDim];
const Nd4jLong incy = Y->stridesOf()[yLenDim];
const auto xType = X->dataType();
const auto yType = Y->dataType();
const auto zType = Z->dataType();
if(!X->isActualOnDeviceSide()) X->syncToDevice();
if(!Y->isActualOnDeviceSide()) Y->syncToDevice();
if(!Z->isActualOnDeviceSide()) Z->syncToDevice();
cudaStream_t* stream = X->getContext()->getCudaStream();
dim3 threadsPerBlock(512);
dim3 blocksPerGrid(1);
if (length > 512)
threadsPerBlock.x = math::nd4j_ceil<double, int>(static_cast<double>(length) / 512);
NDArray::prepareSpecialUse({Z}, {X, Y});
[WIP] multi-device support (#80) * fix pad javadoc and @see links. (#72) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * [WIP] More fixes (#73) * special tests for ConstantTadHelper/ConstantShapeHelper Signed-off-by: raver119 <raver119@gmail.com> * release methods for data buffers Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary TadPack C++/Java side (#74) Signed-off-by: raver119 <raver119@gmail.com> * Zoo model TF import test updates (#75) * argLine fix, update compression_gru comment * updated comment for xception * undid but commented argLine change * updated xlnet comment * copyright headers * - new NDArray methods like()/ulike() (#77) - fix for depthwise_conv2d_bp + special test Signed-off-by: raver119 <raver119@gmail.com> * upsampling2d fix CUDA Signed-off-by: raver119 <raver119@gmail.com> * DL4J trace logging (#79) * MLN/CG trace logging for debugging Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tiny tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * strided_slice_bp shape fn leak fix Signed-off-by: raver119 <raver119@gmail.com> * SameDiff fixes and naming (#78) * remove SDVariable inplace methods * import methods * npe fix in OpVal * removed SameDiff inplace ops from tests * Naming updates, moved to centralized methods in SameDiff, should use op_#:# for everything * quick fixes * javadoc * SDVariable eval with placeholders * use regex match * better matching * initial commit Signed-off-by: raver119 <raver119@gmail.com> * initial commit Signed-off-by: raver119 <raver119@gmail.com> * fix javadoc. (#76) * fix javadoc. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace most @see with @link s. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * 4 additional tests Signed-off-by: raver119 <raver119@gmail.com> * launch context reorganization Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext reorganization Signed-off-by: raver119 <raver119@gmail.com> * per-device LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * Various DL4J/ND4J fixes (#81) * #7954 Force refresh of UI when switching tabs on overview page Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8017 Concurrent modification exception (synchronize) fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8033 Don't initialize updater in middle of writing memory crash dump Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8208 Fix shape checks for ND4J int[] creator methods Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6385 #7992 Keras import naming fixes + cleanup Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8016 Upsampling3D - add NDHWC format support Signed-off-by: AlexDBlack <blacka101@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Refactor NativeOps.h to export C functions * Actually export functions from NativeOps.h * Adapt the Java wrappers in ND4J generated with JavaCPP * Create C wrappers for some of the C++ classes currently used by ND4J * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * remove duplicate code in createBufferDetached. (#83) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Keras model import - updater lr fix (#84) * Keras model import - updater lr fix Signed-off-by: eraly <susan.eraly@gmail.com> * Keras model import - updater lr fix, cleanup Signed-off-by: eraly <susan.eraly@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Fix functions of OpaqueVariablesSet * thread-local buffers/affinity Signed-off-by: raver119 <raver119@gmail.com> * thread safety for LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * more of thread safety Signed-off-by: raver119 <raver119@gmail.com> * one more multi threaded test Signed-off-by: raver119 <raver119@gmail.com> * SameDiff Convolution Config validation, better output methods (#82) * Conv Config validation & tests Signed-off-by: Ryan Nett <rnett@skymind.io> * stackOutputs utility method Signed-off-by: Ryan Nett <rnett@skymind.io> * use constructor for validation, support negative kernel sizes (infered from weights) Signed-off-by: Ryan Nett <rnett@skymind.io> * better output methods Signed-off-by: Ryan Nett <rnett@skymind.io> * move output to be with fit and evaluate Signed-off-by: Ryan Nett <rnett@skymind.io> * fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * more fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * refactor duplicate code from pad methods. (#86) * refactor duplicate code from pad methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace switch with if. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes and improvements (#87) * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6488 ElementWiseVertex broadcast support Signed-off-by: AlexDBlack <blacka101@gmail.com> * Constructors and broadcast supported it Transforms.max/min Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8054 ElementWiseVertex now supports broadcast inputs Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8057 Nd4j.create overload dtype fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7551 ND4J Shape validation fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Numpy boolean import (#91) * numpy bool type Signed-off-by: raver119 <raver119@gmail.com> * numpy bool java side Signed-off-by: raver119 <raver119@gmail.com> * remove create method with unused parameter. (#89) * remove create method with unused parameter. * removed more unused methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * removing more unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * last removal of unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * remove createSparse methods. (#92) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes (#90) * Deprecate Old*Op instances Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8063 #8054 Broadcast exceptions + cleanup inplace ops Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Remove bad test condition Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7993 Fix shape function issue in crop_and_resize op Signed-off-by: AlexDBlack <blacka101@gmail.com> * DL4J SameDiff lambda layer fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8029 Fix for pnorm backprop math Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8038 Fix Op profiler NaN/Inf triggering + add tests (#93) Signed-off-by: AlexDBlack <blacka101@gmail.com> * createUninitializedDetached refactoring. (#94) * wip * update interface, add null implementations. * Breaking one test in a weird way. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * createUninitializedDetached refactored. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * cuda build fix for issues introduced by recent refactoring Signed-off-by: raver119 <raver119@gmail.com> * [WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io> * build fix Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI (#97) Signed-off-by: raver119 <raver119@gmail.com> * temporary stack fix Signed-off-by: raver119 <raver119@gmail.com> * round robin affinity test Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy CudaContext methods Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy ContextPool classes/methods Signed-off-by: raver119 <raver119@gmail.com> * one legacy test removed Signed-off-by: raver119 <raver119@gmail.com> * few more fields rearranged Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext++ Signed-off-by: raver119 <raver119@gmail.com> * more of OpaqueLaunchContext methods Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext -> CudaContext Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * cusolver handles Signed-off-by: raver119 <raver119@gmail.com> * typo Signed-off-by: raver119 <raver119@gmail.com> * cusolver method Signed-off-by: raver119 <raver119@gmail.com> * cusolver handle propagated Signed-off-by: raver119 <raver119@gmail.com> * blas/solver handles Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * legacy concat implementations replaced with new CustomOp Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * concat now uses way more blocks Signed-off-by: raver119 <raver119@gmail.com> * print Signed-off-by: raver119 <raver119@gmail.com> * no more triple template mmul Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bitonic sort reorganized Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * type conversions moved to generic impl Signed-off-by: raver119 <raver119@gmail.com> * cpu data types pass Signed-off-by: raver119 <raver119@gmail.com> * non_max_suppression Signed-off-by: raver119 <raver119@gmail.com> * sortByValue fix Signed-off-by: raver119 <raver119@gmail.com> * ignore all mixed datatype tests for mmul Signed-off-by: raver119 <raver119@gmail.com> * special handling of OpProfiler exceptions Signed-off-by: raver119 <raver119@gmail.com> * - one failing concat test in cpp - Nd4j.tile now uses op internally Signed-off-by: raver119 <raver119@gmail.com> * get back dtype exception for legacy arrays deserialization Signed-off-by: raver119 <raver119@gmail.com>
2019-08-14 15:52:34 +02:00
//BUILD_TRIPLE_SELECTOR(xType, yType, zType, usualDot, (blocksPerGrid, threadsPerBlock, stream, length, alpha, X->getSpecialBuffer(), incx, Y->getSpecialBuffer(), incy, beta, Z->getSpecialBuffer()), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
BUILD_SINGLE_SELECTOR_THRICE(xType, usualDot, (blocksPerGrid, threadsPerBlock, stream, length, alpha, X->getSpecialBuffer(), incx, Y->getSpecialBuffer(), incy, beta, Z->getSpecialBuffer()), NUMERIC_TYPES)
2019-06-06 14:21:15 +02:00
auto cudaResult = cudaStreamSynchronize(*stream);
if (cudaResult != 0) throw cuda_exception::build("MmulHelper::dot cuda failed !", cudaResult);
NDArray::registerSpecialUse({Z}, {X, Y});
return Z;
}
[WIP] multi-device support (#80) * fix pad javadoc and @see links. (#72) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * [WIP] More fixes (#73) * special tests for ConstantTadHelper/ConstantShapeHelper Signed-off-by: raver119 <raver119@gmail.com> * release methods for data buffers Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary buffer Java side Signed-off-by: raver119 <raver119@gmail.com> * delete temporary TadPack C++/Java side (#74) Signed-off-by: raver119 <raver119@gmail.com> * Zoo model TF import test updates (#75) * argLine fix, update compression_gru comment * updated comment for xception * undid but commented argLine change * updated xlnet comment * copyright headers * - new NDArray methods like()/ulike() (#77) - fix for depthwise_conv2d_bp + special test Signed-off-by: raver119 <raver119@gmail.com> * upsampling2d fix CUDA Signed-off-by: raver119 <raver119@gmail.com> * DL4J trace logging (#79) * MLN/CG trace logging for debugging Signed-off-by: AlexDBlack <blacka101@gmail.com> * Tiny tweak Signed-off-by: AlexDBlack <blacka101@gmail.com> * strided_slice_bp shape fn leak fix Signed-off-by: raver119 <raver119@gmail.com> * SameDiff fixes and naming (#78) * remove SDVariable inplace methods * import methods * npe fix in OpVal * removed SameDiff inplace ops from tests * Naming updates, moved to centralized methods in SameDiff, should use op_#:# for everything * quick fixes * javadoc * SDVariable eval with placeholders * use regex match * better matching * initial commit Signed-off-by: raver119 <raver119@gmail.com> * initial commit Signed-off-by: raver119 <raver119@gmail.com> * fix javadoc. (#76) * fix javadoc. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace most @see with @link s. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * 4 additional tests Signed-off-by: raver119 <raver119@gmail.com> * launch context reorganization Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext reorganization Signed-off-by: raver119 <raver119@gmail.com> * per-device LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * Various DL4J/ND4J fixes (#81) * #7954 Force refresh of UI when switching tabs on overview page Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8017 Concurrent modification exception (synchronize) fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8033 Don't initialize updater in middle of writing memory crash dump Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8208 Fix shape checks for ND4J int[] creator methods Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6385 #7992 Keras import naming fixes + cleanup Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8016 Upsampling3D - add NDHWC format support Signed-off-by: AlexDBlack <blacka101@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Refactor NativeOps.h to export C functions * Actually export functions from NativeOps.h * Adapt the Java wrappers in ND4J generated with JavaCPP * Create C wrappers for some of the C++ classes currently used by ND4J * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * remove duplicate code in createBufferDetached. (#83) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Keras model import - updater lr fix (#84) * Keras model import - updater lr fix Signed-off-by: eraly <susan.eraly@gmail.com> * Keras model import - updater lr fix, cleanup Signed-off-by: eraly <susan.eraly@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * ContextBuffers as separate entity Signed-off-by: raver119 <raver119@gmail.com> * Fix functions of OpaqueVariablesSet * thread-local buffers/affinity Signed-off-by: raver119 <raver119@gmail.com> * thread safety for LaunchContext Signed-off-by: raver119 <raver119@gmail.com> * more of thread safety Signed-off-by: raver119 <raver119@gmail.com> * one more multi threaded test Signed-off-by: raver119 <raver119@gmail.com> * SameDiff Convolution Config validation, better output methods (#82) * Conv Config validation & tests Signed-off-by: Ryan Nett <rnett@skymind.io> * stackOutputs utility method Signed-off-by: Ryan Nett <rnett@skymind.io> * use constructor for validation, support negative kernel sizes (infered from weights) Signed-off-by: Ryan Nett <rnett@skymind.io> * better output methods Signed-off-by: Ryan Nett <rnett@skymind.io> * move output to be with fit and evaluate Signed-off-by: Ryan Nett <rnett@skymind.io> * fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * more fixes Signed-off-by: Ryan Nett <rnett@skymind.io> * refactor duplicate code from pad methods. (#86) * refactor duplicate code from pad methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * replace switch with if. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes and improvements (#87) * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * Reshape and reallocate - small fixes Signed-off-by: AlexDBlack <blacka101@gmail.com> * #6488 ElementWiseVertex broadcast support Signed-off-by: AlexDBlack <blacka101@gmail.com> * Constructors and broadcast supported it Transforms.max/min Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8054 ElementWiseVertex now supports broadcast inputs Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8057 Nd4j.create overload dtype fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7551 ND4J Shape validation fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Numpy boolean import (#91) * numpy bool type Signed-off-by: raver119 <raver119@gmail.com> * numpy bool java side Signed-off-by: raver119 <raver119@gmail.com> * remove create method with unused parameter. (#89) * remove create method with unused parameter. * removed more unused methods. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * removing more unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * last removal of unused code. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * remove createSparse methods. (#92) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Various ND4J/DL4J fixes (#90) * Deprecate Old*Op instances Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8063 #8054 Broadcast exceptions + cleanup inplace ops Signed-off-by: AlexDBlack <blacka101@gmail.com> * Small fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * Remove bad test condition Signed-off-by: AlexDBlack <blacka101@gmail.com> * #7993 Fix shape function issue in crop_and_resize op Signed-off-by: AlexDBlack <blacka101@gmail.com> * DL4J SameDiff lambda layer fix Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8029 Fix for pnorm backprop math Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8038 Fix Op profiler NaN/Inf triggering + add tests (#93) Signed-off-by: AlexDBlack <blacka101@gmail.com> * createUninitializedDetached refactoring. (#94) * wip * update interface, add null implementations. * Breaking one test in a weird way. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * createUninitializedDetached refactored. Signed-off-by: Robert Altena <Rob@Ra-ai.com> * cuda build fix for issues introduced by recent refactoring Signed-off-by: raver119 <raver119@gmail.com> * [WIP] More of CUDA (#95) * initial commit Signed-off-by: raver119 <raver119@gmail.com> * Implementation of hashcode cuda helper. Working edition. * Fixed parallel test input arangements. * Fixed tests for hashcode op. * Fixed shape calculation for image:crop_and_resize op and test. * NativeOps tests. Initial test suite. * Added tests for indexReduce methods. * Added test on execBroadcast with NDArray as dimensions. * Added test on execBroadcastBool with NDArray as dimensions. * Added tests on execPairwiseTransform and execPairwiseTransofrmBool. * Added tests for execReduce with scalar results. * Added reduce tests for non-empty dims array. * Added tests for reduce3. * Added tests for execScalar. * Added tests for execSummaryStats. * - provide cpu/cuda code for batch_to_space - testing it Signed-off-by: Yurii <yurii@skymind.io> * - remove old test for batch_to_space (had wrong format and numbers were not checked) Signed-off-by: Yurii <yurii@skymind.io> * Fixed complilation errors with test. * Added test for execTransformFloat. * Added test for execTransformSame. * Added test for execTransformBool. * Added test for execTransformStrict. * Added tests for execScalar/execScalarBool with TADs. * Added test for flatten. * - provide cpu/cuda code for space_to_Batch operaion Signed-off-by: Yurii <yurii@skymind.io> * Added test for concat. * comment unnecessary stuff in s_t_b Signed-off-by: Yurii <yurii@skymind.io> * Added test for specialConcat. * Added tests for memcpy/set routines. * Fixed pullRow cuda test. * Added pullRow test. * Added average test. * - correct typo in NDArray::applyPairwiseTransform(nd4j::pairwise::BoolOps op...) Signed-off-by: Yurii <yurii@skymind.io> * - debugging and fixing cuda tests in JavaInteropTests file Signed-off-by: Yurii <yurii@skymind.io> * - correct some tests Signed-off-by: Yurii <yurii@skymind.io> * Added test for shuffle. * Fixed ops declarations. * Restored omp and added shuffle test. * Added convertTypes test. * Added tests for execRandom. Eliminated usage of RandomBuffer with NativeOps. * Added sort tests. * Added tests for execCustomOp. * - further debuging and fixing tests terminated with crash Signed-off-by: Yurii <yurii@skymind.io> * Added tests for calculateOutputShapes. * Addded Benchmarks test. * Commented benchmark tests. * change assertion Signed-off-by: raver119 <raver119@gmail.com> * Added tests for apply_sgd op. Added cpu helper for that op. * Implement cuda helper for aplly_sgd op. Fixed tests for NativeOps. * Added test for assign broadcastable. * Added tests for assign_bp op. * Added tests for axpy op. * - assign/execScalar/execTransformAny signature change - minor test fix Signed-off-by: raver119 <raver119@gmail.com> * Fixed axpy op. * meh Signed-off-by: raver119 <raver119@gmail.com> * - fix tests for nativeOps::concat Signed-off-by: Yurii <yurii@skymind.io> * sequential transform/scalar Signed-off-by: raver119 <raver119@gmail.com> * allow nested parallelism Signed-off-by: raver119 <raver119@gmail.com> * assign_bp leak fix Signed-off-by: raver119 <raver119@gmail.com> * block setRNG fix Signed-off-by: raver119 <raver119@gmail.com> * enable parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * enable nested parallelism by default Signed-off-by: raver119 <raver119@gmail.com> * Added cuda implementation for row_count helper. * Added implementation for tnse gains op helper. * - take into account possible situations when input arrays are empty in reduce_ cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implemented tsne/edge_forces op cuda-based helper. Parallelized cpu-based helper for edge_forces. * Added kernel for tsne/symmetrized op heleper. * Implementation of tsne/symmetrized op cuda helper. Working edition. * Eliminated waste printfs. * Added test for broadcastgradientargs op. * host-only fallback for empty reduce float Signed-off-by: raver119 <raver119@gmail.com> * - some tests fixes Signed-off-by: Yurii <yurii@skymind.io> * - correct the rest of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * - further correction of reduce_ stuff Signed-off-by: Yurii <yurii@skymind.io> * Added test for Cbow op. Also added cuda implementation for cbow helpers. * - improve code of stack operation for scalar case Signed-off-by: Yurii <yurii@skymind.io> * - provide cuda kernel for gatherND operation Signed-off-by: Yurii <yurii@skymind.io> * Implementation of cbow helpers with cuda kernels. * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * minor tests tweaks Signed-off-by: raver119 <raver119@gmail.com> * - further correction of cuda stuff Signed-off-by: Yurii <yurii@skymind.io> * Implementatation of cbow op helper with cuda kernels. Working edition. * Skip random testing for cudablas case. * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for ELU and ELU_BP ops. * Added tests for eq_scalar, gt_scalar, gte_scalar and lte_scalar ops. * Added tests for neq_scalar. * Added test for noop. * - further work on clipbynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * - get rid of concat op call, use instead direct concat helper call Signed-off-by: Yurii <yurii@skymind.io> * lstmBlockCell context fix Signed-off-by: raver119 <raver119@gmail.com> * Added tests for lrelu and lrelu_bp. * Added tests for selu and selu_bp. * Fixed lrelu derivative helpers. * - some corrections in lstm Signed-off-by: Yurii <yurii@skymind.io> * operator * result shape fix Signed-off-by: raver119 <raver119@gmail.com> * - correct typo in lstmCell Signed-off-by: Yurii <yurii@skymind.io> * few tests fixed Signed-off-by: raver119 <raver119@gmail.com> * CUDA inverse broadcast bool fix Signed-off-by: raver119 <raver119@gmail.com> * disable MMAP test for CUDA Signed-off-by: raver119 <raver119@gmail.com> * BooleanOp syncToDevice Signed-off-by: raver119 <raver119@gmail.com> * meh Signed-off-by: raver119 <raver119@gmail.com> * additional data types for im2col/col2im Signed-off-by: raver119 <raver119@gmail.com> * Added test for firas_sparse op. * one more RandomBuffer test excluded Signed-off-by: raver119 <raver119@gmail.com> * Added tests for flatten op. * Added test for Floor op. * bunch of tests fixed Signed-off-by: raver119 <raver119@gmail.com> * mmulDot tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Implemented floordiv_bp op and tests. * Fixed scalar case with cuda implementation for bds. * - work on cuda kernel for clip_by_norm backprop op is completed Signed-off-by: Yurii <yurii@skymind.io> * Eliminate cbow crach. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Eliminated abortion with batched nlp test. * more tests fixed Signed-off-by: raver119 <raver119@gmail.com> * Fixed shared flag initializing. * disabled bunch of cpu workspaces tests Signed-off-by: raver119 <raver119@gmail.com> * scalar operators fix: missing registerSpecialUse call Signed-off-by: raver119 <raver119@gmail.com> * Fixed logdet for cuda and tests. * - correct clipBynorm_bp Signed-off-by: Yurii <yurii@skymind.io> * Fixed crop_and_resize shape datatype. * - correct some mmul tests Signed-off-by: Yurii <yurii@skymind.io> * build fix Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI Signed-off-by: raver119 <raver119@gmail.com> * exclude two methods for JNI (#97) Signed-off-by: raver119 <raver119@gmail.com> * temporary stack fix Signed-off-by: raver119 <raver119@gmail.com> * round robin affinity test Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy CudaContext methods Signed-off-by: raver119 <raver119@gmail.com> * get rid of legacy ContextPool classes/methods Signed-off-by: raver119 <raver119@gmail.com> * one legacy test removed Signed-off-by: raver119 <raver119@gmail.com> * few more fields rearranged Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext Signed-off-by: raver119 <raver119@gmail.com> * OpaqueLaunchContext++ Signed-off-by: raver119 <raver119@gmail.com> * more of OpaqueLaunchContext methods Signed-off-by: raver119 <raver119@gmail.com> * LaunchContext -> CudaContext Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * AffinityManger changes Signed-off-by: raver119 <raver119@gmail.com> * cusolver handles Signed-off-by: raver119 <raver119@gmail.com> * typo Signed-off-by: raver119 <raver119@gmail.com> * cusolver method Signed-off-by: raver119 <raver119@gmail.com> * cusolver handle propagated Signed-off-by: raver119 <raver119@gmail.com> * blas/solver handles Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * legacy concat implementations replaced with new CustomOp Signed-off-by: raver119 <raver119@gmail.com> * one more test Signed-off-by: raver119 <raver119@gmail.com> * concat now uses way more blocks Signed-off-by: raver119 <raver119@gmail.com> * print Signed-off-by: raver119 <raver119@gmail.com> * no more triple template mmul Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bunch of kernels have dtypes reconsidered Signed-off-by: raver119 <raver119@gmail.com> * bitonic sort reorganized Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * bunch of cpu stuff removed from cuda scope Signed-off-by: raver119 <raver119@gmail.com> * type conversions moved to generic impl Signed-off-by: raver119 <raver119@gmail.com> * cpu data types pass Signed-off-by: raver119 <raver119@gmail.com> * non_max_suppression Signed-off-by: raver119 <raver119@gmail.com> * sortByValue fix Signed-off-by: raver119 <raver119@gmail.com> * ignore all mixed datatype tests for mmul Signed-off-by: raver119 <raver119@gmail.com> * special handling of OpProfiler exceptions Signed-off-by: raver119 <raver119@gmail.com> * - one failing concat test in cpp - Nd4j.tile now uses op internally Signed-off-by: raver119 <raver119@gmail.com> * get back dtype exception for legacy arrays deserialization Signed-off-by: raver119 <raver119@gmail.com>
2019-08-14 15:52:34 +02:00
//BUILD_TRIPLE_TEMPLATE(template void usualGemm, (const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const bool transA, const bool transB, const int M, const int N, const int K, const double alpha, const void* vA, const int lda, const void* vB, const int ldb, const double beta, void* vC, const int ldc), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
//BUILD_TRIPLE_TEMPLATE(template void usualGemv, (const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const bool transA, const int M, const int N, const double alpha, const void* vA, const int lda, const void* vB, const int incx, const double beta, void* vC, const int incy), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
//BUILD_TRIPLE_TEMPLATE(template void usualDot, (const dim3 &blocksPerGrid, const dim3 &threadsPerBlock, cudaStream_t *stream, const Nd4jLong length, const double alpha, const void* vX, const Nd4jLong incx, const void* vY, const Nd4jLong incy, const double beta, void* vZ), NUMERIC_TYPES, NUMERIC_TYPES, FLOAT_TYPES);
2019-06-06 14:21:15 +02:00
}