Brutex
/
cavis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

- fix wrong calculation of elements offsets in batchnorm op when input arrays have unusual (#169) 

Signed-off-by: Yurii <iuriish@yahoo.com>
master
Yurii Shyrma 2020-01-10 23:14:20 +02:00 committed by raver119
parent c84307a6fe
commit bbf88b53dd
4 changed files with 281 additions and 168 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

202
libnd4j/include/ops/declarable/helpers/cpu/batchnorm.cpp
View File

@ -15,7 +15,7 @@
******************************************************************************/ ******************************************************************************/
// //
// @author Yurii Shyrma, created on 25.02.2018 // @author Yurii Shyrma (iuriish@yahoo.com)
// //
@ -31,112 +31,160 @@ namespace helpers {
////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////
template <typename T> template <typename T>
static void batchnorm_(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta, NDArray* output, const std::vector<int>& axes, const double epsilon) { static void batchnorm_(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta,
NDArray* output,
const std::vector<int>& axes, const double epsilon) {
// formula: output = gamma * ((input - mean) / sqrt(variance + epsilon)) + beta // formula: output = gamma * ((input - mean) / sqrt(variance + epsilon)) + beta
NDArray sigmaInvGam(mean); // do not copy mean's buffer, take only its shapeInfo const T* x = input->bufferAsT<T>();
T eps = epsilon; T* z = output->bufferAsT<T>();
const T* m = mean->bufferAsT<T>();
const T* v = variance->bufferAsT<T>();
const T* g = gamma == nullptr ? nullptr : gamma->bufferAsT<T>();
const T* b = beta == nullptr ? nullptr : beta->bufferAsT<T>();
if(gamma != nullptr) { const bool xzSameOffset = shape::haveSameShapeAndStrides(input->getShapeInfo(), output->getShapeInfo());
auto lambda = LAMBDA_TT(x, y, eps) {return x / nd4j::math::nd4j_sqrt<T, T>(y + eps);};
const_cast<NDArray*>(gamma)->applyPairwiseLambda<T>(*variance, lambda, sigmaInvGam);
}
else {
auto lambda = LAMBDA_T(x, eps) { return 1. / nd4j::math::nd4j_sqrt<T, T>(x + eps); };
const_cast<NDArray*>(variance)->applyLambda<T>(lambda, sigmaInvGam);
}
// auto sigmaInvGam = (*variance + epsilon).transform(transform::RSqrt); // sigmaInvGam = 1 / sqrt(variance + epsilon) bool paramSameOffset = shape::haveSameShapeAndStrides(mean->getShapeInfo(), variance->getShapeInfo());
// if(gamma != nullptr) sigmaInvGam *= *gamma; if(paramSameOffset && gamma != nullptr)
paramSameOffset &= shape::haveSameShapeAndStrides(mean->getShapeInfo(), gamma->getShapeInfo());
const T* sigmaBuff = sigmaInvGam.bufferAsT<T>(); if(paramSameOffset && beta != nullptr)
const T* meanBuff = mean->bufferAsT<T>(); paramSameOffset &= shape::haveSameShapeAndStrides(mean->getShapeInfo(), beta->getShapeInfo());
const T* inBuff = input->bufferAsT<T>();
T* outBuff = output->bufferAsT<T>();
const Nd4jLong lenBig = input->lengthOf(); const Nd4jLong lenBig = input->lengthOf();
const Nd4jLong lenSmall = mean->lengthOf(); const Nd4jLong lenSmall = mean->lengthOf();
const Nd4jLong* inShapeInfo = input->getShapeInfo();
const Nd4jLong* meanShapeInfo = mean->getShapeInfo();
uint inShapeInfoCast[MAX_RANK]; const Nd4jLong steps = lenBig / lenSmall;
uint meanShapeInfoCast[MAX_RANK];
bool canCastIn = nd4j::DataTypeUtils::castShapeInfo(inShapeInfo, inShapeInfoCast);
bool canCastMean = nd4j::DataTypeUtils::castShapeInfo(meanShapeInfo, meanShapeInfoCast);
const Nd4jLong step = lenBig / lenSmall;
std::vector<int> dimsToExclude = ShapeUtils::evalDimsToExclude(input->rankOf(), axes); std::vector<int> dimsToExclude = ShapeUtils::evalDimsToExclude(input->rankOf(), axes);
OmpLaunchHelper info(lenBig, lenSmall); OmpLaunchHelper info(lenBig, lenSmall);
if(beta != nullptr) { auto func = PRAGMA_THREADS_DO {
const T* betaBuff = beta->bufferAsT<T>();
auto func = PRAGMA_THREADS_DO {
const auto threadNum = thread_id;
Nd4jLong* inOffsets = new Nd4jLong[step];
Nd4jLong* memBuff = new Nd4jLong[2 * inShapeInfo[0]];
for (int j = 0; j < lenSmall; ++j) { Nd4jLong* xOffsets = new Nd4jLong[steps];
Nd4jLong* zOffsets = xzSameOffset ? xOffsets : new Nd4jLong[steps];
Nd4jLong* auxBuff = new Nd4jLong[2 * input->rankOf()];
const bool isOwner = j < info._numThreads ? threadNum == j : threadNum == j % info._numThreads; for (int j = 0; j < lenSmall; ++j) {
if (!isOwner) continue;
const Nd4jLong start = j * step; const bool isOwner = (j < info._numThreads) ? thread_id == j : thread_id == (j % info._numThreads);
const Nd4jLong end = start + step;
// calculate offset for mean, variance, gamma, beta (all of them have the same shape) if(!isOwner)
auto offsetSmall = shape::indexOffset(j, meanShapeInfo, meanShapeInfoCast, canCastMean); continue;
// calculate offset for input and output (all of them have the same shape)
shape::outerArrayOffsets(inOffsets, j, inShapeInfo, meanShapeInfo, memBuff, dimsToExclude.data());
PRAGMA_OMP_SIMD const auto meanOffset = shape::getIndexOffset(j, mean->getShapeInfo());
for (Nd4jLong i = 0; i < step; ++i) { const auto varOffset = paramSameOffset ? meanOffset : shape::getIndexOffset(j, variance->getShapeInfo());
auto offsetBig = inOffsets[i];
outBuff[offsetBig] = (inBuff[offsetBig] - meanBuff[offsetSmall]) * sigmaBuff[offsetSmall] + betaBuff[offsetSmall]; const auto meanVal = m[meanOffset];
auto sigmaInvGam = static_cast<T>(1) / nd4j::math::nd4j_sqrt<T, T>(v[varOffset] + epsilon);
if(g != nullptr) {
const auto gammaOffset = paramSameOffset ? meanOffset : shape::getIndexOffset(j, gamma->getShapeInfo());
sigmaInvGam *= g[gammaOffset];
}
T betaVal = static_cast<T>(0);
if(b != nullptr) {
const auto betaOffset = paramSameOffset ? meanOffset : shape::getIndexOffset(j, beta->getShapeInfo());
betaVal = b[betaOffset];
}
// calculate offsets for input and output
shape::outerArrayOffsets(xOffsets, j, input->getShapeInfo(), mean->getShapeInfo(), auxBuff, dimsToExclude.data());
if(!xzSameOffset)
shape::outerArrayOffsets(zOffsets, j, output->getShapeInfo(), mean->getShapeInfo(), auxBuff, dimsToExclude.data());
PRAGMA_OMP_SIMD
for (uint i = 0; i < steps; ++i)
z[zOffsets[i]] = (x[xOffsets[i]] - meanVal) * sigmaInvGam + betaVal;
}
delete []auxBuff;
delete []xOffsets;
if(!xzSameOffset)
delete []zOffsets;
};
samediff::Threads::parallel_do(func, info._numThreads);
}
//////////////////////////////////////////////////////////////////////////
template <typename T>
static void batchnorm2_(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta,
NDArray* output,
const std::vector<int>& axes, const double epsilon) {
// formula: output = gamma * ((input - mean) / sqrt(variance + epsilon)) + beta
const auto x = input->bufferAsT<T>();
auto z = output->bufferAsT<T>();
const auto m = mean->bufferAsT<T>();
const auto v = variance->bufferAsT<T>();
const auto g = gamma == nullptr ? nullptr : gamma->bufferAsT<T>();
const auto b = beta == nullptr ? nullptr : beta->bufferAsT<T>();
// xRank == zRank, minRank = meanRank = varianceRank = gammaRank = betaRank
const uint xRank = input->rankOf();
const uint minRank = mean->rankOf();
const uint numAxes = axes.size();
const bool xzSameOffset = shape::haveSameShapeAndStrides(input->getShapeInfo(), output->getShapeInfo());
bool paramSameOffset = shape::haveSameShapeAndStrides(mean->getShapeInfo(), variance->getShapeInfo());
if(paramSameOffset && gamma != nullptr)
paramSameOffset &= shape::haveSameShapeAndStrides(mean->getShapeInfo(), gamma->getShapeInfo());
if(paramSameOffset && beta != nullptr)
paramSameOffset &= shape::haveSameShapeAndStrides(mean->getShapeInfo(), beta->getShapeInfo());
auto func = PRAGMA_THREADS_FOR {
Nd4jLong coords[MAX_RANK];
for (auto i = start; i < stop; i += increment) {
shape::index2coords(i, input->getShapeInfo(), coords);
const auto xOffset = shape::getOffset(input->getShapeInfo(), coords);
const auto zOffset = xzSameOffset ? xOffset : shape::getOffset(output->getShapeInfo(), coords);
if(minRank == xRank) {
for (uint i = 0, j = 0; i < xRank; ++i) {
if(j < numAxes && i != axes[j])
coords[i] = 0;
else
++j;
} }
} }
delete []inOffsets; else // minRank = numAxes = 1 in this case
delete []memBuff; coords[0] = coords[axes[0]];
};
samediff::Threads::parallel_do(func, info._numThreads); const auto meanOffset = shape::getOffset(mean->getShapeInfo(), coords);
} const auto varianceOffset = paramSameOffset ? meanOffset : shape::getOffset(variance->getShapeInfo(), coords);
else {
auto func = PRAGMA_THREADS_DO {
const auto threadNum = thread_id;
Nd4jLong* inOffsets = new Nd4jLong[step];
Nd4jLong* memBuff = new Nd4jLong[2 * inShapeInfo[0]];
for (int j = 0; j < lenSmall; ++j) { T sigmaInvGam = 1. / nd4j::math::nd4j_sqrt<T, T>(v[varianceOffset] + epsilon);
const bool isOwner = j < info._numThreads ? threadNum == j : threadNum == j % info._numThreads;
if (!isOwner) continue;
const Nd4jLong start = j * step; if(g != nullptr) {
const Nd4jLong end = start + step; const auto gammaOffset = paramSameOffset ? meanOffset : shape::getOffset(gamma->getShapeInfo(), coords);
sigmaInvGam *= g[gammaOffset];
// calculate offset for mean, variance, gamma, beta (all of them have the same shape)
auto offsetSmall = shape::indexOffset(j, meanShapeInfo, meanShapeInfoCast, canCastMean);
// calculate offset for input and output (all of them have the same shape)
shape::outerArrayOffsets(inOffsets, j, inShapeInfo, meanShapeInfo, memBuff, dimsToExclude.data());
PRAGMA_OMP_SIMD
for (Nd4jLong i = 0; i < step; ++i) {
auto offsetBig = inOffsets[i];
outBuff[offsetBig] = (inBuff[offsetBig] - meanBuff[offsetSmall]) * sigmaBuff[offsetSmall];
}
} }
delete []inOffsets;
delete []memBuff;
};
samediff::Threads::parallel_do(func, info._numThreads); z[zOffset] = (x[xOffset] - m[meanOffset]) * sigmaInvGam;
}
if(b != nullptr) {
const auto betaOffset = paramSameOffset ? meanOffset : shape::getOffset(beta->getShapeInfo(), coords);
z[zOffset] += b[betaOffset];
}
}
};
samediff::Threads::parallel_for(func, 0, input->lengthOf());
} }
////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////
void batchnorm(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta, NDArray* output, const std::vector<int>& axes, const double epsilon) { void batchnorm(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta, NDArray* output, const std::vector<int>& axes, const double epsilon) {
// batchnorm2_ is slower
BUILD_SINGLE_SELECTOR(input->dataType(), batchnorm_, (input, mean, variance, gamma, beta, output, axes, epsilon), FLOAT_TYPES); BUILD_SINGLE_SELECTOR(input->dataType(), batchnorm_, (input, mean, variance, gamma, beta, output, axes, epsilon), FLOAT_TYPES);
} }

169
libnd4j/include/ops/declarable/helpers/cuda/batchnorm.cu
View File

@ -31,66 +31,66 @@ namespace helpers {
////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////
template<typename T> // template<typename T>
__global__ static void batchnormCuda(const void* vx, const Nd4jLong* xShapeInfo, // __global__ static void batchnormCuda(const void* vx, const Nd4jLong* xShapeInfo,
const void* vMean, const Nd4jLong* meanShapeInfo, // const void* vMean, const Nd4jLong* meanShapeInfo,
const void* vVariance, const Nd4jLong* varianceShapeInfo, // const void* vVariance, const Nd4jLong* varianceShapeInfo,
const void* vGamma, const Nd4jLong* gammaShapeInfo, // const void* vGamma, const Nd4jLong* gammaShapeInfo,
const void* vBeta, const Nd4jLong* betaShapeInfo, // const void* vBeta, const Nd4jLong* betaShapeInfo,
void* vz, const Nd4jLong* zShapeInfo, // void* vz, const Nd4jLong* zShapeInfo,
const Nd4jLong* xTadShapeInfo, const Nd4jLong* xTadOffsets, // const Nd4jLong* xTadShapeInfo, const Nd4jLong* xTadOffsets,
const Nd4jLong* zTadShapeInfo, const Nd4jLong* zTadOffsets, // const Nd4jLong* zTadShapeInfo, const Nd4jLong* zTadOffsets,
const T epsilon) { // const T epsilon) {
const auto x = reinterpret_cast<const T*>(vx); // const auto x = reinterpret_cast<const T*>(vx);
auto z = reinterpret_cast<T*>(vz); // auto z = reinterpret_cast<T*>(vz);
const auto mean = reinterpret_cast<const T*>(vMean); // const auto mean = reinterpret_cast<const T*>(vMean);
const auto variance = reinterpret_cast<const T*>(vVariance); // const auto variance = reinterpret_cast<const T*>(vVariance);
const auto gamma = reinterpret_cast<const T*>(vGamma); // const auto gamma = reinterpret_cast<const T*>(vGamma);
const auto beta = reinterpret_cast<const T*>(vBeta); // const auto beta = reinterpret_cast<const T*>(vBeta);
// maxRank = xRank = zRank, minRank = meanRank = varianceRank = gammaRank = betaRank // // maxRank = xRank = zRank, minRank = meanRank = varianceRank = gammaRank = betaRank
__shared__ Nd4jLong minLen, tadLen, totalThreads; // __shared__ Nd4jLong minLen, tadLen, totalThreads;
if (threadIdx.x == 0) { // if (threadIdx.x == 0) {
totalThreads = gridDim.x * blockDim.x; // totalThreads = gridDim.x * blockDim.x;
minLen = shape::length(meanShapeInfo); // minLen = shape::length(meanShapeInfo);
tadLen = shape::length(xShapeInfo) / minLen; // tadLen = shape::length(xShapeInfo) / minLen;
} // }
__syncthreads(); // __syncthreads();
const auto tid = blockIdx.x * blockDim.x + threadIdx.x; // const auto tid = blockIdx.x * blockDim.x + threadIdx.x;
for (uint i = tid; i < minLen; i += totalThreads) { // for (uint i = tid; i < minLen; i += totalThreads) {
const auto meanOffset = shape::getIndexOffset(i, meanShapeInfo); // const auto meanOffset = shape::getIndexOffset(i, meanShapeInfo);
const auto varianceOffset = shape::getIndexOffset(i, varianceShapeInfo); // const auto varianceOffset = shape::getIndexOffset(i, varianceShapeInfo);
T sigmaInvGam = 1. / nd4j::math::nd4j_sqrt<T, T>(variance[varianceOffset] + epsilon); // T sigmaInvGam = 1. / nd4j::math::nd4j_sqrt<T, T>(variance[varianceOffset] + epsilon);
if(gamma != nullptr) // if(gamma != nullptr)
sigmaInvGam *= gamma[shape::getIndexOffset(i, gammaShapeInfo)]; // sigmaInvGam *= gamma[shape::getIndexOffset(i, gammaShapeInfo)];
auto betaOffset = 0; // auto betaOffset = 0;
if(beta != nullptr) // if(beta != nullptr)
betaOffset = shape::getIndexOffset(i, betaShapeInfo); // betaOffset = shape::getIndexOffset(i, betaShapeInfo);
const auto xTad = x + xTadOffsets[i]; // const auto xTad = x + xTadOffsets[i];
auto zTad = z + zTadOffsets[i]; // auto zTad = z + zTadOffsets[i];
for (uint j = 0; j < tadLen; ++j) { // for (uint j = 0; j < tadLen; ++j) {
const auto xTadOffset = shape::getIndexOffset(j, xTadShapeInfo); // const auto xTadOffset = shape::getIndexOffset(j, xTadShapeInfo);
const auto zTadOffset = shape::getIndexOffset(j, zTadShapeInfo); // const auto zTadOffset = shape::getIndexOffset(j, zTadShapeInfo);
zTad[zTadOffset] = (xTad[xTadOffset] - mean[meanOffset]) * sigmaInvGam; // zTad[zTadOffset] = (xTad[xTadOffset] - mean[meanOffset]) * sigmaInvGam;
if(beta != nullptr) // if(beta != nullptr)
zTad[zTadOffset] += beta[betaOffset]; // zTad[zTadOffset] += beta[betaOffset];
} // }
} // }
} // }
////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////
template<typename T> template<typename T>
@ -110,13 +110,12 @@ __global__ static void batchnormCuda2(const void* vx, const Nd4jLong* xShapeInfo
const auto gamma = reinterpret_cast<const T*>(vGamma); const auto gamma = reinterpret_cast<const T*>(vGamma);
const auto beta = reinterpret_cast<const T*>(vBeta); const auto beta = reinterpret_cast<const T*>(vBeta);
__shared__ int xRank, minRank; // xRank == zRank. minRank = meanRank = varianceRank = gammaRank = betaRank __shared__ int xRank, minRank; // xRank == zRank, minRank = meanRank = varianceRank = gammaRank = betaRank
__shared__ Nd4jLong xLen, totalThreads, *sharedMem; // xLen = zLen __shared__ Nd4jLong xLen, totalThreads; // xLen = zLen
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
extern __shared__ unsigned char shmem[];
sharedMem = reinterpret_cast<Nd4jLong*>(shmem);
totalThreads = gridDim.x * blockDim.x; totalThreads = gridDim.x * blockDim.x;
xLen = shape::length(xShapeInfo); xLen = shape::length(xShapeInfo);
@ -125,7 +124,8 @@ __global__ static void batchnormCuda2(const void* vx, const Nd4jLong* xShapeInfo
} }
__syncthreads(); __syncthreads();
auto coords = sharedMem + threadIdx.x * xRank; Nd4jLong coords[MAX_RANK];
const auto tid = blockIdx.x * blockDim.x + threadIdx.x; const auto tid = blockIdx.x * blockDim.x + threadIdx.x;
for (uint i = tid; i < xLen; i += totalThreads) { for (uint i = tid; i < xLen; i += totalThreads) {
@ -166,24 +166,24 @@ __global__ static void batchnormCuda2(const void* vx, const Nd4jLong* xShapeInfo
} }
/////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////
template<typename T> // template<typename T>
__host__ static void batchnormCudaLauncher(const int blocksPerGrid, const int threadsPerBlock, const cudaStream_t *stream, // __host__ static void batchnormCudaLauncher(const int blocksPerGrid, const int threadsPerBlock, const cudaStream_t *stream,
const void* vx, const Nd4jLong* xShapeInfo, // const void* vx, const Nd4jLong* xShapeInfo,
const void* vMean, const Nd4jLong* meanShapeInfo, // const void* vMean, const Nd4jLong* meanShapeInfo,
const void* vVariance, const Nd4jLong* varianceShapeInfo, // const void* vVariance, const Nd4jLong* varianceShapeInfo,
const void* vGamma, const Nd4jLong* gammaShapeInfo, // const void* vGamma, const Nd4jLong* gammaShapeInfo,
const void* vBeta, const Nd4jLong* betaShapeInfo, // const void* vBeta, const Nd4jLong* betaShapeInfo,
void* vz, const Nd4jLong* zShapeInfo, // void* vz, const Nd4jLong* zShapeInfo,
const Nd4jLong* xTadShapeInfo, const Nd4jLong* xTadOffsets, // const Nd4jLong* xTadShapeInfo, const Nd4jLong* xTadOffsets,
const Nd4jLong* zTadShapeInfo, const Nd4jLong* zTadOffsets, // const Nd4jLong* zTadShapeInfo, const Nd4jLong* zTadOffsets,
const double epsilon) { // const double epsilon) {
batchnormCuda<T><<<blocksPerGrid, threadsPerBlock, 1024, *stream>>>(vx, xShapeInfo, vMean, meanShapeInfo, vVariance, varianceShapeInfo, vGamma, gammaShapeInfo, vBeta, betaShapeInfo, vz, zShapeInfo, xTadShapeInfo, xTadOffsets, zTadShapeInfo, zTadOffsets, static_cast<T>(epsilon)); // batchnormCuda<T><<<blocksPerGrid, threadsPerBlock, 1024, *stream>>>(vx, xShapeInfo, vMean, meanShapeInfo, vVariance, varianceShapeInfo, vGamma, gammaShapeInfo, vBeta, betaShapeInfo, vz, zShapeInfo, xTadShapeInfo, xTadOffsets, zTadShapeInfo, zTadOffsets, static_cast<T>(epsilon));
} // }
/////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////
template<typename T> template<typename T>
__host__ static void batchnormCudaLauncher2(const int blocksPerGrid, const int threadsPerBlock, const int sharedMem, const cudaStream_t *stream, __host__ static void batchnormCudaLauncher2(const int blocksPerGrid, const int threadsPerBlock, const cudaStream_t *stream,
const void* vx, const Nd4jLong* xShapeInfo, const void* vx, const Nd4jLong* xShapeInfo,
const void* vMean, const Nd4jLong* meanShapeInfo, const void* vMean, const Nd4jLong* meanShapeInfo,
const void* vVariance, const Nd4jLong* varianceShapeInfo, const void* vVariance, const Nd4jLong* varianceShapeInfo,
@ -193,42 +193,41 @@ __host__ static void batchnormCudaLauncher2(const int blocksPerGrid, const int t
const int numDims, const int* dims, const int numDims, const int* dims,
const double epsilon) { const double epsilon) {
batchnormCuda2<T><<<blocksPerGrid, threadsPerBlock, sharedMem, *stream>>>(vx, xShapeInfo, vMean, meanShapeInfo, vVariance, varianceShapeInfo, vGamma, gammaShapeInfo, vBeta, betaShapeInfo, vz, zShapeInfo, numDims, dims, static_cast<T>(epsilon)); batchnormCuda2<T><<<blocksPerGrid, threadsPerBlock, 512, *stream>>>(vx, xShapeInfo, vMean, meanShapeInfo, vVariance, varianceShapeInfo, vGamma, gammaShapeInfo, vBeta, betaShapeInfo, vz, zShapeInfo, numDims, dims, static_cast<T>(epsilon));
} }
////////////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////////////
void batchnorm(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta, NDArray* output, const std::vector<int>& axes, const double epsilon) { void batchnorm(const NDArray* input, const NDArray* mean, const NDArray* variance, const NDArray* gamma, const NDArray* beta, NDArray* output, const std::vector<int>& axes, const double epsilon) {
std::vector<int> dimsToExclude = ShapeUtils::evalDimsToExclude(input->rankOf(), axes); // std::vector<int> dimsToExclude = ShapeUtils::evalDimsToExclude(input->rankOf(), axes);
// auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimsToExclude);
// auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->shapeInfo(), dimsToExclude);
// const int threadsPerBlock = MAX_NUM_THREADS / 2;
// const int blocksPerGrid = (mean->lengthOf() + threadsPerBlock - 1) / threadsPerBlock;
// PointersManager manager(input->getContext(), "batchnorm");
// NDArray::prepareSpecialUse({output}, {input, mean, variance, gamma, beta});
// BUILD_SINGLE_SELECTOR(input->dataType(), batchnormCudaLauncher, (blocksPerGrid, threadsPerBlock, input->getContext()->getCudaStream(), input->getSpecialBuffer(), input->getSpecialShapeInfo(), mean->getSpecialBuffer(), mean->getSpecialShapeInfo(), variance->getSpecialBuffer(), variance->getSpecialShapeInfo(), gamma ? gamma->getSpecialBuffer() : nullptr, gamma ? gamma->getSpecialShapeInfo() : nullptr, beta ? beta->getSpecialBuffer() : nullptr, beta ? beta->getSpecialShapeInfo() : nullptr, output->specialBuffer(), output->specialShapeInfo(), packX.platformShapeInfo(), packX.platformOffsets(), packZ.platformShapeInfo(), packZ.platformOffsets(), epsilon), FLOAT_TYPES);
// NDArray::registerSpecialUse({output}, {input, mean, variance, gamma, beta});
// manager.synchronize();
auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimsToExclude);
auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->shapeInfo(), dimsToExclude);
const int threadsPerBlock = MAX_NUM_THREADS / 2; const int threadsPerBlock = MAX_NUM_THREADS / 2;
const int blocksPerGrid = (mean->lengthOf() + threadsPerBlock - 1) / threadsPerBlock; const int blocksPerGrid = (input->lengthOf() + threadsPerBlock - 1) / threadsPerBlock;
PointersManager manager(input->getContext(), "batchnorm"); PointersManager manager(input->getContext(), "batchnorm");
const int* dims = reinterpret_cast<int*>(manager.replicatePointer(axes.data(), axes.size() * sizeof(int)));
NDArray::prepareSpecialUse({output}, {input, mean, variance, gamma, beta}); NDArray::prepareSpecialUse({output}, {input, mean, variance, gamma, beta});
BUILD_SINGLE_SELECTOR(input->dataType(), batchnormCudaLauncher, (blocksPerGrid, threadsPerBlock, input->getContext()->getCudaStream(), input->getSpecialBuffer(), input->getSpecialShapeInfo(), mean->getSpecialBuffer(), mean->getSpecialShapeInfo(), variance->getSpecialBuffer(), variance->getSpecialShapeInfo(), gamma ? gamma->getSpecialBuffer() : nullptr, gamma ? gamma->getSpecialShapeInfo() : nullptr, beta ? beta->getSpecialBuffer() : nullptr, beta ? beta->getSpecialShapeInfo() : nullptr, output->specialBuffer(), output->specialShapeInfo(), packX.platformShapeInfo(), packX.platformOffsets(), packZ.platformShapeInfo(), packZ.platformOffsets(), epsilon), FLOAT_TYPES); BUILD_SINGLE_SELECTOR(input->dataType(), batchnormCudaLauncher2, (blocksPerGrid, threadsPerBlock, input->getContext()->getCudaStream(), input->getSpecialBuffer(), input->getSpecialShapeInfo(), mean->getSpecialBuffer(), mean->getSpecialShapeInfo(), variance->getSpecialBuffer(), variance->getSpecialShapeInfo(), gamma ? gamma->getSpecialBuffer() : nullptr, gamma ? gamma->getSpecialShapeInfo() : nullptr, beta ? beta->getSpecialBuffer() : nullptr, beta ? beta->getSpecialShapeInfo() : nullptr, output->specialBuffer(), output->specialShapeInfo(), axes.size(), dims, epsilon), FLOAT_TYPES);
NDArray::registerSpecialUse({output}, {input, mean, variance, gamma, beta}); NDArray::registerSpecialUse({output}, {input, mean, variance, gamma, beta});
manager.synchronize(); manager.synchronize();
// const int threadsPerBlock = MAX_NUM_THREADS / 4;
// const int blocksPerGrid = (input->lengthOf() + threadsPerBlock - 1) / threadsPerBlock;
// const int sharedMem = sizeof(Nd4jLong) * threadsPerBlock * input->rankOf() + 128;
// PointersManager manager(input->getContext(), "batchnorm");
// const int* dims = reinterpret_cast<int*>(manager.replicatePointer(axes.data(), axes.size() * sizeof(int)));
// NDArray::prepareSpecialUse({output}, {input, mean, variance, gamma, beta});
// BUILD_SINGLE_SELECTOR(input->dataType(), batchnormCudaLauncher2, (blocksPerGrid, threadsPerBlock, sharedMem, input->getContext()->getCudaStream(), input->getSpecialBuffer(), input->getSpecialShapeInfo(), mean->getSpecialBuffer(), mean->getSpecialShapeInfo(), variance->getSpecialBuffer(), variance->getSpecialShapeInfo(), gamma ? gamma->getSpecialBuffer() : nullptr, gamma ? gamma->getSpecialShapeInfo() : nullptr, beta ? beta->getSpecialBuffer() : nullptr, beta ? beta->getSpecialShapeInfo() : nullptr, output->specialBuffer(), output->specialShapeInfo(), axes.size(), dims, epsilon), FLOAT_TYPES);
// NDArray::registerSpecialUse({output}, {input, mean, variance, gamma, beta});
// manager.synchronize();
} }

29
libnd4j/tests_cpu/layers_tests/DeclarableOpsTests10.cpp
View File

@ -3431,6 +3431,35 @@ TEST_F(DeclarableOpsTests10, batchnorm_test6) {
delete results; delete results;
} }
////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests10, batchnorm_test7) {
NDArray input1('c', {3,3,15,15}, nd4j::DataType::FLOAT32);
NDArray input2('c', {3,15,15,3}, nd4j::DataType::FLOAT32);
input2.permutei({0,3,1,2});
NDArray mean ('c', {3}, {0, 0, 0}, nd4j::DataType::FLOAT32);
NDArray variance('c', {3}, {1, 1, 1}, nd4j::DataType::FLOAT32);
NDArray gamma ('c', {3}, {1, 1, 1}, nd4j::DataType::FLOAT32);
NDArray beta ('c', {3}, {0, 0, 0}, nd4j::DataType::FLOAT32);
NDArray out1('c', {3,3,15,15}, nd4j::DataType::FLOAT32);
NDArray out2('c', {3,3,15,15}, nd4j::DataType::FLOAT32);
input1.linspace(-1012, 1);
input2.assign(input1);
nd4j::ops::batchnorm op;
auto res1 = op.execute({&input1, &mean, &variance, &gamma, &beta}, {&out1}, {1e-5}, {1,1,1}, {});
ASSERT_EQ(ND4J_STATUS_OK, res1);
auto res2 = op.execute({&input2, &mean, &variance, &gamma, &beta}, {&out2}, {1e-5}, {1,1,1}, {});
ASSERT_EQ(ND4J_STATUS_OK, res2);
ASSERT_TRUE(out1.equalsTo(out2));
}
/////////////////////////////////////////////////////////////////// ///////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests10, bool_broadcast_test_1) { TEST_F(DeclarableOpsTests10, bool_broadcast_test_1) {

49
libnd4j/tests_cpu/layers_tests/PlaygroundTests.cpp
View File

@ -422,13 +422,50 @@ TEST_F(PlaygroundTests, my) {
delete variableSpace; delete variableSpace;
} }
*/
#include<ops/declarable/helpers/batchnorm.h>
TEST_F(PlaygroundTests, my) { TEST_F(PlaygroundTests, my) {
NDArray a('c',{2,3,4}, nd4j::DataType::DOUBLE); const int N = 10000;
a({0,0, 0,1, 0,1}).printShapeInfo(); const Nd4jLong dim0(128), dim1(128), dim2(128);
a({0,1, 0,0, 0,1}).printShapeInfo();
a({0,0, 0,1, 0,1}).printShapeInfo(); NDArray input('c', {dim0,dim1,dim2}, nd4j::DataType::DOUBLE);
NDArray mean('c', {dim1}, nd4j::DataType::DOUBLE);
NDArray variance('c', {dim1}, nd4j::DataType::DOUBLE);
NDArray gamma('c', {dim1}, nd4j::DataType::DOUBLE);
NDArray beta ('c', {dim1}, nd4j::DataType::DOUBLE);
NDArray output('c', {dim0,dim1,dim2}, nd4j::DataType::DOUBLE);
input.linspace(-100, 0.1);
mean.linspace(-50, 0.15);
variance.linspace(-5, 0.2);
gamma = 1.5;
beta = -2.5;
// warm up
ops::helpers::batchnorm(&input, &mean, &variance, &gamma, &beta, &output, {1}, 1e-5);
auto timeStart = std::chrono::system_clock::now();
for (int i = 0; i < N; ++i)
ops::helpers::batchnorm(&input, &mean, &variance, &gamma, &beta, &output, {1}, 1e-5);
auto timeEnd = std::chrono::system_clock::now();
auto time = std::chrono::duration_cast<std::chrono::microseconds> ((timeEnd - timeStart)/N).count();
printf("time: %li \n", time);
}
*/
}