cavis/libnd4j/include/helpers/cpu/ConstantShapeHelper.cpp

207 lines
7.4 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
//
#ifndef __CUDABLAS__
#include <helpers/ConstantShapeHelper.h>
#include <helpers/logger.h>
#include <helpers/ShapeBuilders.h>
#include <helpers/ShapeUtils.h>
2019-06-06 14:21:15 +02:00
namespace sd {
2019-06-06 14:21:15 +02:00
ConstantShapeHelper::ConstantShapeHelper() {
_cache.resize(32);
for (int e = 0; e < 32; e++) {
MAP_IMPL<ShapeDescriptor, ConstantDataBuffer> cache;
2019-06-06 14:21:15 +02:00
_cache[e] = cache;
}
}
ConstantShapeHelper* ConstantShapeHelper::getInstance() {
if (!_INSTANCE)
_INSTANCE = new ConstantShapeHelper();
return _INSTANCE;
}
ConstantDataBuffer ConstantShapeHelper::bufferForShapeInfo(sd::DataType dataType, char order, const std::vector<Nd4jLong> &shape) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(dataType, order, shape);
return bufferForShapeInfo(descriptor);
}
ConstantDataBuffer ConstantShapeHelper::bufferForShapeInfo(const sd::DataType dataType, const char order, const int rank, const Nd4jLong* shape) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(dataType, order, shape, rank);
return bufferForShapeInfo(descriptor);
}
Snapshot update (#8194) * fix double consumption of rng on cpu Signed-off-by: raver119 <raver119@gmail.com> * Shyrma docs (#222) * - documenting and profiling matrix_set_diag cuda kernel Signed-off-by: Yurii <yurii@skymind.io> * - correct formula of pnorm pooling in cuda 2d/3d kernels - remove helper matrix_diag which duplicates work of helper matrix_set_diag Signed-off-by: Yurii <yurii@skymind.io> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * Documentation from serialization/deserialization in NLP (#221) * refactoring Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadocs Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadoc fixed Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Cleanup Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * dedicated lock for getCudaCublasHandle Signed-off-by: raver119 <raver119@gmail.com> * Small fixes (#223) Signed-off-by: AlexDBlack <blacka101@gmail.com> * ELU DL4J fixes (#224) Signed-off-by: AlexDBlack <blacka101@gmail.com> * javadoc (#225) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Small test compilation fix (#226) Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8182 remove spark version suffix (#227) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Thread safety (#229) * sync after cublas*gemm Signed-off-by: raver119 <raver119@gmail.com> * mutex for CublasHelper Signed-off-by: raver119 <raver119@gmail.com> * don't store cublasHandle in LaunchContext, it's per-device anyway Signed-off-by: raver119 <raver119@gmail.com> * some printout Signed-off-by: raver119 <raver119@gmail.com> * check for field instead Signed-off-by: raver119 <raver119@gmail.com> * pew-pew Signed-off-by: raver119 <raver119@gmail.com> * don't release ContextBuffers until device changed Signed-off-by: raver119 <raver119@gmail.com> * small tweak Signed-off-by: raver119 <raver119@gmail.com> * some logging in sgemm Signed-off-by: raver119 <raver119@gmail.com> * stream sync Signed-off-by: raver119 <raver119@gmail.com> * some more logging Signed-off-by: raver119 <raver119@gmail.com> * some more error checks Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * minor AffinityManager fix Signed-off-by: raver119 <raver119@gmail.com> * cudaEvent error logging improvement Signed-off-by: raver119 <raver119@gmail.com> * ConstantHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * - minor corrections in ConstantTadHelper Signed-off-by: Yurii <yurii@skymind.io> * ConstantShapeHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * ConstantTadHelper.cu updated Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com>
2019-09-03 21:02:02 +02:00
ConstantDataBuffer ConstantShapeHelper::bufferForShapeInfo(const ShapeDescriptor &descriptor) {
2019-06-06 14:21:15 +02:00
int deviceId = 0;
_mutex.lock();
if (_cache[deviceId].count(descriptor) == 0) {
auto hPtr = descriptor.toShapeInfo();
ConstantDataBuffer buffer(hPtr, nullptr, shape::shapeInfoLength(hPtr)*sizeof(Nd4jLong), DataType::INT64);
ShapeDescriptor descriptor1(descriptor);
_cache[deviceId][descriptor1] = buffer;
Snapshot update (#8194) * fix double consumption of rng on cpu Signed-off-by: raver119 <raver119@gmail.com> * Shyrma docs (#222) * - documenting and profiling matrix_set_diag cuda kernel Signed-off-by: Yurii <yurii@skymind.io> * - correct formula of pnorm pooling in cuda 2d/3d kernels - remove helper matrix_diag which duplicates work of helper matrix_set_diag Signed-off-by: Yurii <yurii@skymind.io> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * Documentation from serialization/deserialization in NLP (#221) * refactoring Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadocs Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadoc fixed Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Cleanup Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * dedicated lock for getCudaCublasHandle Signed-off-by: raver119 <raver119@gmail.com> * Small fixes (#223) Signed-off-by: AlexDBlack <blacka101@gmail.com> * ELU DL4J fixes (#224) Signed-off-by: AlexDBlack <blacka101@gmail.com> * javadoc (#225) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Small test compilation fix (#226) Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8182 remove spark version suffix (#227) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Thread safety (#229) * sync after cublas*gemm Signed-off-by: raver119 <raver119@gmail.com> * mutex for CublasHelper Signed-off-by: raver119 <raver119@gmail.com> * don't store cublasHandle in LaunchContext, it's per-device anyway Signed-off-by: raver119 <raver119@gmail.com> * some printout Signed-off-by: raver119 <raver119@gmail.com> * check for field instead Signed-off-by: raver119 <raver119@gmail.com> * pew-pew Signed-off-by: raver119 <raver119@gmail.com> * don't release ContextBuffers until device changed Signed-off-by: raver119 <raver119@gmail.com> * small tweak Signed-off-by: raver119 <raver119@gmail.com> * some logging in sgemm Signed-off-by: raver119 <raver119@gmail.com> * stream sync Signed-off-by: raver119 <raver119@gmail.com> * some more logging Signed-off-by: raver119 <raver119@gmail.com> * some more error checks Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * minor AffinityManager fix Signed-off-by: raver119 <raver119@gmail.com> * cudaEvent error logging improvement Signed-off-by: raver119 <raver119@gmail.com> * ConstantHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * - minor corrections in ConstantTadHelper Signed-off-by: Yurii <yurii@skymind.io> * ConstantShapeHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * ConstantTadHelper.cu updated Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com>
2019-09-03 21:02:02 +02:00
auto r = _cache[deviceId][descriptor1];
2019-06-06 14:21:15 +02:00
_mutex.unlock();
return r;
} else {
Snapshot update (#8194) * fix double consumption of rng on cpu Signed-off-by: raver119 <raver119@gmail.com> * Shyrma docs (#222) * - documenting and profiling matrix_set_diag cuda kernel Signed-off-by: Yurii <yurii@skymind.io> * - correct formula of pnorm pooling in cuda 2d/3d kernels - remove helper matrix_diag which duplicates work of helper matrix_set_diag Signed-off-by: Yurii <yurii@skymind.io> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * Documentation from serialization/deserialization in NLP (#221) * refactoring Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadocs Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadoc fixed Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Cleanup Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * dedicated lock for getCudaCublasHandle Signed-off-by: raver119 <raver119@gmail.com> * Small fixes (#223) Signed-off-by: AlexDBlack <blacka101@gmail.com> * ELU DL4J fixes (#224) Signed-off-by: AlexDBlack <blacka101@gmail.com> * javadoc (#225) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Small test compilation fix (#226) Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8182 remove spark version suffix (#227) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Thread safety (#229) * sync after cublas*gemm Signed-off-by: raver119 <raver119@gmail.com> * mutex for CublasHelper Signed-off-by: raver119 <raver119@gmail.com> * don't store cublasHandle in LaunchContext, it's per-device anyway Signed-off-by: raver119 <raver119@gmail.com> * some printout Signed-off-by: raver119 <raver119@gmail.com> * check for field instead Signed-off-by: raver119 <raver119@gmail.com> * pew-pew Signed-off-by: raver119 <raver119@gmail.com> * don't release ContextBuffers until device changed Signed-off-by: raver119 <raver119@gmail.com> * small tweak Signed-off-by: raver119 <raver119@gmail.com> * some logging in sgemm Signed-off-by: raver119 <raver119@gmail.com> * stream sync Signed-off-by: raver119 <raver119@gmail.com> * some more logging Signed-off-by: raver119 <raver119@gmail.com> * some more error checks Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * minor AffinityManager fix Signed-off-by: raver119 <raver119@gmail.com> * cudaEvent error logging improvement Signed-off-by: raver119 <raver119@gmail.com> * ConstantHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * - minor corrections in ConstantTadHelper Signed-off-by: Yurii <yurii@skymind.io> * ConstantShapeHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * ConstantTadHelper.cu updated Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com>
2019-09-03 21:02:02 +02:00
auto r = _cache[deviceId].at(descriptor);
2019-06-06 14:21:15 +02:00
_mutex.unlock();
return r;
}
}
Snapshot update (#8194) * fix double consumption of rng on cpu Signed-off-by: raver119 <raver119@gmail.com> * Shyrma docs (#222) * - documenting and profiling matrix_set_diag cuda kernel Signed-off-by: Yurii <yurii@skymind.io> * - correct formula of pnorm pooling in cuda 2d/3d kernels - remove helper matrix_diag which duplicates work of helper matrix_set_diag Signed-off-by: Yurii <yurii@skymind.io> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * cublasHandle sharing + lock Signed-off-by: raver119 <raver119@gmail.com> * Documentation from serialization/deserialization in NLP (#221) * refactoring Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadocs Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Javadoc fixed Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * Cleanup Signed-off-by: Alexander Stoyakin <alexander.stoyakin@gmail.com> * dedicated lock for getCudaCublasHandle Signed-off-by: raver119 <raver119@gmail.com> * Small fixes (#223) Signed-off-by: AlexDBlack <blacka101@gmail.com> * ELU DL4J fixes (#224) Signed-off-by: AlexDBlack <blacka101@gmail.com> * javadoc (#225) Signed-off-by: Robert Altena <Rob@Ra-ai.com> * Small test compilation fix (#226) Signed-off-by: AlexDBlack <blacka101@gmail.com> * #8182 remove spark version suffix (#227) Signed-off-by: AlexDBlack <blacka101@gmail.com> * [WIP] Thread safety (#229) * sync after cublas*gemm Signed-off-by: raver119 <raver119@gmail.com> * mutex for CublasHelper Signed-off-by: raver119 <raver119@gmail.com> * don't store cublasHandle in LaunchContext, it's per-device anyway Signed-off-by: raver119 <raver119@gmail.com> * some printout Signed-off-by: raver119 <raver119@gmail.com> * check for field instead Signed-off-by: raver119 <raver119@gmail.com> * pew-pew Signed-off-by: raver119 <raver119@gmail.com> * don't release ContextBuffers until device changed Signed-off-by: raver119 <raver119@gmail.com> * small tweak Signed-off-by: raver119 <raver119@gmail.com> * some logging in sgemm Signed-off-by: raver119 <raver119@gmail.com> * stream sync Signed-off-by: raver119 <raver119@gmail.com> * some more logging Signed-off-by: raver119 <raver119@gmail.com> * some more error checks Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * one fancy test Signed-off-by: raver119 <raver119@gmail.com> * minor AffinityManager fix Signed-off-by: raver119 <raver119@gmail.com> * cudaEvent error logging improvement Signed-off-by: raver119 <raver119@gmail.com> * ConstantHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * - minor corrections in ConstantTadHelper Signed-off-by: Yurii <yurii@skymind.io> * ConstantShapeHelper thread safety Signed-off-by: raver119 <raver119@gmail.com> * ConstantTadHelper.cu updated Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com> * logging off Signed-off-by: raver119 <raver119@gmail.com>
2019-09-03 21:02:02 +02:00
ConstantDataBuffer ConstantShapeHelper::bufferForShapeInfo(const Nd4jLong *shapeInfo) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(shapeInfo);
return bufferForShapeInfo(descriptor);
}
bool ConstantShapeHelper::checkBufferExistenceForShapeInfo(ShapeDescriptor &descriptor) {
bool result;
int deviceId = 0;
_mutex.lock();
if (_cache[deviceId].count(descriptor) == 0)
result = false;
else
result = true;
_mutex.unlock();
return result;
}
Nd4jLong* ConstantShapeHelper::createShapeInfo(const sd::DataType dataType, const char order, const int rank, const Nd4jLong* shape) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(dataType, order, shape, rank);
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::createShapeInfo(const sd::DataType dataType, const Nd4jLong* shapeInfo) {
return ConstantShapeHelper::createShapeInfo(dataType, shape::order(shapeInfo), shape::rank(shapeInfo), shape::shapeOf(const_cast<Nd4jLong*>(shapeInfo)));
}
Nd4jLong* ConstantShapeHelper::emptyShapeInfo(const sd::DataType dataType) {
2019-06-06 14:21:15 +02:00
auto descriptor = ShapeDescriptor::emptyDescriptor(dataType);
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::scalarShapeInfo(const sd::DataType dataType) {
2019-06-06 14:21:15 +02:00
auto descriptor = ShapeDescriptor::scalarDescriptor(dataType);
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::vectorShapeInfo(const Nd4jLong length, const sd::DataType dataType) {
2019-06-06 14:21:15 +02:00
auto descriptor = ShapeDescriptor::vectorDescriptor(length, dataType);
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::createShapeInfo(const sd::DataType dataType, const char order, const std::vector<Nd4jLong> &shape) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(dataType, order, shape);
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::createShapeInfo(const ShapeDescriptor &descriptor) {
return bufferForShapeInfo(descriptor).primaryAsT<Nd4jLong>();
}
Nd4jLong* ConstantShapeHelper::createFromExisting(Nd4jLong *shapeInfo, bool destroyOriginal) {
ShapeDescriptor descriptor(shapeInfo);
auto result = createShapeInfo(descriptor);
if (destroyOriginal)
RELEASE(shapeInfo, nullptr)
return result;
}
Nd4jLong* ConstantShapeHelper::createFromExisting(Nd4jLong *shapeInfo, sd::memory::Workspace *workspace) {
2019-06-06 14:21:15 +02:00
ShapeDescriptor descriptor(shapeInfo);
auto result = createShapeInfo(descriptor);
RELEASE(shapeInfo, workspace);
return result;
}
////////////////////////////////////////////////////////////////////////
ConstantDataBuffer ConstantShapeHelper::createShapeInfoWithUnitiesForBroadcast(const Nd4jLong* maxShapeInfo, const Nd4jLong* minShapeInfo, sd::memory::Workspace* workspace, const std::vector<int> dimensions) {
Nd4jLong* newShapeInfo = nullptr;
ALLOCATE(newShapeInfo, workspace, shape::shapeInfoLength(shape::rank(maxShapeInfo)), Nd4jLong);
newShapeInfo[0] = shape::rank(maxShapeInfo);
sd::ArrayOptions::copyDataType(newShapeInfo, minShapeInfo); // type
newShapeInfo[2 * newShapeInfo[0] + 2] = shape::elementWiseStride(minShapeInfo); // ews
newShapeInfo[2 * newShapeInfo[0] + 3] = shape::order(minShapeInfo); // order
if(!dimensions.empty()) {
for(uint k = 0, j = 0, i = 0; i < shape::rank(maxShapeInfo); ++i) {
if(j < dimensions.size() && dimensions[j] == i) {
shape::shapeOf(newShapeInfo)[i] = shape::shapeOf(minShapeInfo)[k];
shape::stride(newShapeInfo)[i] = shape::stride(minShapeInfo)[k++];
++j;
}
else{
shape::shapeOf(newShapeInfo)[i] = 1;
shape::stride(newShapeInfo)[i] = 0;
if(shape::sizeAt(minShapeInfo, k) == 1 && dimensions.size() != shape::rank(minShapeInfo))
++k;
}
}
}
else{
for(int j = shape::rank(minShapeInfo) - 1, i = shape::rank(maxShapeInfo) - 1; i >=0 ; --i) {
if(j >= 0) {
shape::shapeOf(newShapeInfo)[i] = shape::shapeOf(minShapeInfo)[j];
shape::stride(newShapeInfo)[i] = shape::shapeOf(minShapeInfo)[j] == 1 ? 0 : shape::stride(minShapeInfo)[j];
--j;
}
else {
shape::shapeOf(newShapeInfo)[i] = 1;
shape::stride(newShapeInfo)[i] = 0;
}
}
}
ShapeDescriptor descriptor(newShapeInfo);
RELEASE(newShapeInfo, workspace);
return bufferForShapeInfo(descriptor);
}
sd::ConstantShapeHelper* sd::ConstantShapeHelper::_INSTANCE = 0;
2019-06-06 14:21:15 +02:00
}
#endif