cavis/libnd4j/include/loops/cpu/reduce/reduce_bool.cpp

241 lines
10 KiB
C++
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 <types/types.h>
#include <system/op_boilerplate.h>
2019-06-06 14:21:15 +02:00
#include <loops/reduce_bool.h>
#include <loops/legacy_ops.h>
#include <helpers/OmpLaunchHelper.h>
2019-06-06 14:21:15 +02:00
#include <helpers/Loops.h>
#include <helpers/ConstantTadHelper.h>
using namespace simdOps;
namespace functions {
namespace reduce {
template <typename X, typename Z>
template <typename OpType>
void _CUDA_H ReduceBoolFunction<X,Z>::execScalar(void *vx,
Nd4jLong *xShapeInfo,
void *vextraParams,
void *vz,
Nd4jLong *zShapeInfo) {
auto x = reinterpret_cast<X *>(vx);
auto z = reinterpret_cast<Z *>(vz);
auto extraParams = reinterpret_cast<X *>(vextraParams);
const Nd4jLong length = shape::length(xShapeInfo);
auto xEws = shape::elementWiseStride(xShapeInfo);
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
if (shape::isEmpty(xShapeInfo)) {
z[0] = OpType::startingValue(x);
return;
}
if(sd::ArrayOptions::arrayType(xShapeInfo) == sd::ArrayType::EMPTY) {
if(sd::ArrayOptions::arrayType(zShapeInfo) == sd::ArrayType::EMPTY)
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
return;
const auto startingVal = OpType::startingValue(x);
for (Nd4jLong i = 0; i < length; i++)
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
z[i] = startingVal;
return;
}
2019-06-06 14:21:15 +02:00
if (xEws >= 1) {
z[0] = execScalar<OpType>(x, xEws, length, extraParams);
}
else {
auto startingValue = OpType::startingValue(x);
2019-06-06 14:21:15 +02:00
uint xShapeInfoCast[MAX_RANK];
const bool canCastX = sd::DataTypeUtils::castShapeInfo(xShapeInfo, xShapeInfoCast);
2019-06-06 14:21:15 +02:00
for (Nd4jLong i = 0; i < length; i++)
startingValue = OpType::update(startingValue, OpType::op(x[shape::indexOffset(i, xShapeInfo, xShapeInfoCast, canCastX)], extraParams), extraParams);
2019-06-06 14:21:15 +02:00
z[0] = OpType::postProcess(startingValue, length, extraParams);
2019-06-06 14:21:15 +02:00
}
}
template <typename X, typename Z>
template <typename OpType>
Z _CUDA_H ReduceBoolFunction<X, Z>::execScalar(void *vx, Nd4jLong *xShapeInfo, void *vextraParams) {
auto x = reinterpret_cast<X *>(vx);
auto extraParams = reinterpret_cast<X *>(vextraParams);
const Nd4jLong length = shape::length(xShapeInfo);
auto xEws = shape::elementWiseStride(xShapeInfo);
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
2019-06-06 14:21:15 +02:00
if (xEws >= 1) {
return execScalar<OpType>(x, xEws, length, extraParams);
}
else {
auto startingValue = OpType::startingValue(x);
2019-06-06 14:21:15 +02:00
uint xShapeInfoCast[MAX_RANK];
bool canCastX = sd::DataTypeUtils::castShapeInfo(xShapeInfo, xShapeInfoCast);
2019-06-06 14:21:15 +02:00
for (Nd4jLong i = 0; i < length; i++)
startingValue = OpType::update(startingValue, OpType::op(x[shape::indexOffset(i, xShapeInfo, xShapeInfoCast, canCastX)], extraParams), extraParams);
2019-06-06 14:21:15 +02:00
return OpType::postProcess(startingValue, length, extraParams);
2019-06-06 14:21:15 +02:00
}
}
template <typename X, typename Y>
Y ReduceBoolFunction<X, Y>::execScalar(const int opNum,
void *x,
Nd4jLong *xShapeInfo,
void *extraParams) {
RETURNING_DISPATCH_BY_OPNUM_TT(execScalar, PARAMS(x, xShapeInfo, extraParams), REDUCE_BOOL_OPS);
}
template <typename X, typename Y>
void ReduceBoolFunction<X, Y>::execScalar(const int opNum,
void *x,
Nd4jLong *xShapeInfo,
void *extraParams,
void *z,
Nd4jLong *zShapeInfo) {
DISPATCH_BY_OPNUM_TT(execScalar, PARAMS(x, xShapeInfo, extraParams, z, zShapeInfo), REDUCE_BOOL_OPS);
}
template <typename X, typename Y>
void ReduceBoolFunction<X, Y>::exec(const int opNum,
void *x,
Nd4jLong *xShapeInfo,
void *extraParams,
void *z,
Nd4jLong *zShapeInfo,
int *dimension,
int dimensionLength,
Nd4jLong *tadShapeInfo,
Nd4jLong *tadOffset, int64_t start, int64_t stop) {
DISPATCH_BY_OPNUM_TT(exec, PARAMS(x, xShapeInfo, extraParams, z, zShapeInfo, dimension, dimensionLength, tadShapeInfo, tadOffset, start, stop), REDUCE_BOOL_OPS);
2019-06-06 14:21:15 +02:00
}
template <typename X, typename Z>
template <typename OpType>
void _CUDA_H ReduceBoolFunction<X,Z>::exec(void *vx,
Nd4jLong *xShapeInfo,
void *vextraParams,
void *vresult,
Nd4jLong *zShapeInfo,
int *dimension,
int dimensionLength,
Nd4jLong *tadShapeInfo,
Nd4jLong *tadOffset, int64_t start, int64_t stop) {
2019-06-06 14:21:15 +02:00
auto x = reinterpret_cast<X *>(vx);
auto z = reinterpret_cast<Z *>(vresult);
auto extraParams = reinterpret_cast<X *>(vextraParams);
auto resultLength = shape::length(zShapeInfo);
if(sd::ArrayOptions::arrayType(xShapeInfo) == sd::ArrayType::EMPTY) {
if(sd::ArrayOptions::arrayType(zShapeInfo) == sd::ArrayType::EMPTY)
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
return;
const auto startingVal = OpType::startingValue(x);
for (Nd4jLong i = 0; i < resultLength; i++)
Dev branch merge: dev_20190606 (#7904) * 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
2019-06-15 13:34:34 +02:00
z[i] = startingVal;
return;
}
2019-06-06 14:21:15 +02:00
//pre squeezed: this is for keeping the pointer to the original
//shape information for tad offset
//the squeezed information doesn't render the right strides for
//tad offset
// || tad.wholeThing
if (resultLength == 1 || dimension == nullptr || dimensionLength == shape::rank(xShapeInfo)) {
z[0] = execScalar<OpType>(x, xShapeInfo, extraParams);
return;
}
auto tadOnlyShapeInfo = tadShapeInfo;
auto tadOffsets = tadOffset;
if (tadOnlyShapeInfo == nullptr || tadOffsets == nullptr) {
if (dimensionLength < 1)
return;
auto tadPack = sd::ConstantTadHelper::getInstance()->tadForDimensions(xShapeInfo, dimension, dimensionLength);
2019-06-06 14:21:15 +02:00
tadOnlyShapeInfo = tadPack.primaryShapeInfo();
tadOffsets = tadPack.primaryOffsets();
}
#ifdef INLINE_LOOPS
sd::ReductionLoops<X,Z,X>::template loopReduce<OpType>(x, xShapeInfo, z, zShapeInfo, tadOnlyShapeInfo, tadOffsets, extraParams, start, stop);
2019-06-06 14:21:15 +02:00
#else
sd::ReductionBoolLoops<X,Z>::template innerloopReduce<OpType>(x, xShapeInfo, z, zShapeInfo, tadOnlyShapeInfo, tadOffsets, extraParams, start, stop);
2019-06-06 14:21:15 +02:00
#endif
}
template <typename X, typename Z>
template<typename OpType>
void _CUDA_H ReduceBoolFunction<X,Z>::exec(void *x,
Nd4jLong *xShapeInfo,
void *extraParams,
void *vresult,
Nd4jLong *resultShapeInfo) {
// FIXME: wtf???
auto z = reinterpret_cast<Z*>(vresult);
z[0] = execScalar<OpType>(x, xShapeInfo, extraParams);
}
template <typename X, typename Z>
template <typename OpType>
Z _CUDA_H ReduceBoolFunction<X, Z>::execScalar(void *vx, Nd4jLong xEws, Nd4jLong length, void *vextraParams) {
auto x = reinterpret_cast<X *>(vx);
auto extraParams = reinterpret_cast<X *>(vextraParams);
int maxThreads = sd::math::nd4j_min<int>(64, sd::Environment::getInstance()->maxThreads());
Z intermediate[64];
PRAGMA_OMP_SIMD
for (auto e = 0; e < maxThreads; e++)
intermediate[e] = OpType::startingValue(x);
auto func = PRAGMA_THREADS_FOR {
if (xEws == 1) {
for (auto i = start; i < stop; i++)
intermediate[thread_id] = OpType::update(intermediate[thread_id], OpType::op(x[i], extraParams), extraParams);
} else {
for (auto i = start; i < stop; i++)
intermediate[thread_id] = OpType::update(intermediate[thread_id], OpType::op(x[i * xEws], extraParams), extraParams);
2019-06-06 14:21:15 +02:00
}
};
2019-06-06 14:21:15 +02:00
maxThreads = samediff::Threads::parallel_for(func, 0, length, 1, maxThreads);
2019-06-06 14:21:15 +02:00
// merge results
for (int e = 1; e < maxThreads; e++)
intermediate[0] = OpType::update(intermediate[0], intermediate[e], extraParams);
2019-06-06 14:21:15 +02:00
// return result
return OpType::postProcess(intermediate[0], length, extraParams);
2019-06-06 14:21:15 +02:00
}
BUILD_DOUBLE_TEMPLATE(template class ND4J_EXPORT ReduceBoolFunction, , LIBND4J_TYPES, BOOL_TYPES);
}
}