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

Shugeo_release_fixes3 (#81) 

* Implementation for non_max_suppression_v3 was added. Initial version

* Added check for overcome threshold.

* Added definition for V3 method.

* java remapping for NonMaxSuppressionV3

Signed-off-by: raver119 <raver119@gmail.com>

* Fixed proporly processing of an empty output and test.

* Refactored op to less threshold data to float.

* Implemented cuda-based helper for non_max_suppression_v3 op.

* Fixed fake_quant_with_min_max_vars op.

* Fixed tests with float numbers.

* - assert now stops execution
- sortByKey/sortByValue now have input validation

Signed-off-by: raver119 <raver119@gmail.com>

* missing var

Signed-off-by: raver119 <raver119@gmail.com>

* Fixed proper processing for zero max_size inputs.

* Refactored kernel callers.

* Fixed return statement for logdet op helper.

* Refactored unsorted segment SqrtN op.

* get back 8 tail bytes on CUDA

Signed-off-by: raver119 <raver119@gmail.com>

* Refactored segment prod ops and helpers for cuda and tests.

* Additional test.

* CudaWorkspace tests updated for 8 tail bytes

Signed-off-by: raver119 <raver119@gmail.com>

* special atomic test

Signed-off-by: raver119 <raver119@gmail.com>

* atomicMul/atomicDiv fix for 16bit values

Signed-off-by: raver119 <raver119@gmail.com>

* Eliminated waste prints.
master
shugeo 2019-11-28 20:08:51 +02:00 committed by raver119
parent abd2017a0a
commit 009007120b
24 changed files with 945 additions and 440 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
libnd4j/blas/cuda/NativeOps.cu
View File

@ -2202,10 +2202,17 @@ void sortByKey(Nd4jPointer *extraPointers,
auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]); auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
auto xLength = shape::length(xShapeInfo); auto xLength = shape::length(xShapeInfo);
auto yLength = shape::length(yShapeInfo);
auto xEWS = shape::elementWiseStride(xShapeInfo); auto xEWS = shape::elementWiseStride(xShapeInfo);
auto xType = nd4j::ArrayOptions::dataType(xShapeInfo); auto xType = nd4j::ArrayOptions::dataType(xShapeInfo);
auto yType = nd4j::ArrayOptions::dataType(yShapeInfo); auto yType = nd4j::ArrayOptions::dataType(yShapeInfo);
if (shape::isEmpty(xShapeInfo) || shape::isEmpty(yShapeInfo))
return;
if (xLength != yLength)
throw std::runtime_error("sortByKey: keys and values must have the same size");
// check if xLength is a power of 2, and use bitonic sort, if that's the case // check if xLength is a power of 2, and use bitonic sort, if that's the case
if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) { if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) {
@ -2269,10 +2276,17 @@ void sortByValue(Nd4jPointer *extraPointers,
auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]); auto stream = reinterpret_cast<cudaStream_t *>(extraPointers[1]);
auto xLength = shape::length(xShapeInfo); auto xLength = shape::length(xShapeInfo);
auto yLength = shape::length(yShapeInfo);
auto xEWS = shape::elementWiseStride(xShapeInfo); auto xEWS = shape::elementWiseStride(xShapeInfo);
auto xType = nd4j::ArrayOptions::dataType(yShapeInfo); auto xType = nd4j::ArrayOptions::dataType(yShapeInfo);
auto yType = nd4j::ArrayOptions::dataType(xShapeInfo); auto yType = nd4j::ArrayOptions::dataType(xShapeInfo);
if (shape::isEmpty(xShapeInfo) || shape::isEmpty(yShapeInfo))
return;
if (xLength != yLength)
throw std::runtime_error("sortByValue: keys and values must have the same size");
// check if xLength is a power of 2, and use bitonic sort, if that's the case // check if xLength is a power of 2, and use bitonic sort, if that's the case
if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) { if ((xLength != 0) && ((xLength & (xLength - 1)) == 0) && (xLength <= 1024 * 1024 * 10)) {

6
libnd4j/include/op_boilerplate.h
View File

@ -1461,12 +1461,14 @@
#ifdef _RELEASE #ifdef _RELEASE
#define ALLOCATE_SPECIAL(VARIABLE, WORKSPACE, LENGTH, TT) if (WORKSPACE == nullptr) {auto erc_##VARIABLE = cudaMalloc(reinterpret_cast<void**>(&VARIABLE), LENGTH * sizeof(TT)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] allocation failed", erc_##VARIABLE);} else { }; } else {VARIABLE = reinterpret_cast<TT *>(WORKSPACE->allocateBytes(nd4j::memory::MemoryType::DEVICE, LENGTH * sizeof(TT))); } // we intentionally add 8 tail bytes here to avoid problems with atomic operations
#define ALLOCATE_SPECIAL(VARIABLE, WORKSPACE, LENGTH, TT) if (WORKSPACE == nullptr) {auto erc_##VARIABLE = cudaMalloc(reinterpret_cast<void**>(&VARIABLE), LENGTH * sizeof(TT) + 8); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] allocation failed", erc_##VARIABLE);} else { }; } else {VARIABLE = reinterpret_cast<TT *>(WORKSPACE->allocateBytes(nd4j::memory::MemoryType::DEVICE, LENGTH * sizeof(TT) + 8)); }
#define RELEASE_SPECIAL(VARIABLE, WORKSPACE) if (VARIABLE != nullptr) {if (WORKSPACE == nullptr) { auto erc_##VARIABLE = cudaFree(reinterpret_cast<void *>(VARIABLE)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] deallocation failed", erc_##VARIABLE);}; }; }; #define RELEASE_SPECIAL(VARIABLE, WORKSPACE) if (VARIABLE != nullptr) {if (WORKSPACE == nullptr) { auto erc_##VARIABLE = cudaFree(reinterpret_cast<void *>(VARIABLE)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] deallocation failed", erc_##VARIABLE);}; }; };
#else #else
#define ALLOCATE_SPECIAL(VARIABLE, WORKSPACE, LENGTH, TT) if (WORKSPACE == nullptr) {auto erc_##VARIABLE = cudaMalloc(reinterpret_cast<void**>(&VARIABLE), LENGTH * sizeof(TT)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] allocation failed", erc_##VARIABLE);} else { nd4j::memory::MemoryTracker::getInstance()->countIn(nd4j::memory::MemoryType::DEVICE, VARIABLE, LENGTH * sizeof(TT)); }; } else {VARIABLE = reinterpret_cast<TT *>(WORKSPACE->allocateBytes(nd4j::memory::MemoryType::DEVICE, LENGTH * sizeof(TT))); } // we intentionally add 8 tail bytes here to avoid problems with atomic operations
#define ALLOCATE_SPECIAL(VARIABLE, WORKSPACE, LENGTH, TT) if (WORKSPACE == nullptr) {auto erc_##VARIABLE = cudaMalloc(reinterpret_cast<void**>(&VARIABLE), LENGTH * sizeof(TT) + 8); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] allocation failed", erc_##VARIABLE);} else { nd4j::memory::MemoryTracker::getInstance()->countIn(nd4j::memory::MemoryType::DEVICE, VARIABLE, LENGTH * sizeof(TT)); }; } else {VARIABLE = reinterpret_cast<TT *>(WORKSPACE->allocateBytes(nd4j::memory::MemoryType::DEVICE, LENGTH * sizeof(TT) + 8)); }
#define RELEASE_SPECIAL(VARIABLE, WORKSPACE) if (VARIABLE != nullptr) {if (WORKSPACE == nullptr) { nd4j::memory::MemoryTracker::getInstance()->countOut(VARIABLE); auto erc_##VARIABLE = cudaFree(reinterpret_cast<void *>(VARIABLE)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] deallocation failed", erc_##VARIABLE);}; }; }; #define RELEASE_SPECIAL(VARIABLE, WORKSPACE) if (VARIABLE != nullptr) {if (WORKSPACE == nullptr) { nd4j::memory::MemoryTracker::getInstance()->countOut(VARIABLE); auto erc_##VARIABLE = cudaFree(reinterpret_cast<void *>(VARIABLE)); if (erc_##VARIABLE != 0) {throw cuda_exception::build("[DEVICE] deallocation failed", erc_##VARIABLE);}; }; };
#endif #endif

4
libnd4j/include/ops/declarable/generic/parity_ops/assert.cpp
View File

@ -29,8 +29,8 @@ namespace nd4j {
OP_IMPL(Assert, 1, 1, false) { OP_IMPL(Assert, 1, 1, false) {
auto x = INPUT_VARIABLE(0); auto x = INPUT_VARIABLE(0);
if (x->e<float>(0) == 0.0f) { if (!x->e<bool>(0)) {
nd4j_printf("Assertion failed for node [%i]\n", block.getNodeId()); REQUIRE_TRUE(false, 0, "Assertion failed for node [%i]\n", block.getNodeId());
} }
return Status::OK(); return Status::OK();

133
libnd4j/include/ops/declarable/generic/parity_ops/non_max_suppression.cpp
View File

@ -21,10 +21,10 @@
#include <ops/declarable/CustomOperations.h> #include <ops/declarable/CustomOperations.h>
#include <ops/declarable/helpers/image_suppression.h> #include <ops/declarable/helpers/image_suppression.h>
#if NOT_EXCLUDED(OP_image_non_max_suppression)
namespace nd4j { namespace nd4j {
namespace ops { namespace ops {
#if NOT_EXCLUDED(OP_image_non_max_suppression)
CUSTOM_OP_IMPL(non_max_suppression, 2, 1, false, 0, 0) { CUSTOM_OP_IMPL(non_max_suppression, 2, 1, false, 0, 0) {
auto boxes = INPUT_VARIABLE(0); auto boxes = INPUT_VARIABLE(0);
auto scales = INPUT_VARIABLE(1); auto scales = INPUT_VARIABLE(1);
@ -56,11 +56,24 @@ namespace nd4j {
if (boxes->isEmpty() || scales->isEmpty()) if (boxes->isEmpty() || scales->isEmpty())
return Status::OK(); return Status::OK();
REQUIRE_TRUE(boxes->rankOf() == 2, 0, "image.non_max_suppression: The rank of boxes array should be 2, but %i is given", boxes->rankOf()); if (output->isEmpty())
REQUIRE_TRUE(boxes->sizeAt(1) == 4, 0, "image.non_max_suppression: The last dimension of boxes array should be 4, but %i is given", boxes->sizeAt(1)); return Status::OK();
REQUIRE_TRUE(scales->rankOf() == 1 && scales->lengthOf() == boxes->sizeAt(0), 0, "image.non_max_suppression: The rank of scales array should be 1, but %i is given", boxes->rankOf());
helpers::nonMaxSuppression(block.launchContext(), boxes, scales, maxOutputSize, overlayThreshold, scoreThreshold, output); REQUIRE_TRUE(boxes->rankOf() == 2, 0, "image.non_max_suppression: The rank of boxes array should be 2, "
"but %i is given", boxes->rankOf());
REQUIRE_TRUE(boxes->sizeAt(1) == 4, 0, "image.non_max_suppression: The last dimension of boxes array "
"should be 4, but %i is given", boxes->sizeAt(1));
REQUIRE_TRUE(scales->rankOf() == 1 && scales->lengthOf() == boxes->sizeAt(0), 0,
"image.non_max_suppression: The rank of scales array should be 1, but %i is given", boxes->rankOf());
REQUIRE_TRUE(overlayThreshold >= 0. && overlayThreshold <= 1., 0, "image.non_max_suppressio: The overlay "
"threashold should be in [0, 1], but "
"%lf is given.", overlayThreshold);
REQUIRE_TRUE(boxes->dataType() == scales->dataType(), 0,
"image.non_max_suppression: Boxes and scores inputs should have the same data type, but %s and %s "
"were given.", DataTypeUtils::asString(boxes->dataType()).c_str(),
DataTypeUtils::asString(scales->dataType()).c_str());
helpers::nonMaxSuppression(block.launchContext(), boxes, scales, maxOutputSize, overlayThreshold,
scoreThreshold, output);
return Status::OK(); return Status::OK();
} }
@ -77,20 +90,22 @@ namespace nd4j {
else else
REQUIRE_TRUE(false, 0, "image.non_max_suppression: Max output size argument cannot be retrieved."); REQUIRE_TRUE(false, 0, "image.non_max_suppression: Max output size argument cannot be retrieved.");
if (maxOutputSize > 0) {
auto actualIndicesCount = shape::sizeAt(in, 0); auto actualIndicesCount = shape::sizeAt(in, 0);
if (block.getTArguments()->size() > 1 || block.width() > 4) { if (block.getTArguments()->size() > 1 || block.width() > 4) {
auto scoreThreshold = block.getTArguments()->size() > 1?T_ARG(1):INPUT_VARIABLE(4)->e<double>(0); auto scoreThreshold =
block.getTArguments()->size() > 1 ? T_ARG(1) : INPUT_VARIABLE(4)->e<double>(0);
auto scales = INPUT_VARIABLE(1); auto scales = INPUT_VARIABLE(1);
scales->syncToHost(); scales->syncToHost();
for (auto e = 0; e < scales->lengthOf(); e++) { for (auto e = 0; e < scales->lengthOf(); e++) {
if (scales->e<float>(e) < (float)scoreThreshold) { if (scales->e<float>(e) < (float) scoreThreshold) {
actualIndicesCount--; actualIndicesCount--;
} }
} }
} }
if (actualIndicesCount < maxOutputSize) if (actualIndicesCount < maxOutputSize)
maxOutputSize = actualIndicesCount; maxOutputSize = actualIndicesCount;
}
outputShape = ConstantShapeHelper::getInstance()->vectorShapeInfo(maxOutputSize, DataType::INT32); outputShape = ConstantShapeHelper::getInstance()->vectorShapeInfo(maxOutputSize, DataType::INT32);
return SHAPELIST(outputShape); return SHAPELIST(outputShape);
@ -100,7 +115,107 @@ namespace nd4j {
->setAllowedInputTypes(nd4j::DataType::ANY) ->setAllowedInputTypes(nd4j::DataType::ANY)
->setAllowedOutputTypes({ALL_INDICES}); ->setAllowedOutputTypes({ALL_INDICES});
} }
#endif
#if NOT_EXCLUDED(OP_image_non_max_suppression_v3)
DECLARE_TYPES(non_max_suppression_v3) {
getOpDescriptor()
->setAllowedInputTypes(nd4j::DataType::ANY)
->setAllowedOutputTypes({ALL_INDICES});
}
CUSTOM_OP_IMPL(non_max_suppression_v3, 2, 1, false, 0, 0) {
auto boxes = INPUT_VARIABLE(0);
auto scales = INPUT_VARIABLE(1);
auto output = OUTPUT_VARIABLE(0);
int maxOutputSize; // = INT_ARG(0);
if (block.width() > 2)
maxOutputSize = INPUT_VARIABLE(2)->e<int>(0);
else if (block.getIArguments()->size() == 1)
maxOutputSize = INT_ARG(0);
else
REQUIRE_TRUE(false, 0, "image.non_max_suppression: Max output size argument cannot be retrieved.");
double overlayThreshold = 0.5;
double scoreThreshold = - DataTypeUtils::infOrMax<float>();
if (block.width() > 3) {
overlayThreshold = INPUT_VARIABLE(3)->e<double>(0);
}
else if (block.getTArguments()->size() > 0) {
overlayThreshold = T_ARG(0);
}
if (block.width() > 4) {
scoreThreshold = INPUT_VARIABLE(4)->e<double>(0);
}
else if (block.getTArguments()->size() > 1) {
scoreThreshold = T_ARG(1);
}
if (boxes->isEmpty() || scales->isEmpty())
return Status::OK();
if (output->isEmpty())
return Status::OK();
REQUIRE_TRUE(boxes->rankOf() == 2, 0, "image.non_max_suppression: The rank of boxes array should be 2, but "
"%i is given", boxes->rankOf());
REQUIRE_TRUE(boxes->sizeAt(1) == 4, 0, "image.non_max_suppression: The last dimension of boxes array should "
"be 4, but %i is given", boxes->sizeAt(1));
REQUIRE_TRUE(scales->rankOf() == 1 && scales->lengthOf() == boxes->sizeAt(0), 0,
"image.non_max_suppression: The rank of scales array should be 1, but %i is given", boxes->rankOf());
REQUIRE_TRUE(overlayThreshold >= 0. && overlayThreshold <= 1., 0,
"image.non_max_suppression_v3: The overlay threashold should be in [0, 1], but %lf given.",
overlayThreshold);
REQUIRE_TRUE(boxes->dataType() == scales->dataType(), 0,
"image.non_max_suppression_v3: Boxes and scores inputs should have the same data type, but %s and %s "
"were given.", DataTypeUtils::asString(boxes->dataType()).c_str(),
DataTypeUtils::asString(scales->dataType()).c_str());
helpers::nonMaxSuppressionV3(block.launchContext(), boxes, scales, maxOutputSize, overlayThreshold,
scoreThreshold, output);
return Status::OK();
}
DECLARE_SHAPE_FN(non_max_suppression_v3) {
auto in = inputShape->at(0);
int outRank = shape::rank(in);
Nd4jLong *outputShape = nullptr;
int maxOutputSize;
if (block.width() > 2)
maxOutputSize = INPUT_VARIABLE(2)->e<int>(0);
else if (block.getIArguments()->size() == 1)
maxOutputSize = INT_ARG(0);
else
REQUIRE_TRUE(false, 0, "image.non_max_suppression: Max output size argument cannot be retrieved.");
auto boxes = INPUT_VARIABLE(0);
auto scales = INPUT_VARIABLE(1);
double overlayThreshold = 0.5;
double scoreThreshold = - DataTypeUtils::infOrMax<float>();
if (block.width() > 3) {
overlayThreshold = INPUT_VARIABLE(3)->e<double>(0);
}
else if (block.getTArguments()->size() > 0) {
overlayThreshold = T_ARG(0);
}
if (block.width() > 4) {
scoreThreshold = INPUT_VARIABLE(4)->e<double>(0);
}
else if (block.getTArguments()->size() > 1) {
scoreThreshold = T_ARG(1);
}
auto len = maxOutputSize;
if (len > 0)
len = helpers::nonMaxSuppressionV3(block.launchContext(), boxes, scales, maxOutputSize, overlayThreshold, scoreThreshold, nullptr);
outputShape = ConstantShapeHelper::getInstance()->vectorShapeInfo(len, DataType::INT32);
return SHAPELIST(outputShape);
}
#endif
} }
} }
#endif

12
libnd4j/include/ops/declarable/generic/parity_ops/unsorted_segment_prod.cpp
View File

@ -61,9 +61,9 @@ namespace nd4j {
} }
DECLARE_TYPES(unsorted_segment_prod) { DECLARE_TYPES(unsorted_segment_prod) {
getOpDescriptor() getOpDescriptor()
->setAllowedOutputTypes({ALL_FLOATS}) ->setAllowedOutputTypes({ALL_FLOATS, ALL_INTS})
->setAllowedInputTypes(0, {ALL_FLOATS}) ->setAllowedInputTypes(0, {ALL_FLOATS, ALL_INTS})
->setAllowedInputTypes(1, {ALL_INTS}) ->setAllowedInputTypes(1, {ALL_INDICES})
->setSameMode(false); ->setSameMode(false);
} }
@ -88,10 +88,10 @@ namespace nd4j {
DECLARE_TYPES(unsorted_segment_prod_bp) { DECLARE_TYPES(unsorted_segment_prod_bp) {
getOpDescriptor() getOpDescriptor()
->setAllowedOutputTypes(0, {ALL_FLOATS}) ->setAllowedOutputTypes(0, {ALL_FLOATS})
->setAllowedOutputTypes(1, {ALL_INTS}) ->setAllowedOutputTypes(1, {ALL_INDICES})
->setAllowedInputTypes(0, {ALL_FLOATS}) ->setAllowedInputTypes(0, {ALL_FLOATS})
->setAllowedInputTypes(1, {ALL_INTS}) ->setAllowedInputTypes(1, {ALL_INDICES})
->setAllowedInputTypes(2,{ALL_FLOATS}) ->setAllowedInputTypes(2,{ALL_FLOATS, ALL_INTS})
->setSameMode(false); ->setSameMode(false);
} }

5
libnd4j/include/ops/declarable/headers/parity_ops.h
View File

@ -1723,7 +1723,7 @@ namespace nd4j {
#endif #endif
/** /**
* image.non_max_suppression op. * image.non_max_suppression ops.
* input: * input:
* 0 - boxes - 2D-tensor with shape (num_boxes, 4) by float type * 0 - boxes - 2D-tensor with shape (num_boxes, 4) by float type
* 1 - scales - 1D-tensor with shape (num_boxes) by float type * 1 - scales - 1D-tensor with shape (num_boxes) by float type
@ -1741,6 +1741,9 @@ namespace nd4j {
#if NOT_EXCLUDED(OP_image_non_max_suppression) #if NOT_EXCLUDED(OP_image_non_max_suppression)
DECLARE_CUSTOM_OP(non_max_suppression, 2, 1, false, 0, 0); DECLARE_CUSTOM_OP(non_max_suppression, 2, 1, false, 0, 0);
#endif #endif
#if NOT_EXCLUDED(OP_image_non_max_suppression_v3)
DECLARE_CUSTOM_OP(non_max_suppression_v3, 2, 1, false, 0, 0);
#endif
/* /*
* image.non_max_suppression_overlaps op. * image.non_max_suppression_overlaps op.

11
libnd4j/include/ops/declarable/helpers/cpu/fake_quantization.cpp
View File

@ -77,7 +77,13 @@ namespace helpers {
} }
} }
} }
//
//const auto clamped = inputs.cwiseMin(nudged_max).cwiseMax(nudged_min);
// const auto clamped_shifted = clamped - nudged_min;
// outputs.device(d) = (clamped_shifted / nudged_scale_repl + 0.5f).floor() *
// nudged_scale_repl +
// nudged_min;
//
template <typename T> template <typename T>
void fakeQuantWithMinMaxVars_(NDArray* input, NDArray* min, NDArray* max, int numBits, bool narrowed, NDArray* output) { void fakeQuantWithMinMaxVars_(NDArray* input, NDArray* min, NDArray* max, int numBits, bool narrowed, NDArray* output) {
int lowIntBound = narrowed ? 1 : 0; int lowIntBound = narrowed ? 1 : 0;
@ -95,7 +101,8 @@ namespace helpers {
else if (val > nudgedMax) else if (val > nudgedMax)
val = nudgedMax; val = nudgedMax;
// converse value with scale and shifted with nudged min // converse value with scale and shifted with nudged min
return (nd4j::math::nd4j_floor<T,T>((val - nudgedMin)/scale + T(0.5)) * scale + nudgedMin); val -= nudgedMin;
return (nd4j::math::nd4j_floor<T,T>(val / scale + T(0.5f)) * scale + nudgedMin);
}; };
input->applyLambda<T>(fakeQuantizationWithMinMax, output); input->applyLambda<T>(fakeQuantizationWithMinMax, output);

77
libnd4j/include/ops/declarable/helpers/cpu/image_suppression.cpp
View File

@ -19,7 +19,7 @@
// //
#include <ops/declarable/helpers/image_suppression.h> #include <ops/declarable/helpers/image_suppression.h>
//#include <blas/NDArray.h> #include <NDArrayFactory.h>
#include <algorithm> #include <algorithm>
#include <numeric> #include <numeric>
#include <queue> #include <queue>
@ -90,17 +90,61 @@ namespace helpers {
} }
} }
} }
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// Return intersection-over-union overlap between boxes i and j
template <typename T>
static inline T similirityV3_(NDArray const& boxes, Nd4jLong i, Nd4jLong j) {
const T zero = static_cast<T>(0.f);
const T yminI = math::nd4j_min(boxes.t<T>(i, 0), boxes.t<T>(i, 2));
const T xminI = math::nd4j_min(boxes.t<T>(i, 1), boxes.t<T>(i, 3));
const T ymaxI = math::nd4j_max(boxes.t<T>(i, 0), boxes.t<T>(i, 2));
const T xmaxI = math::nd4j_max(boxes.t<T>(i, 1), boxes.t<T>(i, 3));
const T yminJ = math::nd4j_min(boxes.t<T>(j, 0), boxes.t<T>(j, 2));
const T xminJ = math::nd4j_min(boxes.t<T>(j, 1), boxes.t<T>(j, 3));
const T ymaxJ = math::nd4j_max(boxes.t<T>(j, 0), boxes.t<T>(j, 2));
const T xmaxJ = math::nd4j_max(boxes.t<T>(j, 1), boxes.t<T>(j, 3));
const T areaI = (ymaxI - yminI) * (xmaxI - xminI);
const T areaJ = (ymaxJ - yminJ) * (xmaxJ - xminJ);
if (areaI <= zero || areaJ <= zero) {
return zero;
}
const T intersectionYmin = math::nd4j_max(yminI, yminJ);
const T intersectionXmin = math::nd4j_max(xminI, xminJ);
const T intersectionYmax = math::nd4j_min(ymaxI, ymaxJ);
const T intersectionXmax = math::nd4j_min(xmaxI, xmaxJ);
const T intersectionY = intersectionYmax - intersectionYmin;
const T intersectionX = intersectionXmax - intersectionXmin;
const T intersectionArea = math::nd4j_max(intersectionY, zero) * math::nd4j_max(intersectionX, zero);
return intersectionArea / (areaI + areaJ - intersectionArea);
}
template <typename T>
static inline T similiratyOverlaps_(NDArray const& boxes, Nd4jLong i, Nd4jLong j) {
return boxes.t<T>(i, j);
}
typedef NDArray (*SimiliratyFunc)(NDArray const& boxes, Nd4jLong i, Nd4jLong j);
static NDArray similiratyOverlaps(NDArray const& boxes, Nd4jLong i, Nd4jLong j) {
NDArray res(boxes.dataType(), boxes.getContext()); // = NDArrayFactory::create(0.);
BUILD_SINGLE_SELECTOR(boxes.dataType(), res = similiratyOverlaps_, (boxes, i, j) , FLOAT_TYPES);
return res;
}
static NDArray similiratyV3(NDArray const& boxes, Nd4jLong i, Nd4jLong j) {
NDArray res(boxes.dataType(), boxes.getContext()); // = NDArrayFactory::create(0.);
BUILD_SINGLE_SELECTOR(boxes.dataType(), res = similirityV3_, (boxes, i, j) , FLOAT_TYPES);
return res;
}
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
template <typename T, typename I> template <typename T, typename I>
static Nd4jLong static Nd4jLong
nonMaxSuppressionGeneric_(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int outputSize, nonMaxSuppressionGeneric_(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int outputSize,
double overlapThreshold, double scoreThreshold, NDArray* output) { float overlapThreshold, float scoreThreshold, NDArray* output, SimiliratyFunc f) {
// const int outputSize = maxSize->e<int>(0);
auto numBoxes = boxes->sizeAt(0); auto numBoxes = boxes->sizeAt(0);
//std::vector<T> scoresData(numBoxes);
T* scoresData = scores->dataBuffer()->primaryAsT<T>(); T* scoresData = scores->dataBuffer()->primaryAsT<T>();
//std::copy_n(scores->getDataBuffer()->primaryAsT<T>(), numBoxes, scoresData.begin());
// Data structure for a selection candidate in NMS. // Data structure for a selection candidate in NMS.
struct Candidate { struct Candidate {
@ -113,9 +157,10 @@ namespace helpers {
return ((bsI._score == bsJ._score) && (bsI._boxIndex > bsJ._boxIndex)) || return ((bsI._score == bsJ._score) && (bsI._boxIndex > bsJ._boxIndex)) ||
(bsI._score < bsJ._score); (bsI._score < bsJ._score);
}; };
std::priority_queue<Candidate, std::deque<Candidate>, decltype(cmp)> candidatePriorityQueue(cmp); std::priority_queue<Candidate, std::deque<Candidate>, decltype(cmp)> candidatePriorityQueue(cmp);
for (auto i = 0; i < scores->lengthOf(); ++i) { for (auto i = 0; i < scores->lengthOf(); ++i) {
if (scoresData[i] > scoreThreshold) { if ((float)scoresData[i] > (float)scoreThreshold) {
candidatePriorityQueue.emplace(Candidate({i, scoresData[i], 0})); candidatePriorityQueue.emplace(Candidate({i, scoresData[i], 0}));
} }
} }
@ -139,17 +184,18 @@ namespace helpers {
// following loop. // following loop.
bool shouldHardSuppress = false; bool shouldHardSuppress = false;
for (int j = static_cast<int>(selected.size()) - 1; j >= nextCandidate._suppressBeginIndex; --j) { for (int j = static_cast<int>(selected.size()) - 1; j >= nextCandidate._suppressBeginIndex; --j) {
similarity = boxes->t<T>(nextCandidate._boxIndex, selected[j]); auto similarityA = f(*boxes, nextCandidate._boxIndex, selected[j]); //boxes->t<T>(nextCandidate._boxIndex, selected[j]);
similarity = similarityA.template t<T>(0);
nextCandidate._score *= T(similarity <= overlapThreshold?1.0:0.); //suppressWeightFunc(similarity); nextCandidate._score *= T(similarity <= overlapThreshold?1.0:0.); //suppressWeightFunc(similarity);
// First decide whether to perform hard suppression // First decide whether to perform hard suppression
if (similarity >= static_cast<T>(overlapThreshold)) { if ((float)similarity >= static_cast<float>(overlapThreshold)) {
shouldHardSuppress = true; shouldHardSuppress = true;
break; break;
} }
// If next_candidate survives hard suppression, apply soft suppression // If next_candidate survives hard suppression, apply soft suppression
if (nextCandidate._score <= scoreThreshold) break; if ((float)nextCandidate._score <= (float)scoreThreshold) break;
} }
// If `nextCandidate._score` has not dropped below `scoreThreshold` // If `nextCandidate._score` has not dropped below `scoreThreshold`
// by this point, then we know that we went through all of the previous // by this point, then we know that we went through all of the previous
@ -169,7 +215,7 @@ namespace helpers {
selected.push_back(nextCandidate._boxIndex); selected.push_back(nextCandidate._boxIndex);
// selected_scores.push_back(nextCandidate._score); // selected_scores.push_back(nextCandidate._score);
} }
if (nextCandidate._score > scoreThreshold) { if ((float)nextCandidate._score > (float)scoreThreshold) {
// Soft suppression has occurred and current score is still greater than // Soft suppression has occurred and current score is still greater than
// score_threshold; add next_candidate back onto priority queue. // score_threshold; add next_candidate back onto priority queue.
candidatePriorityQueue.push(nextCandidate); candidatePriorityQueue.push(nextCandidate);
@ -188,12 +234,19 @@ namespace helpers {
Nd4jLong Nd4jLong
nonMaxSuppressionGeneric(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize, nonMaxSuppressionGeneric(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output) { double overlapThreshold, double scoreThreshold, NDArray* output) {
BUILD_DOUBLE_SELECTOR(boxes->dataType(), output == nullptr?DataType::INT32:output->dataType(), return nonMaxSuppressionGeneric_, (context, boxes, scores, maxSize, overlapThreshold, scoreThreshold, output), FLOAT_TYPES, INTEGER_TYPES); BUILD_DOUBLE_SELECTOR(boxes->dataType(), output == nullptr?DataType::INT32:output->dataType(), return nonMaxSuppressionGeneric_, (context, boxes, scores, maxSize, overlapThreshold, scoreThreshold, output, similiratyOverlaps), FLOAT_TYPES, INTEGER_TYPES);
return 0;
}
Nd4jLong
nonMaxSuppressionV3(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output) {
BUILD_DOUBLE_SELECTOR(boxes->dataType(), output == nullptr?DataType::INT32:output->dataType(), return nonMaxSuppressionGeneric_, (context, boxes, scores, maxSize, overlapThreshold, scoreThreshold, output, similiratyV3), FLOAT_TYPES, INTEGER_TYPES);
return 0; return 0;
} }
BUILD_DOUBLE_TEMPLATE(template Nd4jLong nonMaxSuppressionGeneric_, (nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize, BUILD_DOUBLE_TEMPLATE(template Nd4jLong nonMaxSuppressionGeneric_, (nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output), FLOAT_TYPES, INTEGER_TYPES); float overlapThreshold, float scoreThreshold, NDArray* output, SimiliratyFunc similiratyFunc), FLOAT_TYPES, INTEGER_TYPES);
void void
nonMaxSuppression(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize, nonMaxSuppression(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize,

2
libnd4j/include/ops/declarable/helpers/cuda/fake_quantization.cu
View File

@ -100,7 +100,7 @@ namespace helpers {
val = nudgedMax; val = nudgedMax;
} }
output[shape::getIndexOffset(b * channels + i, outputShape)] = output[shape::getIndexOffset(b * channels + i, outputShape)] =
(math::nd4j_floor<T, T>((val - nudgedMin) / scale + T(0.5)) * scale + nudgedMin); (math::nd4j_floor<T, T>((val - nudgedMin) / scale + T(0.5f)) * scale + nudgedMin);
}; };
} }
} }

88
libnd4j/include/ops/declarable/helpers/cuda/image_suppression.cu
View File

@ -79,7 +79,49 @@ namespace helpers {
return intersectionValue > threshold; return intersectionValue > threshold;
} }
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// template <typename T>
static __device__ T similirityV3(T* boxes, Nd4jLong* boxesShape, int previousIndex, int nextIndex) {
Nd4jLong previous0[] = {previousIndex, 0};
Nd4jLong previous1[] = {previousIndex, 1};
Nd4jLong previous2[] = {previousIndex, 2};
Nd4jLong previous3[] = {previousIndex, 3};
Nd4jLong next0[] = {nextIndex, 0};
Nd4jLong next1[] = {nextIndex, 1};
Nd4jLong next2[] = {nextIndex, 2};
Nd4jLong next3[] = {nextIndex, 3};
// we have rectangle with given max values. Compute vexes of rectangle first
T minYPrev = nd4j::math::nd4j_min(boxes[shape::getOffset(boxesShape, previous0)], boxes[shape::getOffset(boxesShape, previous2)]);
T minXPrev = nd4j::math::nd4j_min(boxes[shape::getOffset(boxesShape, previous1)], boxes[shape::getOffset(boxesShape, previous3)]);
T maxYPrev = nd4j::math::nd4j_max(boxes[shape::getOffset(boxesShape, previous0)], boxes[shape::getOffset(boxesShape, previous2)]);
T maxXPrev = nd4j::math::nd4j_max(boxes[shape::getOffset(boxesShape, previous1)], boxes[shape::getOffset(boxesShape, previous3)]);
T minYNext = nd4j::math::nd4j_min(boxes[shape::getOffset(boxesShape, next0)], boxes[shape::getOffset(boxesShape, next2)]);
T minXNext = nd4j::math::nd4j_min(boxes[shape::getOffset(boxesShape, next1)], boxes[shape::getOffset(boxesShape, next3)]);
T maxYNext = nd4j::math::nd4j_max(boxes[shape::getOffset(boxesShape, next0)], boxes[shape::getOffset(boxesShape, next2)]);
T maxXNext = nd4j::math::nd4j_max(boxes[shape::getOffset(boxesShape, next1)], boxes[shape::getOffset(boxesShape, next3)]);
// compute areas for comparation
T areaPrev = (maxYPrev - minYPrev) * (maxXPrev - minXPrev);
T areaNext = (maxYNext - minYNext) * (maxXNext - minXNext);
// of course, areas should be positive
if (areaNext <= T(0.f) || areaPrev <= T(0.f)) return false;
// compute intersection of rectangles
T minIntersectionY = nd4j::math::nd4j_max(minYPrev, minYNext);
T minIntersectionX = nd4j::math::nd4j_max(minXPrev, minXNext);
T maxIntersectionY = nd4j::math::nd4j_min(maxYPrev, maxYNext);
T maxIntersectionX = nd4j::math::nd4j_min(maxXPrev, maxXNext);
T intersectionArea =
nd4j::math::nd4j_max(T(maxIntersectionY - minIntersectionY), T(0.0f)) *
nd4j::math::nd4j_max(T(maxIntersectionX - minIntersectionX), T(0.0f));
T intersectionValue = intersectionArea / (areaPrev + areaNext - intersectionArea);
// final check
return intersectionValue;
}
////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// shouldSelectKernel - compute status for all selected rectangles (boxes) // shouldSelectKernel - compute status for all selected rectangles (boxes)
// //
// we compute boolean flag as shared uint32 and return it on final only for the first thread // we compute boolean flag as shared uint32 and return it on final only for the first thread
@ -139,7 +181,7 @@ namespace helpers {
} }
} }
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
// nonMaxSuppressionV2 algorithm - given from TF NonMaxSuppressionV2 implementation // nonMaxSuppressionV2 algorithm - given from TF NonMaxSuppressionV2 implementation
// //
template <typename T, typename I> template <typename T, typename I>
@ -200,24 +242,33 @@ namespace helpers {
//////////////////////////////////////////////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////
template <typename T, typename I> template <typename T, typename I>
static __device__ bool checkOverlapBoxes(T* boxes, Nd4jLong* shape, T* scores, I* indices, I* selectedIndices, I* startIndices, I selectedSize, I nextCandidateIndex, T overlapThreshold, T scoreThreshold) { static __device__ bool checkOverlapBoxes(T* boxes, Nd4jLong* shape, T* scores, I* indices, I* selectedIndices, I* startIndices, I selectedSize, I nextCandidateIndex, T overlapThreshold, T scoreThreshold, bool simple) {
bool shouldHardSuppress = false; bool shouldHardSuppress = false;
T& nextCandidateScore = scores[nextCandidateIndex]; T& nextCandidateScore = scores[nextCandidateIndex];
I selectedIndex = indices[nextCandidateIndex]; I selectedIndex = indices[nextCandidateIndex];
I finish = startIndices[nextCandidateIndex]; I finish = startIndices[nextCandidateIndex];
for (int j = selectedSize; j > finish; --j) { for (int j = selectedSize; j > finish; --j) {
T boxVal;
if (simple) {
Nd4jLong xPos[] = {selectedIndex, selectedIndices[j - 1]}; Nd4jLong xPos[] = {selectedIndex, selectedIndices[j - 1]};
auto xShift = shape::getOffset(shape, xPos, 0); auto xShift = shape::getOffset(shape, xPos, 0);
nextCandidateScore *= (boxes[xShift] <= static_cast<T>(overlapThreshold)?T(1.):T(0.));// boxVal = boxes[xShift];
}
else {
boxVal = similirityV3(boxes, shape, selectedIndex, selectedIndices[j - 1]);
}
if (boxVal > static_cast<T>(overlapThreshold))
nextCandidateScore = static_cast<T>(0.f);
// First decide whether to perform hard suppression // First decide whether to perform hard suppression
if (boxes[xShift] >= overlapThreshold) { if (boxVal >= overlapThreshold) {
shouldHardSuppress = true; shouldHardSuppress = true;
break; break;
} }
// If nextCandidate survives hard suppression, apply soft suppression // If nextCandidate survives hard suppression, apply soft suppression
if (nextCandidateScore <= scoreThreshold) break; if (nextCandidateScore <= static_cast<T>(scoreThreshold)) break;
} }
return shouldHardSuppress; return shouldHardSuppress;
@ -226,7 +277,7 @@ namespace helpers {
template <typename T, typename I> template <typename T, typename I>
static __global__ void static __global__ void
suppressNonMaxOverlapKernel(T* boxes, Nd4jLong* boxesShape, T* scoresData, I* indices, I* startIndices, Nd4jLong length, I maxOutputLen, suppressNonMaxOverlapKernel(T* boxes, Nd4jLong* boxesShape, T* scoresData, I* indices, I* startIndices, Nd4jLong length, I maxOutputLen,
T overlapThreshold, T scoreThreshold, I* output, Nd4jLong* outputShape, I* outputLength) { T overlapThreshold, T scoreThreshold, I* output, Nd4jLong* outputShape, I* outputLength, bool simple) {
__shared__ I selectedSize; __shared__ I selectedSize;
__shared__ I* tempOutput; __shared__ I* tempOutput;
@ -253,7 +304,7 @@ namespace helpers {
} }
// check for overlaps // check for overlaps
bool shouldHardSuppress = checkOverlapBoxes(boxes, boxesShape, scoresData, indices, tempOutput, startIndices, selectedSize, bool shouldHardSuppress = checkOverlapBoxes(boxes, boxesShape, scoresData, indices, tempOutput, startIndices, selectedSize,
nextCandidateIndex, overlapThreshold, scoreThreshold);//false; nextCandidateIndex, overlapThreshold, scoreThreshold, simple);//false;
T nextCandidateScore = scoresData[nextCandidateIndex]; T nextCandidateScore = scoresData[nextCandidateIndex];
startIndices[nextCandidateIndex] = selectedSize; startIndices[nextCandidateIndex] = selectedSize;
@ -285,7 +336,7 @@ namespace helpers {
template <typename T, typename I> template <typename T, typename I>
static Nd4jLong static Nd4jLong
nonMaxSuppressionGeneric_(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int outputSize, nonMaxSuppressionGeneric_(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int outputSize,
double overlapThreshold, double scoreThreshold, NDArray* output) { double overlapThreshold, double scoreThreshold, NDArray* output, bool simple) {
auto stream = context->getCudaStream(); auto stream = context->getCudaStream();
if (output) if (output)
NDArray::prepareSpecialUse({output}, {boxes, scores}); NDArray::prepareSpecialUse({output}, {boxes, scores});
@ -315,16 +366,16 @@ namespace helpers {
Nd4jLong res = 0; Nd4jLong res = 0;
if (output) { // this part used when output shape already calculated to fill up values on output if (output) { // this part used when output shape already calculated to fill up values on output
DataBuffer selectedSizeBuf(&selectedSize, sizeof(I), DataTypeUtils::fromT<I>()); DataBuffer selectedSizeBuf(&selectedSize, sizeof(I), DataTypeUtils::fromT<I>());
suppressNonMaxOverlapKernel <<<1, 1, 1024, *stream >>> (boxes->dataBuffer()->specialAsT<T>(), suppressNonMaxOverlapKernel<<<1, 1, 1024, *stream >>> (boxes->dataBuffer()->specialAsT<T>(),
boxes->specialShapeInfo(), scoresData, indexBuf, startIndices, scores->lengthOf(), (I) outputSize, boxes->specialShapeInfo(), scoresData, indexBuf, startIndices, scores->lengthOf(), (I) outputSize,
T(overlapThreshold), T(scoreThreshold), output->dataBuffer()->specialAsT<I>(), output->specialShapeInfo(), T(overlapThreshold), T(scoreThreshold), output->dataBuffer()->specialAsT<I>(), output->specialShapeInfo(),
selectedSizeBuf.specialAsT<I>()); selectedSizeBuf.specialAsT<I>(), simple);
} }
else { // this case used on calculation of output shape. Output and output shape shoulde be nullptr. else { // this case used on calculation of output shape. Output and output shape shoulde be nullptr.
DataBuffer selectedSizeBuf(&selectedSize, sizeof(I), DataTypeUtils::fromT<I>()); DataBuffer selectedSizeBuf(&selectedSize, sizeof(I), DataTypeUtils::fromT<I>());
suppressNonMaxOverlapKernel <<<1, 1, 1024, *stream >>> (boxes->dataBuffer()->specialAsT<T>(), suppressNonMaxOverlapKernel<<<1, 1, 1024, *stream >>> (boxes->dataBuffer()->specialAsT<T>(),
boxes->specialShapeInfo(), scoresData, indexBuf, startIndices, scores->lengthOf(), (I)outputSize, boxes->specialShapeInfo(), scoresData, indexBuf, startIndices, scores->lengthOf(), (I)outputSize,
T(overlapThreshold), T(scoreThreshold), (I*)nullptr, (Nd4jLong*) nullptr, selectedSizeBuf.specialAsT<I>()); T(overlapThreshold), T(scoreThreshold), (I*)nullptr, (Nd4jLong*) nullptr, selectedSizeBuf.specialAsT<I>(), simple);
selectedSizeBuf.syncToPrimary(context, true); selectedSizeBuf.syncToPrimary(context, true);
res = *selectedSizeBuf.primaryAsT<I>(); res = *selectedSizeBuf.primaryAsT<I>();
} }
@ -344,7 +395,16 @@ namespace helpers {
Nd4jLong nonMaxSuppressionGeneric(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize, double threshold, double scoreThreshold, NDArray* output) { Nd4jLong nonMaxSuppressionGeneric(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize, double threshold, double scoreThreshold, NDArray* output) {
BUILD_DOUBLE_SELECTOR(boxes->dataType(), output ? output->dataType():DataType::INT32, return nonMaxSuppressionGeneric_, BUILD_DOUBLE_SELECTOR(boxes->dataType(), output ? output->dataType():DataType::INT32, return nonMaxSuppressionGeneric_,
(context, boxes, scales, maxSize, threshold, scoreThreshold, output), (context, boxes, scales, maxSize, threshold, scoreThreshold, output, true),
FLOAT_TYPES, INDEXING_TYPES);
return boxes->sizeAt(0);
}
Nd4jLong
nonMaxSuppressionV3(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output) {
BUILD_DOUBLE_SELECTOR(boxes->dataType(), output ? output->dataType():DataType::INT32, return nonMaxSuppressionGeneric_,
(context, boxes, scores, maxSize, overlapThreshold, scoreThreshold, output, false),
FLOAT_TYPES, INDEXING_TYPES); FLOAT_TYPES, INDEXING_TYPES);
return boxes->sizeAt(0); return boxes->sizeAt(0);
} }

41
libnd4j/include/ops/declarable/helpers/cuda/lup.cu
View File

@ -825,37 +825,30 @@ namespace helpers {
NDArray::prepareSpecialUse({output}, {input}); NDArray::prepareSpecialUse({output}, {input});
auto n2 = input->sizeAt(-1) * input->sizeAt(-2); auto n2 = input->sizeAt(-1) * input->sizeAt(-2);
auto stream = context->getCudaStream(); auto stream = context->getCudaStream();
std::unique_ptr<NDArray> tempOutput(input->dup()); NDArray tempOutput(*input);
// auto inputs = tempOutput->allTensorsAlongDimension({input->rankOf() - 2, input->rankOf() - 1});
// for (Nd4jLong e = 0; e < packX.numberOfTads(); e++) { cholesky(context, input, &tempOutput, false);
// auto subArray = inputs->at(e);
// cholesky(context, subArray, subArray, true); auto outputBuf = output->dataBuffer()->specialAsT<T>(); //reinterpret_cast<T*>(output->specialBuffer()); // + e * n2; // + e * n2;
// } auto inputBuf = tempOutput.dataBuffer()->specialAsT<T>(); //reinterpret_cast<T*>(tempOutput->specialBuffer());
// delete inputs; output->nullify();
cholesky(context, input, tempOutput.get(), false); auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(tempOutput.getShapeInfo(),
tempOutput->syncToHost(); {tempOutput.rankOf() - 2,
tempOutput->printIndexedBuffer("Cholesky res!!!"); tempOutput.rankOf() - 1});
auto outputBuf = reinterpret_cast<T*>(output->specialBuffer()); // + e * n2; // + e * n2; logDetKernel<T> <<< 128, 512, 256, *stream >>>(inputBuf, tempOutput.specialShapeInfo(),
auto inputBuf = reinterpret_cast<T*>(tempOutput->specialBuffer()); packX.numberOfTads(), packX.specialShapeInfo(),
output->assign(0); packX.specialOffsets(), outputBuf, output->specialShapeInfo());
output->syncToDevice(); output->tickWriteDevice();
auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(tempOutput->getShapeInfo(),
{input->rankOf() - 2,
input->rankOf() - 1});
logDetKernel<T> << < packX.numberOfTads(), n2, 128, *stream >> >
(inputBuf, tempOutput->specialShapeInfo(), packX.numberOfTads(), packX.specialShapeInfo(), packX.specialOffsets(), outputBuf, output->specialShapeInfo());
// }
NDArray::registerSpecialUse({output}, {input}); NDArray::registerSpecialUse({output}, {input});
//delete tempOutput;
return Status::OK(); return Status::OK();
} }
int logdetFunctor(nd4j::LaunchContext *context, NDArray *input, NDArray *output) { int logdetFunctor(nd4j::LaunchContext *context, NDArray *input, NDArray *output) {
BUILD_SINGLE_SELECTOR(output->dataType(), logdetFunctor_, (context, input, output), FLOAT_NATIVE); BUILD_SINGLE_SELECTOR(output->dataType(), return logdetFunctor_, (context, input, output), FLOAT_NATIVE);
} }
BUILD_SINGLE_TEMPLATE(template int logdetFunctor_, // BUILD_SINGLE_TEMPLATE(template int logdetFunctor_,
(nd4j::LaunchContext * context, NDArray * input, NDArray * output), FLOAT_NATIVE); // (nd4j::LaunchContext * context, NDArray * input, NDArray * output), FLOAT_NATIVE);
} }
} }
} }

132
libnd4j/include/ops/declarable/helpers/cuda/segment_prod.cu
View File

@ -35,129 +35,86 @@ namespace helpers {
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
template <typename T, typename I> template <typename T, typename I>
static __global__ void segmentProdLinearKernel(void* input, Nd4jLong* inputShape, int* starts, int* lengths, Nd4jLong numOfClasses, void* output, Nd4jLong* outputShape) { static __global__ void segmentProdLinearKernel(void* input, Nd4jLong* inputShape, int* starts, int* lengths,
__shared__ T* val; Nd4jLong numOfClasses, void* output, Nd4jLong* outputShape) {
__shared__ Nd4jLong xLen, zLen, segment, zIndex;
__shared__ Nd4jLong xLen, zLen;
__shared__ T* x; __shared__ T* x;
__shared__ T* z; __shared__ T* z;
__shared__ int threadsPerSegment, start, finish;
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
threadsPerSegment = (gridDim.x + numOfClasses - 1) / numOfClasses;
segment = blockIdx.x / threadsPerSegment;
x = reinterpret_cast<T*>(input); x = reinterpret_cast<T*>(input);
z = reinterpret_cast<T*>(output); z = reinterpret_cast<T*>(output);
extern __shared__ unsigned char shmem[];
val = reinterpret_cast<T*>(shmem);
xLen = shape::length(inputShape); xLen = shape::length(inputShape);
zLen = shape::length(outputShape); zLen = shape::length(outputShape);
if (segment < numOfClasses) {
zIndex = shape::getIndexOffset(segment, outputShape);
start = starts[segment];
finish = start + lengths[segment];
//val[segment] = ;
z[zIndex] = x[shape::getIndexOffset(start, inputShape)];
val[segment] = z[zIndex];
}
} }
__syncthreads(); __syncthreads();
// auto tid = threadIdx.x + blockIdx.x * blockDim.x;
// auto step = blockDim.x * gridDim.x;
for (auto e = start + threadIdx.x + 1; e < finish; e += blockDim.x) { for(auto segment = blockIdx.x; segment < numOfClasses; segment += gridDim.x) {
auto zIndex = shape::getIndexOffset(segment, outputShape);
auto start = starts[segment];
auto finish = start + lengths[segment];
if (lengths[segment] == 0) {
continue;
}
for (auto e = start + threadIdx.x; e < finish; e += blockDim.x) {
auto xIndex = shape::getIndexOffset(e, inputShape); auto xIndex = shape::getIndexOffset(e, inputShape);
nd4j::math::atomics::nd4j_atomicMul(&val[segment], x[xIndex]); nd4j::math::atomics::nd4j_atomicMul(&z[segment], x[xIndex]);
}
} }
__syncthreads();
if (threadIdx.x == 0) {
z[zIndex] = val[segment];
}
__syncthreads();
} }
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
template <typename T, typename I> template <typename T, typename I>
static __global__ void unsortedSegmentProdLinearKernel(void* input, Nd4jLong* inputShape, void* indices, Nd4jLong* indicesShape, int* starts, int* lengths, Nd4jLong numOfClasses, void* output, Nd4jLong* outputShape) { static __global__ void unsortedSegmentProdLinearKernel(T* input, Nd4jLong* inputShape, I* indices, Nd4jLong* indicesShape, int* starts, int* lengths, Nd4jLong numOfClasses, T* output, Nd4jLong* outputShape) {
__shared__ T* val; __shared__ Nd4jLong xLen, zLen;
__shared__ Nd4jLong xLen, zLen, segment, zIndex;
__shared__ T* x;
__shared__ T* z;
__shared__ I* y; //int threadsPerSegment, start, finish;
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
// threadsPerSegment = (gridDim.x + numOfClasses - 1) / numOfClasses;
segment = blockIdx.x;// / threadsPerSegment;
x = reinterpret_cast<T*>(input);
z = reinterpret_cast<T*>(output);
y = reinterpret_cast<I*>(indices);
// extern __shared__ unsigned char shmem[];
// val = reinterpret_cast<T*>(shmem);
xLen = shape::length(inputShape); xLen = shape::length(inputShape);
zLen = shape::length(outputShape); zLen = shape::length(outputShape);
// if (segment < numOfClasses) {
zIndex = shape::getIndexOffset(segment, outputShape);
//start = starts[segment];
//finish = start + lengths[segment];
if (lengths[segment] > 0)
z[zIndex] = x[shape::getIndexOffset(starts[segment], inputShape)];
else
z[zIndex] = 0; //DataTypeUtils::max<T>();
// val[segment] = z[zIndex];
// }
} }
__syncthreads(); __syncthreads();
if (lengths[segment] > 0) auto start = threadIdx.x + blockIdx.x * blockDim.x;
for (auto e = threadIdx.x; e < xLen; e += blockDim.x) { auto step = blockDim.x * gridDim.x;
auto xIndex = shape::getIndexOffset(e, inputShape); for (auto idx = start; idx < xLen; idx += step) {
auto yIndex = shape::getIndexOffset(e, indicesShape); auto xIndex = shape::getIndexOffset(idx, inputShape);
if (y[yIndex] == segment && e != starts[segment]) { auto yIndex = shape::getIndexOffset(idx, indicesShape);
nd4j::math::atomics::nd4j_atomicMul(&z[zIndex], x[xIndex]); auto segment = indices[yIndex];
auto zIndex = shape::getIndexOffset(segment, outputShape);
if (lengths[segment] == 0) {
continue;
} }
nd4j::math::atomics::nd4j_atomicMul(&output[zIndex], input[xIndex]);
} }
} }
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
// SegmentProd kernel // SegmentProd kernel
template <typename T, typename I> template <typename T, typename I>
static __global__ void segmentProdTadKernel(void* inputBuf, Nd4jLong* inputShape, Nd4jLong* inputTads, Nd4jLong* inputTadOffsets, I* indices, int* starts, int* lengths, Nd4jLong numOfClasses, void* outputBuf, Nd4jLong* outputShape, Nd4jLong* outputTads, Nd4jLong* outputTadOffsets) { static __global__ void segmentProdTadKernel(void* inputBuf, Nd4jLong* inputShape, Nd4jLong* inputTads,
__shared__ T* val; Nd4jLong* inputTadOffsets, I* indices, int* starts, int* lengths, Nd4jLong numOfClasses, void* outputBuf,
__shared__ Nd4jLong len, segment, zIndex, total; Nd4jLong* outputShape, Nd4jLong* outputTads, Nd4jLong* outputTadOffsets) {
__shared__ T* z;
__shared__ int threadsPerSegment, start, finish; __shared__ Nd4jLong len, total;
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
segment = indices[blockIdx.x]; // / threadsPerSegment;
z = reinterpret_cast<T*>(outputBuf) + outputTadOffsets[segment];
len = shape::length(inputTads);
start = starts[segment];
finish = start + lengths[segment];
total = shape::sizeAt(inputShape, 0); total = shape::sizeAt(inputShape, 0);
len = shape::length(inputTads);
} }
__syncthreads(); __syncthreads();
auto idx = blockIdx.x; for (auto idx = blockIdx.x; idx < total; idx += gridDim.x) {
if (blockIdx.x <= total) {
auto x = reinterpret_cast<T *>(inputBuf) + inputTadOffsets[idx]; auto x = reinterpret_cast<T *>(inputBuf) + inputTadOffsets[idx];
if (blockIdx.x == start) { auto segment = indices[idx]; // / threadsPerSegment;
auto z = reinterpret_cast<T *>(outputBuf) + outputTadOffsets[segment];
auto start = starts[segment];
auto finish = start + lengths[segment];
if (lengths[segment] == 0) continue;
for (auto e = threadIdx.x; e < len; e += blockDim.x) { for (auto e = threadIdx.x; e < len; e += blockDim.x) {
auto xIndex = shape::getIndexOffset(e, inputTads); auto xIndex = shape::getIndexOffset(e, inputTads);
auto zIndex = shape::getIndexOffset(e, outputTads); auto zIndex = shape::getIndexOffset(e, outputTads);
nd4j::math::atomics::nd4j_atomicMul(&z[zIndex], x[xIndex]); nd4j::math::atomics::nd4j_atomicMul(&z[zIndex], x[xIndex]);
} }
} }
else {
for (auto e = threadIdx.x; e < len; e += blockDim.x) {
auto xIndex = shape::getIndexOffset(e, inputTads);
auto zIndex = shape::getIndexOffset(e, outputTads);
if (lengths[segment] > 0)
nd4j::math::atomics::nd4j_atomicMul(&z[zIndex], x[xIndex]);
}
}
}
} }
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
@ -177,7 +134,7 @@ namespace helpers {
int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer()); int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer());
if (input->isVector()) { if (input->isVector()) {
segmentProdLinearKernel<T,I><<<numClasses, input->lengthOf(), numClasses * 32 + 32, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), begins, lengths, numClasses, output->specialBuffer(), output->specialShapeInfo()); segmentProdLinearKernel<T,I><<<128, 256, 128, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), begins, lengths, numClasses, output->specialBuffer(), output->specialShapeInfo());
} }
else { else {
std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0}); std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0});
@ -187,7 +144,7 @@ namespace helpers {
Nd4jLong* inputTadOffsets = packX.specialOffsets(); Nd4jLong* inputTadOffsets = packX.specialOffsets();
Nd4jLong* outputTads = packZ.specialShapeInfo(); Nd4jLong* outputTads = packZ.specialShapeInfo();
Nd4jLong* outputTadOffsets = packZ.specialOffsets(); Nd4jLong* outputTadOffsets = packZ.specialOffsets();
segmentProdTadKernel<T,I><<<input->sizeAt(0), 512, 2048, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), inputTads, inputTadOffsets, reinterpret_cast<I*>(indices->specialBuffer()), begins, lengths, numClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets); segmentProdTadKernel<T,I><<<128, 512, 2048, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), inputTads, inputTadOffsets, reinterpret_cast<I*>(indices->specialBuffer()), begins, lengths, numClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets);
} }
} }
@ -214,12 +171,15 @@ namespace helpers {
fillUpSegments(indices, numOfClasses, classesRangesBegs, classesRangesLens); fillUpSegments(indices, numOfClasses, classesRangesBegs, classesRangesLens);
int* begins = reinterpret_cast<int*>(classesRangesBegs.specialBuffer()); int* begins = reinterpret_cast<int*>(classesRangesBegs.specialBuffer());
int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer()); int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer());
output->assign(1);
if (input->isVector()) { if (input->isVector()) {
unsortedSegmentProdLinearKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), indices->specialBuffer(), indices->specialShapeInfo(), begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo()); unsortedSegmentProdLinearKernel<T,I><<<128, 256, 256, *stream>>>(
input->dataBuffer()->specialAsT<T>(), input->specialShapeInfo(),
indices->dataBuffer()->specialAsT<I>(), indices->specialShapeInfo(), begins, lengths, numOfClasses,
output->dataBuffer()->specialAsT<T>(), output->specialShapeInfo());
} }
else { else {
output->assign(1);
std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0}); std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0});
auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimensions); auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimensions);
auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->getShapeInfo(), dimensions); auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->getShapeInfo(), dimensions);
@ -228,7 +188,7 @@ namespace helpers {
Nd4jLong* outputTads = packZ.specialShapeInfo(); Nd4jLong* outputTads = packZ.specialShapeInfo();
Nd4jLong* outputTadOffsets = packZ.specialOffsets(); Nd4jLong* outputTadOffsets = packZ.specialOffsets();
dims.x = input->sizeAt(0); dims.x = input->sizeAt(0);
segmentProdTadKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), inputTads, inputTadOffsets, reinterpret_cast<I*>(indices->specialBuffer()), begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets); segmentProdTadKernel<T,I><<<128, 256, 256, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), inputTads, inputTadOffsets, reinterpret_cast<I*>(indices->specialBuffer()), begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets);
} }
} }

96
libnd4j/include/ops/declarable/helpers/cuda/segment_sqrtn.cu
View File

@ -32,77 +32,48 @@ namespace ops {
namespace helpers { namespace helpers {
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
template <typename T, typename I> template <typename T, typename I>
static __global__ void unsortedSegmentSqrtNLinearKernel(void* input, Nd4jLong* inputShape, void* indices, Nd4jLong* indicesShape, int* starts, int* lengths, Nd4jLong numOfClasses, void* output, Nd4jLong* outputShape) { static __global__ void unsortedSegmentSqrtNLinearKernel(T* input, Nd4jLong* inputShape, I* indices, Nd4jLong* indicesShape, int* starts, int* lengths, Nd4jLong numOfClasses, T* output, Nd4jLong* outputShape) {
__shared__ T* val; __shared__ Nd4jLong xLen, zLen;
__shared__ Nd4jLong xLen, zLen, segment, zIndex;
__shared__ T* x;
__shared__ T* z;
__shared__ I* y; //int threadsPerSegment, start, finish;
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
// threadsPerSegment = (gridDim.x + numOfClasses - 1) / numOfClasses;
segment = blockIdx.x;// / threadsPerSegment;
x = reinterpret_cast<T*>(input);
z = reinterpret_cast<T*>(output);
y = reinterpret_cast<I*>(indices);
// extern __shared__ unsigned char shmem[];
// val = reinterpret_cast<T*>(shmem);
xLen = shape::length(inputShape); xLen = shape::length(inputShape);
zLen = shape::length(outputShape); zLen = shape::length(outputShape);
// if (segment < numOfClasses) {
zIndex = shape::getIndexOffset(segment, outputShape);
//start = starts[segment];
//finish = start + lengths[segment];
if (lengths[segment] > 0)
z[zIndex] = x[shape::getIndexOffset(starts[segment], inputShape)] / nd4j::math::nd4j_sqrt<int, T>(lengths[segment]);
else
z[zIndex] = 0; //DataTypeUtils::max<T>();
// val[segment] = z[zIndex];
// }
} }
__syncthreads(); __syncthreads();
if (lengths[segment] > 0)
for (auto e = threadIdx.x + 1; e < xLen; e += blockDim.x) { auto start = threadIdx.x + blockIdx.x * blockDim.x;
auto xIndex = shape::getIndexOffset(e, inputShape); auto step = blockDim.x * gridDim.x;
auto yIndex = shape::getIndexOffset(e, indicesShape);
if (y[yIndex] == segment && e != starts[segment]) { for (auto idx = start; idx < xLen; idx += step) {
nd4j::math::atomics::nd4j_atomicAdd(&z[zIndex], x[xIndex] / nd4j::math::nd4j_sqrt<int, T>(lengths[segment])); auto yIndex = shape::getIndexOffset(idx, indicesShape);
} auto segment = indices[yIndex];
auto zIndex = shape::getIndexOffset(segment, outputShape);
if (lengths[segment] == 0) continue;
auto xIndex = shape::getIndexOffset(idx, inputShape);
nd4j::math::atomics::nd4j_atomicAdd(&output[zIndex], input[xIndex] / nd4j::math::nd4j_sqrt<int, T>(lengths[segment]));
} }
} }
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
// SegmentSqrtN kernel // SegmentSqrtN kernel
template <typename T, typename I> template <typename T, typename I>
static __global__ void segmentSqrtNTadKernel(void* inputBuf, Nd4jLong* inputShape, Nd4jLong* inputTads, Nd4jLong* inputTadOffsets, I* indices, int* starts, int* lengths, Nd4jLong numOfClasses, void* outputBuf, Nd4jLong* outputShape, Nd4jLong* outputTads, Nd4jLong* outputTadOffsets) { static __global__ void segmentSqrtNTadKernel(T* inputBuf, Nd4jLong* inputShape, Nd4jLong* inputTads, Nd4jLong* inputTadOffsets, I* indices, int* starts, int* lengths, Nd4jLong numOfClasses, void* outputBuf, Nd4jLong* outputShape, Nd4jLong* outputTads, Nd4jLong* outputTadOffsets) {
__shared__ T* val;
__shared__ Nd4jLong len, segment, zIndex, total; __shared__ Nd4jLong len, total;
__shared__ T* z;
__shared__ int threadsPerSegment, start, finish;
if (threadIdx.x == 0) { if (threadIdx.x == 0) {
segment = indices[blockIdx.x]; // / threadsPerSegment;
z = reinterpret_cast<T*>(outputBuf) + outputTadOffsets[segment];
len = shape::length(inputTads);
start = starts[segment];
finish = start + lengths[segment];
total = shape::sizeAt(inputShape, 0); total = shape::sizeAt(inputShape, 0);
len = shape::length(inputTads);
} }
__syncthreads(); __syncthreads();
auto idx = blockIdx.x; for (auto idx = blockIdx.x; idx < total; idx += gridDim.x) {
if (blockIdx.x <= total) { auto segment = indices[idx];
auto x = reinterpret_cast<T *>(inputBuf) + inputTadOffsets[idx]; auto x = inputBuf + inputTadOffsets[idx];
if (blockIdx.x == start) { auto z = reinterpret_cast<T *>(outputBuf) + outputTadOffsets[segment];
for (auto e = threadIdx.x; e < len; e += blockDim.x) { auto start = starts[segment];
auto xIndex = shape::getIndexOffset(e, inputTads); auto finish = start + lengths[segment];
auto zIndex = shape::getIndexOffset(e, outputTads);
z[zIndex] = x[xIndex] / nd4j::math::nd4j_sqrt<int, T>(lengths[segment]);
}
}
else {
for (auto e = threadIdx.x; e < len; e += blockDim.x) { for (auto e = threadIdx.x; e < len; e += blockDim.x) {
auto xIndex = shape::getIndexOffset(e, inputTads); auto xIndex = shape::getIndexOffset(e, inputTads);
auto zIndex = shape::getIndexOffset(e, outputTads); auto zIndex = shape::getIndexOffset(e, outputTads);
@ -110,7 +81,6 @@ namespace helpers {
} }
} }
} }
}
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //
template <typename T, typename I> template <typename T, typename I>
static void unsortedSegmentSqrtNFunctor_(nd4j::LaunchContext* context, NDArray* input, NDArray* indices, Nd4jLong numOfClasses, NDArray* output) { static void unsortedSegmentSqrtNFunctor_(nd4j::LaunchContext* context, NDArray* input, NDArray* indices, Nd4jLong numOfClasses, NDArray* output) {
@ -122,17 +92,21 @@ namespace helpers {
// classes.applyTrueBroadcast(nd4j::BroadcastOpsTuple::Assign(), &row, &classes); // classes.applyTrueBroadcast(nd4j::BroadcastOpsTuple::Assign(), &row, &classes);
classesRangesBegs.assign(indices->lengthOf()); classesRangesBegs.assign(indices->lengthOf());
classesRangesLens.assign(0); classesRangesLens.assign(0);
dim3 dims(numOfClasses, indices->lengthOf(), numOfClasses * 32 + 32); // dim3 dims(numOfClasses, indices->lengthOf(), numOfClasses * 32 + 32);
dim3 dims(128, 256, 256);
// int* classesBuf = reinterpret_cast<int*>(classes.specialBuffer()); // int* classesBuf = reinterpret_cast<int*>(classes.specialBuffer());
fillUpSegments(indices, numOfClasses, classesRangesBegs, classesRangesLens); fillUpSegments(indices, numOfClasses, classesRangesBegs, classesRangesLens);
int* begins = reinterpret_cast<int*>(classesRangesBegs.specialBuffer()); int* begins = reinterpret_cast<int*>(classesRangesBegs.specialBuffer());
int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer()); int* lengths = reinterpret_cast<int*>(classesRangesLens.specialBuffer());
output->nullify();
if (input->isVector()) { if (input->isVector()) {
unsortedSegmentSqrtNLinearKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), indices->specialBuffer(), indices->specialShapeInfo(), begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo()); unsortedSegmentSqrtNLinearKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(
input->dataBuffer()->specialAsT<T>(), input->specialShapeInfo(),
indices->dataBuffer()->specialAsT<I>(), indices->specialShapeInfo(), begins, lengths, numOfClasses,
output->dataBuffer()->specialAsT<T>(), output->specialShapeInfo());
} }
else { else {
output->assign(0); output->nullify();
std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0}); std::vector<int> dimensions = ShapeUtils::evalDimsToExclude(input->rankOf(), {0});
auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimensions); auto packX = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(input->getShapeInfo(), dimensions);
auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->getShapeInfo(), dimensions); auto packZ = nd4j::ConstantTadHelper::getInstance()->tadForDimensions(output->getShapeInfo(), dimensions);
@ -141,7 +115,9 @@ namespace helpers {
Nd4jLong* outputTads = packZ.specialShapeInfo(); Nd4jLong* outputTads = packZ.specialShapeInfo();
Nd4jLong* outputTadOffsets = packZ.specialOffsets(); Nd4jLong* outputTadOffsets = packZ.specialOffsets();
dims.x = input->sizeAt(0); dims.x = input->sizeAt(0);
segmentSqrtNTadKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(input->specialBuffer(), input->specialShapeInfo(), inputTads, inputTadOffsets, reinterpret_cast<I*>(indices->specialBuffer()), begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets); segmentSqrtNTadKernel<T,I><<<dims.x, dims.y, dims.z, *stream>>>(
input->dataBuffer()->specialAsT<T>(), input->specialShapeInfo(), inputTads, inputTadOffsets, indices->dataBuffer()->specialAsT<I>(),
begins, lengths, numOfClasses, output->specialBuffer(), output->specialShapeInfo(), outputTads, outputTadOffsets);
} }
} }
// -------------------------------------------------------------------------------------------------------------- // // -------------------------------------------------------------------------------------------------------------- //

2
libnd4j/include/ops/declarable/helpers/image_suppression.h
View File

@ -28,6 +28,8 @@ namespace helpers {
void nonMaxSuppression(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize, void nonMaxSuppression(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output); double overlapThreshold, double scoreThreshold, NDArray* output);
Nd4jLong nonMaxSuppressionV3(nd4j::LaunchContext * context, NDArray* boxes, NDArray* scales, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output);
Nd4jLong nonMaxSuppressionGeneric(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize, Nd4jLong nonMaxSuppressionGeneric(nd4j::LaunchContext* context, NDArray* boxes, NDArray* scores, int maxSize,
double overlapThreshold, double scoreThreshold, NDArray* output); double overlapThreshold, double scoreThreshold, NDArray* output);

158
libnd4j/include/templatemath.h
View File

@ -1446,35 +1446,63 @@ inline __device__ unsigned char nd4j_atomicMul<unsigned char>(unsigned char* add
return (uint8_t)old; return (uint8_t)old;
} }
template <> template <typename T>
inline __device__ int16_t nd4j_atomicMul<int16_t>(int16_t* address, int16_t val) { static inline __device__ T internal_16bit_atomicMul(T* address, int16_t val) {
size_t shift = ((size_t)address & 2); size_t shift = ((size_t)address & 2);
int *base_address = (int *)((char*)address - shift); int *base_address = (int *)((char*)address - shift);
int old = val, assumed;
//printf("%u %x", *base_address); union I16PAIR {
struct {
T H;
T L;
} B;
int W;
__host__ __device__
I16PAIR() {};
__host__ __device__
~I16PAIR() {};
};
I16PAIR pairNew, pairOld, pairAssumed;
pairOld.W = (int) val;
if (reinterpret_cast<int*>(address) == base_address) {
do { do {
assumed = old; pairNew.B.L = pairOld.B.L;
old = atomicCAS(base_address, assumed, (old * val)); pairNew.B.H = pairOld.B.H * val;
} while (assumed != old); pairAssumed.W = pairOld.W;
return (int16_t)old; pairOld.W = atomicCAS(base_address, pairAssumed.W, pairNew.W);
} while (pairAssumed.W != pairOld.W);
return (T) pairOld.B.H;
} else {
do {
pairNew.B.H = pairOld.B.H;
pairNew.B.L = pairOld.B.L * val;
pairAssumed.W = pairOld.W;
pairOld.W = atomicCAS(base_address, pairAssumed.W, pairNew.W);
} while (pairAssumed.W != pairOld.W);
return (T) pairOld.B.L;
}
}
template <>
inline __device__ int16_t nd4j_atomicMul<int16_t>(int16_t* address, int16_t val) {
return internal_16bit_atomicMul<int16_t>(address, val);
} }
template <> template <>
inline __device__ uint16_t nd4j_atomicMul<uint16_t>(uint16_t* address, uint16_t val) { inline __device__ uint16_t nd4j_atomicMul<uint16_t>(uint16_t* address, uint16_t val) {
size_t shift = ((size_t)address & 2); return internal_16bit_atomicMul<uint16_t>(address, val);
unsigned int *base_address = (unsigned int *)((char*)address - shift);
unsigned int old = val, assumed;
//printf("%u %x", *base_address);
do {
assumed = old;
old = atomicCAS(base_address, assumed, (old * val));
} while (assumed != old);
return (uint16_t)old;
} }
template <> template <>
@ -1547,105 +1575,27 @@ inline __device__ Nd4jLong nd4j_atomicMul<Nd4jLong>(Nd4jLong* address, Nd4jLong
template <> template <>
inline __device__ bfloat16 nd4j_atomicMul<bfloat16>(bfloat16* address, bfloat16 val) { inline __device__ bfloat16 nd4j_atomicMul<bfloat16>(bfloat16* address, bfloat16 val) {
auto address_as_ull = (int*) address; return internal_16bit_atomicMul<bfloat16>(address, val);
long addr = (long)(address);
bool misaligned = addr & 0x3;
if (misaligned)
address_as_ull = (int *) (address - 1);
BPAIR old, assumed, fresh;
old.W = *address_as_ull;
do {
if (!misaligned) {
bfloat16 res = old.B.H * val;
fresh.B.H = res;
fresh.B.L = old.B.L;
} else {
bfloat16 res = old.B.L * val;
fresh.B.L = res;
fresh.B.H = old.B.H;
}
assumed.W = old.W;
old.W = atomicCAS(address_as_ull, assumed.W, fresh.W);
} while (assumed.W != old.W);
if (!misaligned) return old.B.H;
else return old.B.L;
} }
template <> template <>
inline __device__ float16 nd4j_atomicMul<float16>(float16* address, float16 val) { inline __device__ float16 nd4j_atomicMul<float16>(float16* address, float16 val) {
auto address_as_ull = (int*) address; return internal_16bit_atomicMul<float16>(address, val);
long addr = (long)(address);
bool misaligned = addr & 0x3;
if (misaligned)
address_as_ull = (int *) (address - 1);
BPAIR old, assumed, fresh;
old.W = *address_as_ull;
do {
if (!misaligned) {
bfloat16 res = old.B.H * val;
fresh.B.H = res;
fresh.B.L = old.B.L;
} else {
bfloat16 res = old.B.L * val;
fresh.B.L = res;
fresh.B.H = old.B.H;
}
assumed.W = old.W;
old.W = atomicCAS(address_as_ull, assumed.W, fresh.W);
} while (assumed.W != old.W);
if (!misaligned) return old.B.H;
else return old.B.L;
} }
template <> template <>
inline __device__ float nd4j_atomicDiv<float>(float* address, float val) { inline __device__ float nd4j_atomicDiv<float>(float* address, float val) {
int* address_as_ull = return nd4j_atomicMul<float>(address, (float) 1.f / val);
(int*)address;
int old = *address_as_ull, assumed;
do {
assumed = old;
old = atomicCAS(address_as_ull, assumed, __float_as_int(__int_as_float(assumed) / val ));
} while (assumed != old);
return __int_as_float(old);
} }
template <> template <>
inline __device__ float16 nd4j_atomicDiv<float16>(float16* address, float16 val) { inline __device__ float16 nd4j_atomicDiv<float16>(float16* address, float16 val) {
int* address_as_ull = return nd4j_atomicMul<float16>(address, (float16) 1.f / val);
(int*)address;
int old = *address_as_ull, assumed;
do {
assumed = old;
old = atomicCAS(address_as_ull, assumed, __float_as_int(val *
__float_as_int(assumed)));
} while (assumed != old);
return __int_as_float(old);
} }
template <> template <>
inline __device__ bfloat16 nd4j_atomicDiv<bfloat16>(bfloat16* address, bfloat16 val) { inline __device__ bfloat16 nd4j_atomicDiv<bfloat16>(bfloat16* address, bfloat16 val) {
int* address_as_ull = return nd4j_atomicMul<bfloat16>(address, (bfloat16) 1.f / val);
(int*)address;
int old = *address_as_ull, assumed;
do {
assumed = old;
old = atomicCAS(address_as_ull, assumed, __float_as_int(val *
__float_as_int(assumed)));
} while (assumed != old);
return __int_as_float(old);
} }
} }
#endif #endif

86
libnd4j/tests_cpu/layers_tests/AtomicTests.cu Normal file
View File

@ -0,0 +1,86 @@
/*******************************************************************************
* Copyright (c) 2019 Konduit K.K.
*
* 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
//
#include "testlayers.h"
#include <ops/declarable/CustomOperations.h>
#include <NDArray.h>
#include <ops/ops.h>
#include <GradCheck.h>
#include <helpers/RandomLauncher.h>
#include <exceptions/cuda_exception.h>
using namespace nd4j;
class AtomicTests : public testing::Test {
public:
AtomicTests() {
//
}
};
template <typename T>
static _CUDA_G void multiplyKernel(void *vbuffer, uint64_t length, void *vresult) {
auto buffer = reinterpret_cast<T*>(vbuffer);
auto result = reinterpret_cast<T*>(vresult);
auto tid = blockIdx.x * blockDim.x + threadIdx.x;
for (auto e = tid; e < length; e += gridDim.x * blockDim.x) {
auto rem = e % 4;
auto i = (e - rem) / 4;
nd4j::math::atomics::nd4j_atomicMul<T>(&result[i], buffer[e]);
}
}
template <typename T>
static void multiplyLauncher(void *vbuffer, uint64_t length, void *vresult) {
multiplyKernel<T><<<256, 256, 1024, *nd4j::LaunchContext::defaultContext()->getCudaStream()>>>(vbuffer, length, vresult);
auto err = cudaStreamSynchronize(*nd4j::LaunchContext::defaultContext()->getCudaStream());
if (err != 0)
nd4j::cuda_exception::build("multiply failed", err);
}
static void multiplyHost(NDArray &input, NDArray &output) {
BUILD_SINGLE_SELECTOR(input.dataType(), multiplyLauncher, (input.specialBuffer(), input.lengthOf(), output.specialBuffer()), NUMERIC_TYPES);
}
TEST_F(AtomicTests, test_multiply) {
std::vector<nd4j::DataType> dtypes = {nd4j::DataType::FLOAT32, nd4j::DataType::DOUBLE, nd4j::DataType::INT16};
for (auto t:dtypes) {
nd4j_printf("Trying data type [%s]\n", DataTypeUtils::asString(t).c_str());
NDArray input('c', {4, 25}, t);
NDArray output('c', {input.lengthOf() / 4}, t);
NDArray exp = output.ulike();
input.assign(2);
output.assign(2);
exp.assign(32);
multiplyHost(input, output);
ASSERT_EQ(exp, output);
}
}

272
libnd4j/tests_cpu/layers_tests/DeclarableOpsTests10.cpp
View File

@ -2330,6 +2330,76 @@ TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressing_5) {
delete results; delete results;
} }
TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressing_6) {
NDArray boxes = NDArrayFactory::create<float16>('c', {3, 4}, {0.8115f, 0.4121f, 0.0771f, 0.4863f,
0.7412f, 0.7607f, 0.1543f, 0.5479f,
0.8223f, 0.2246f, 0.0049f, 0.6465f});
NDArray scales = NDArrayFactory::create<float16>('c', {3}, {0.0029f, 0.8135f, 0.4873f}); //3, 0, 1, 2, 4, 5
NDArray expected = NDArrayFactory::create<int>('c', {2}, {1,2});
NDArray maxSize = NDArrayFactory::create(2);
NDArray threshold = NDArrayFactory::create(0.5f);
NDArray scoreThreshold = NDArrayFactory::create(-DataTypeUtils::infOrMax<float>());
nd4j::ops::non_max_suppression_v3 op;
auto results = op.execute({&boxes, &scales, &maxSize, &threshold, &scoreThreshold}, {}, {});
ASSERT_EQ(Status::OK(), results->status());
NDArray* result = results->at(0);
// result->printBuffer("NonMaxSuppression OUtput6");
// result->printShapeInfo("Ouput6 shape is");
ASSERT_TRUE(expected.isSameShapeStrict(result));
ASSERT_TRUE(expected.equalsTo(result));
delete results;
}
TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressing_06) {
NDArray boxes = NDArrayFactory::create<bfloat16>('c', {3, 4}, {0.8115f, 0.4121f, 0.0771f, 0.4863f,
0.7412f, 0.7607f, 0.1543f, 0.5479f,
0.8223f, 0.2246f, 0.0049f, 0.6465f});
NDArray scales = NDArrayFactory::create<bfloat16>('c', {3}, {0.0029f, 0.8135f, 0.4873f}); //3, 0, 1, 2, 4, 5
NDArray expected = NDArrayFactory::create<int>('c', {2}, {1,2});
NDArray maxSize = NDArrayFactory::create(2);
NDArray threshold = NDArrayFactory::create(0.5f);
NDArray scoreThreshold = NDArrayFactory::create(-DataTypeUtils::infOrMax<float>());
nd4j::ops::non_max_suppression_v3 op;
auto results = op.execute({&boxes, &scales, &maxSize, &threshold, &scoreThreshold}, {}, {});
ASSERT_EQ(Status::OK(), results->status());
NDArray* result = results->at(0);
// result->printBuffer("NonMaxSuppression OUtput06");
// result->printShapeInfo("Ouput06 shape is");
ASSERT_TRUE(expected.isSameShapeStrict(result));
ASSERT_TRUE(expected.equalsTo(result));
delete results;
}
TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressing_7) {
NDArray boxes = NDArrayFactory::create<float>('c', {3, 4}, {0.7788f, 0.8012f, 0.7244f, 0.2329f,
0.7271f, 0.1804f, 0.5056f, 0.8929f,
0.5461f, 0.9234f, 0.0856f, 0.7938f});
NDArray scales = NDArrayFactory::create<float>('c', {3}, {0.7717f, 0.9281f, 0.9846f}); //3, 0, 1, 2, 4, 5
NDArray maxSize = NDArrayFactory::create(0);
NDArray threshold = NDArrayFactory::create(0.5f);
NDArray scoreThreshold = NDArrayFactory::create(0.5f);
nd4j::ops::non_max_suppression_v3 op;
auto results = op.execute({&boxes, &scales, &maxSize, &threshold, &scoreThreshold}, {}, {});
ASSERT_EQ(Status::OK(), results->status());
NDArray* result = results->at(0);
// result->printBuffer("NonMaxSuppression OUtput7");
// result->printShapeInfo("Ouput6 shape is");
ASSERT_TRUE(result->isEmpty());
delete results;
}
//////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressingOverlap_1) { TEST_F(DeclarableOpsTests10, Image_NonMaxSuppressingOverlap_1) {
@ -2720,22 +2790,22 @@ TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_4) {
NDArray x = NDArrayFactory::create<float>('c', {2,4,5,3}); NDArray x = NDArrayFactory::create<float>('c', {2,4,5,3});
NDArray exp = NDArrayFactory::create<float>('c', {2,4,5,3},{ NDArray exp = NDArrayFactory::create<float>('c', {2,4,5,3},{
1.0588236, 1.9607843, 3.019608, 4.0588236, 5.098039, 6.039216, 7.0588236, 8.039216, 9.058824, 1.0588236f, 1.9607843f, 3.019608f, 4.0588236f, 5.098039f, 6.039216f, 7.0588236f, 8.039216f, 9.058824f,
10.058824, 10.980392, 12.078432, 13.058824, 13.921569, 15.09804, 16.058825, 17.058825, 18.117647, 10.058824f, 10.980392f, 12.078432f, 13.058824f, 13.921569f, 15.09804f, 16.058825f, 17.058825f, 18.117647f,
19.058825, 20., 21.137257, 22.058825, 22.941177, 23.882355, 25.058825, 26.078432, 26.901962, 19.058825f, 20.f, 21.137257f, 22.058825f, 22.941177f, 23.882355f, 25.058825f, 26.078432f, 26.901962f,
28.058825, 29.019608, 29.92157, 31.058825, 31.960785, 32.941177, 34.058823, 35.09804, 35.960785, 28.058825f, 29.019608f, 29.92157f, 31.058825f, 31.960785f, 32.941177f, 34.058823f, 35.09804f, 35.960785f,
37.058823, 38.039215, 38.980392, 40.058823, 40.980392, 42.000004, 43.058826, 43.92157, 45.01961, 37.058823f, 38.039215f, 38.980392f, 40.058823f, 40.980392f, 42.000004f, 43.058826f, 43.92157f, 45.01961f,
45., 47.058823, 48.03922, 45., 50., 51.058826, 45., 50., 54.078434, 45.f, 47.058823f, 48.03922f, 45.f, 50.f, 51.058826f, 45.f, 50.f, 54.078434f,
45., 50., 57.09804, 45., 50., 60.11765, 45., 50., 62.862747, 45.f, 50.f, 57.09804f, 45.f, 50.f, 60.11765f, 45.f, 50.f, 62.862747f,
45., 50., 65.882355, 45., 50., 68.90196, 45., 50., 70., 45.f, 50.f, 65.882355f, 45.f, 50.f, 68.90196f, 45.f, 50.f, 70.f,
45., 50., 70., 45., 50., 70., 45., 50., 70., 45.f, 50.f, 70.f, 45.f, 50.f, 70.f, 45.f, 50.f, 70.f,
45., 50., 70., 45., 50., 70., 45., 50., 70., 45.f, 50.f, 70.f, 45.f, 50.f, 70.f, 45.f, 50.f, 70.f,
45., 50., 70., 45., 50., 70., 45., 50., 70., 45.f, 50.f, 70.f, 45.f, 50.f, 70.f, 45.f, 50.f, 70.f,
45., 50., 70., 45., 50., 70., 45., 50., 70., 45.f, 50.f, 70.f, 45.f, 50.f, 70.f, 45.f, 50.f, 70.f,
45., 50., 70., 45., 50., 70., 45., 50., 70., 45.f, 50.f, 70.f, 45.f, 50.f, 70.f, 45.f, 50.f, 70.f,
45., 50., 70.}); 45.f, 50.f, 70.f});
NDArray min = NDArrayFactory::create<float>({20., 20., 20.}); NDArray min = NDArrayFactory::create<float>({20.f, 20.f, 20.f});
NDArray max = NDArrayFactory::create<float>({65., 70., 90.}); NDArray max = NDArrayFactory::create<float>({65.f, 70.f, 90.f});
x.linspace(1.); x.linspace(1.);
nd4j::ops::fake_quant_with_min_max_vars_per_channel op; nd4j::ops::fake_quant_with_min_max_vars_per_channel op;
auto results = op.execute({&x, &min, &max}, {}, {}); auto results = op.execute({&x, &min, &max}, {}, {});
@ -2756,36 +2826,36 @@ TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_4) {
TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_5) { TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_5) {
NDArray x = NDArrayFactory::create<float>('c', {2, 3, 5, 4}); NDArray x = NDArrayFactory::create<float>('c', {2, 3, 5, 4});
NDArray exp = NDArrayFactory::create<float>('c', {2, 3, 5, 4},{ NDArray exp = NDArrayFactory::create<float>('c', {2, 3, 5, 4},{
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-19.92157 , -18.980392 , -18.039217 , -16.941177 , -19.92157f, -18.980392f, -18.039217f, -16.941177f,
-16. , -15.058824 , -13.960785 , -13.0196085 , -16.f, -15.058824f, -13.960785f, -13.0196085f,
-11.92157 , -10.980392 , -10.039217 , -8.941177 , -11.92157f, -10.980392f, -10.039217f, -8.941177f,
-8.000001 , -7.0588236 , -5.960785 , -5.0196085 , -8.000001f, -7.0588236f, -5.960785f, -5.0196085f,
-3.9215698 , -2.9803925 , -2.039217 , -0.94117737, -3.9215698f, -2.9803925f, -2.039217f, -0.94117737f,
0. , 0.94117737, 2.039215 , 2.9803925 , 0.f, 0.94117737f, 2.039215f, 2.9803925f,
4.07843 , 5.0196075 , 5.960783 , 7.0588226 , 4.07843f, 5.0196075f, 5.960783f, 7.0588226f,
8. , 8.941177 , 10.039215 , 10.980392 , 8.f, 8.941177f, 10.039215f, 10.980392f,
12.07843 , 13.019608 , 13.960783 , 15.058823 , 12.07843f, 13.019608f, 13.960783f, 15.058823f,
16. , 16.941177 , 18.039217 , 18.980392 , 16.f, 16.941177f, 18.039217f, 18.980392f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 , 20.07843f, 21.019608f, 21.960783f, 23.058823f,
20.07843 , 21.019608 , 21.960783 , 23.058823 20.07843f, 21.019608f, 21.960783f, 23.058823f
}); });
NDArray min = NDArrayFactory::create<float>({-20., -19., -18., -17}); NDArray min = NDArrayFactory::create<float>({-20., -19., -18., -17});
NDArray max = NDArrayFactory::create<float>({20., 21., 22., 23}); NDArray max = NDArrayFactory::create<float>({20., 21., 22., 23});
@ -2833,9 +2903,10 @@ TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_6) {
// 0.0877f, 0.5966f, 0.6600f, 0.3513f, 0.1604f // 0.0877f, 0.5966f, 0.6600f, 0.3513f, 0.1604f
// }); // });
NDArray exp = NDArrayFactory::create<float>('c', {3,5}, {0.77700233, 0.596913, 0.72314, 0.23104, 0.50982356, NDArray exp = NDArrayFactory::create<float>('c', {3,5}, {
0.17930824, 0.50528157, 0.86846, 0.34995764, 0.50982356, 0.77700233f, 0.596913f, 0.72314f, 0.23104f, 0.50982356f,
0.08735529, 0.596913, 0.6574, 0.34995764, 0.15974471}); 0.17930824f, 0.50528157f, 0.86846f, 0.34995764f, 0.50982356f,
0.08735529f, 0.596913f, 0.6574f, 0.34995764f, 0.15974471f});
NDArray min = NDArrayFactory::create<float>('c', {5}, {-0.2283f, -0.0719f, -0.0154f, -0.5162f, -0.3567f}); NDArray min = NDArrayFactory::create<float>('c', {5}, {-0.2283f, -0.0719f, -0.0154f, -0.5162f, -0.3567f});
NDArray max = NDArrayFactory::create<float>('c', {5}, {0.9441f, 0.5957f, 0.8669f, 0.3502f, 0.5100f}); NDArray max = NDArrayFactory::create<float>('c', {5}, {0.9441f, 0.5957f, 0.8669f, 0.3502f, 0.5100f});
// x.linspace(-60.); // x.linspace(-60.);
@ -2856,45 +2927,74 @@ TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_6) {
delete results; delete results;
} }
//////////////////////////////////////////////////////////////////// //////////////////////////////////////////////////////////////////
//TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_7) { TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_7) {
//
// NDArray x = NDArrayFactory::create<double>('c', {100}); NDArray x = NDArrayFactory::create<float>('c', {100});
// NDArray exp = NDArrayFactory::create<double>('c', {100}, { NDArray exp = NDArrayFactory::create<float>('c', {100}, {
// 0.f, 0.f, 0.f , 0.f , 0.06666667f, 0.06666667f , 0.f, 0.01176471f, 0.01960784f, 0.03137255f, 0.03921569f,
// 0.06666667, 0.06666667, 0.06666667, 0.06666667, 0.06666667, 0.13333334 , 0.0509804f, 0.05882353f, 0.07058824f, 0.07843138f, 0.09019608f,
// 0.13333334, 0.13333334, 0.13333334, 0.13333334, 0.13333334, 0.20000002 , 0.09803922f, 0.10980393f, 0.12156864f, 0.12941177f, 0.14117648f,
// 0.20000002, 0.20000002, 0.20000002, 0.20000002, 0.20000002, 0.20000002 , 0.14901961f, 0.16078432f, 0.16862746f, 0.18039216f, 0.18823531f,
// 0.26666668, 0.26666668, 0.26666668, 0.26666668, 0.26666668, 0.26666668 , 0.20000002f, 0.21176472f, 0.21960786f, 0.23137257f, 0.2392157f,
// 0.26666668, 0.33333334, 0.33333334, 0.33333334, 0.33333334, 0.33333334 , 0.2509804f, 0.25882354f, 0.27058825f, 0.2784314f, 0.2901961f,
// 0.33333334, 0.40000004, 0.40000004, 0.40000004, 0.40000004, 0.40000004 , 0.3019608f, 0.30980393f, 0.32156864f, 0.32941177f, 0.34117648f,
// 0.40000004, 0.40000004, 0.4666667 , 0.4666667 , 0.4666667 , 0.4666667 , 0.34901962f, 0.36078432f, 0.36862746f, 0.3803922f, 0.38823533f,
// 0.4666667 , 0.4666667 , 0.4666667 , 0.53333336, 0.53333336, 0.53333336 , 0.40000004f, 0.41176474f, 0.41960788f, 0.43137258f, 0.43921572f,
// 0.53333336, 0.53333336, 0.53333336, 0.6 , 0.6 , 0.6 , 0.45098042f, 0.45882356f, 0.47058827f, 0.4784314f, 0.4901961f,
// 0.6 , 0.6 , 0.6 , 0.6 , 0.6666667 , 0.6666667 , 0.49803925f, 0.50980395f, 0.52156866f, 0.5294118f, 0.5411765f,
// 0.6666667 , 0.6666667 , 0.6666667 , 0.6666667 , 0.6666667 , 0.73333335 , 0.54901963f, 0.56078434f, 0.5686275f, 0.5803922f, 0.5882353f,
// 0.73333335, 0.73333335, 0.73333335, 0.73333335, 0.73333335, 0.8000001 , 0.6f, 0.6117647f, 0.61960787f, 0.6313726f, 0.6392157f,
// 0.8000001 , 0.8000001 , 0.8000001 , 0.8000001 , 0.8000001 , 0.8000001 , 0.6509804f, 0.65882355f, 0.67058825f, 0.6784314f, 0.6901961f,
// 0.86666673, 0.86666673, 0.86666673, 0.86666673, 0.86666673, 0.86666673 , 0.69803923f, 0.70980394f, 0.72156864f, 0.7294118f, 0.7411765f,
// 0.86666673, 0.9333334 , 0.9333334 , 0.9333334 , 0.9333334 , 0.9333334 , 0.7490196f, 0.7607844f, 0.7686275f, 0.7803922f, 0.78823537f,
// 0.9333334 , 1., 1., 1., 0.8000001f, 0.8117648f, 0.8196079f, 0.8313726f, 0.83921576f,
// }); 0.85098046f, 0.8588236f, 0.8705883f, 0.87843144f, 0.89019614f,
// NDArray min = NDArrayFactory::create<float>('c', {1},{0.0f}); 0.8980393f, 0.909804f, 0.9215687f, 0.9294118f, 0.94117653f,
// NDArray max = NDArrayFactory::create<float>('c', {1}, {1.f}); 0.9490197f, 0.9607844f, 0.9686275f, 0.9803922f, 0.98823535f
// x.linspace(0., 0.01); });
// nd4j::ops::fake_quant_with_min_max_vars op; NDArray min = NDArrayFactory::create<float>('c', {1},{0.0f});
// auto results = op.execute({&x, &min, &max}, {}, {}); NDArray max = NDArrayFactory::create<float>('c', {1}, {1.f});
// x.linspace(0., 0.01);
// ASSERT_EQ(ND4J_STATUS_OK, results->status()); nd4j::ops::fake_quant_with_min_max_vars op;
// auto results = op.execute({&x, &min, &max}, {}, {});
// auto result = results->at(0);
ASSERT_EQ(ND4J_STATUS_OK, results->status());
auto result = results->at(0);
// result->printBuffer("Quantized7"); // result->printBuffer("Quantized7");
// exp.printBuffer("Expected 7"); // exp.printBuffer("Expected 7");
// ASSERT_TRUE(exp.isSameShapeStrict(result)); ASSERT_TRUE(exp.isSameShapeStrict(result));
// ASSERT_TRUE(exp.equalsTo(result)); ASSERT_TRUE(exp.equalsTo(result));
//
// delete results; delete results;
//} }
//////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests10, FakeQuantWithMinMaxVars_Test_8) {
NDArray x = NDArrayFactory::create<float>('c', {10});
NDArray exp = NDArrayFactory::create<float>('c', {10}, {
0.f, 0.09803922f, 0.20000002f, 0.3019608f, 0.40000004f, 0.49803925f,
0.6f, 0.69803923f, 0.8000001f, 0.8980393f
});
NDArray min = NDArrayFactory::create<float>('c', {1},{0.0f});
NDArray max = NDArrayFactory::create<float>('c', {1}, {1.f});
x.linspace(0., 0.1);
nd4j::ops::fake_quant_with_min_max_vars op;
auto results = op.execute({&x, &min, &max}, {}, {});
ASSERT_EQ(ND4J_STATUS_OK, results->status());
auto result = results->at(0);
// x.printBuffer("SourInput8");
// result->printBuffer("Quantized8");
// exp.printBuffer("Expected 8");
ASSERT_TRUE(exp.isSameShapeStrict(result));
ASSERT_TRUE(exp.equalsTo(result));
delete results;
}
//////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests10, batchnorm_test1) { TEST_F(DeclarableOpsTests10, batchnorm_test1) {

2
libnd4j/tests_cpu/layers_tests/DeclarableOpsTests6.cpp
View File

@ -1523,14 +1523,12 @@ TEST_F(DeclarableOpsTests6, LogDet_1) {
auto x = NDArrayFactory::create<double>('c', {2, 3, 3}, {4,12,-16,12,37,-43,-16,-43,98, 4,1.2,-1.6,1.2,3.7,-4.3,-1.6,-4.3,9.8}); auto x = NDArrayFactory::create<double>('c', {2, 3, 3}, {4,12,-16,12,37,-43,-16,-43,98, 4,1.2,-1.6,1.2,3.7,-4.3,-1.6,-4.3,9.8});
auto exp = NDArrayFactory::create<double>({ 3.5835189, 4.159008}); auto exp = NDArrayFactory::create<double>({ 3.5835189, 4.159008});
//x.printIndexedBuffer("Input");
nd4j::ops::logdet op; nd4j::ops::logdet op;
auto result = op.execute({&x}, {}, {}); auto result = op.execute({&x}, {}, {});
ASSERT_EQ(ND4J_STATUS_OK, result->status()); ASSERT_EQ(ND4J_STATUS_OK, result->status());
auto z = result->at(0); auto z = result->at(0);
ASSERT_TRUE(exp.isSameShape(z)); ASSERT_TRUE(exp.isSameShape(z));
ASSERT_TRUE(exp.equalsTo(z)); ASSERT_TRUE(exp.equalsTo(z));

93
libnd4j/tests_cpu/layers_tests/DeclarableOpsTests7.cpp
View File

@ -1813,8 +1813,8 @@ TEST_F(DeclarableOpsTests7, TestUnsortedSegmentSqrtN_5) {
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestSegmentSum_1) { TEST_F(DeclarableOpsTests7, TestSegmentSum_1) {
auto x = NDArrayFactory::create<double>({1.8, 2.5,4., 9., 2.1, 2.4,3.,9., 2.1, 2.1,0.7, 0.1, 3., 4.2, 2.2, 1. }); auto x = NDArrayFactory::create<double>({1.8, 2.5, 4., 9., 2.1, 2.4, 3., 9., 2.1, 2.1, 0.7, 0.1, 3., 4.2, 2.2, 1. });
auto idx = NDArrayFactory::create<int>({0, 0, 1, 1, 1, 1, 2, 3, 3, 3, 4, 4, 4, 4, 4, 4}); auto idx = NDArrayFactory::create<int>({ 0, 0, 1, 1, 1, 1, 2, 3, 3, 3, 4, 4, 4, 4, 4, 4});
auto exp = NDArrayFactory::create<double>({4.3, 17.5, 3., 13.2, 11.2}); auto exp = NDArrayFactory::create<double>({4.3, 17.5, 3., 13.2, 11.2});
nd4j::ops::segment_sum op; nd4j::ops::segment_sum op;
@ -2231,6 +2231,27 @@ TEST_F(DeclarableOpsTests7, TestSegmentProd_05) {
auto result = op.execute({&x, &idx}, {}, {}); auto result = op.execute({&x, &idx}, {}, {});
ASSERT_EQ(result->status(), Status::OK()); ASSERT_EQ(result->status(), Status::OK());
auto res = result->at(0);
// res->printIndexedBuffer("Segment prod 05");
ASSERT_TRUE(exp.equalsTo(result->at(0)));
delete result;
}
////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestSegmentProd_05_1) {
auto x = NDArrayFactory::create<int>({1,2,3,4,5,6,7,8 });
// ----------------------------------------------------------------
auto idx = NDArrayFactory::create<int>({0,0,1,2,2,2,3,3});
auto exp = NDArrayFactory::create<int>({ 2, 3, 120, 56});
nd4j::ops::segment_prod op;
auto result = op.execute({&x, &idx}, {}, {});
ASSERT_EQ(result->status(), Status::OK());
auto res = result->at(0);
// res->printIndexedBuffer("Segment prod 05_1");
ASSERT_TRUE(exp.equalsTo(result->at(0))); ASSERT_TRUE(exp.equalsTo(result->at(0)));
delete result; delete result;
@ -2270,6 +2291,23 @@ TEST_F(DeclarableOpsTests7, TestSegmentProd_07) {
delete result; delete result;
} }
////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestSegmentProd_08) {
auto x = NDArrayFactory::create<int>({'\x1','\x2','\x3','\x4','\x5','\x6','\x7','\x8', '\x9', '\xA' });
// ----------------------------------------------------------------
auto idx = NDArrayFactory::create<int>({0,0,2,2,2,2,3,3,3,3});
auto exp = NDArrayFactory::create<int>({ 2, 1,360, 5040});
nd4j::ops::segment_prod op;
auto result = op.execute({&x, &idx}, {}, {});
ASSERT_EQ(result->status(), Status::OK());
ASSERT_TRUE(exp.equalsTo(result->at(0)));
delete result;
}
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_1) { TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_1) {
auto x = NDArrayFactory::create<double>({1.8, 2.5,4., 9., 2.1, 2.4,3.,9., 2.1, 2.1,0.7, 0.1, 3., 4.2, 2.2, 1.}); auto x = NDArrayFactory::create<double>({1.8, 2.5,4., 9., 2.1, 2.4,3.,9., 2.1, 2.1,0.7, 0.1, 3., 4.2, 2.2, 1.});
@ -2341,6 +2379,22 @@ TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_12) {
delete result; delete result;
} }
////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_08) {
auto x = NDArrayFactory::create<int>({'\x1','\x2','\x3','\x4','\x5','\x6','\x7','\x8', '\x9', '\xA' });
// ----------------------------------------------------------------
auto idx = NDArrayFactory::create<int>({0,0,2,2,2,2,3,3,3,3});
auto exp = NDArrayFactory::create<int>({ 2, 1,360, 5040});
nd4j::ops::unsorted_segment_prod op;
auto result = op.execute({&x, &idx}, {}, {4});
ASSERT_EQ(result->status(), Status::OK());
ASSERT_TRUE(exp.equalsTo(result->at(0)));
delete result;
}
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_3) { TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_3) {
@ -2401,6 +2455,41 @@ TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_4) {
delete result; delete result;
} }
////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProd_5) {
auto x = NDArrayFactory::create<double>('c', {8, 15});
// ----------------------------------------------------------------
auto idx = NDArrayFactory::create<int>({3, 1, 2, 1, 2, 3, 2, 1});
auto exp = NDArrayFactory::create<double>('c', {4, 15}, {
1., 1., 1., 1., 1.,
1., 1., 1., 1., 1.,
1., 1., 1., 1., 1.,
78016., 85493., 93312., 101479., 110000.,
118881., 128128., 137747., 147744., 158125.,
168896., 180063., 191632., 203609., 216000.,
172081., 182528., 193347., 204544., 216125.,
228096., 240463., 253232., 266409., 280000.,
294011., 308448., 323317., 338624., 354375.,
76., 154., 234., 316., 400.,
486., 574., 664., 756., 850.,
946., 1044., 1144., 1246., 1350.});
x.linspace(1.);
nd4j::ops::unsorted_segment_prod op;
auto result = op.execute({&x, &idx}, {}, {4});
ASSERT_EQ(result->status(), Status::OK());
//result->at(0)->printIndexedBuffer("Output");
// result->at(0)->printShapeInfo("Out Shape");
//exp.printIndexedBuffer("Expect");
// exp.printShapeInfo("Exp Shape");
ASSERT_TRUE(exp.equalsTo(result->at(0)));
delete result;
}
//////////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////////////////////
TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProdBP_4) { TEST_F(DeclarableOpsTests7, TestUnsortedSegmentProdBP_4) {
auto x = NDArrayFactory::create<double>('c', {8}, { auto x = NDArrayFactory::create<double>('c', {8}, {

4
libnd4j/tests_cpu/layers_tests/WorkspaceTests.cu
View File

@ -39,7 +39,7 @@ TEST_F(CudaWorkspaceTests, Basic_Tests_1) {
ctx.setWorkspace(&workspace); ctx.setWorkspace(&workspace);
auto array = NDArrayFactory::create<float>('c', {5, 5}, &ctx); auto array = NDArrayFactory::create<float>('c', {5, 5}, &ctx);
ASSERT_EQ(100, workspace.getCurrentOffset()); ASSERT_EQ(108, workspace.getCurrentOffset());
ASSERT_EQ(0, workspace.getCurrentSecondaryOffset()); ASSERT_EQ(0, workspace.getCurrentSecondaryOffset());
array.e<int>(0); array.e<int>(0);
@ -55,6 +55,6 @@ TEST_F(CudaWorkspaceTests, Basic_Tests_2) {
ctx.setWorkspace(&workspace); ctx.setWorkspace(&workspace);
auto array = NDArrayFactory::create<float>('c', {5, 5}, {0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f}, &ctx); auto array = NDArrayFactory::create<float>('c', {5, 5}, {0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f, 0.0f}, &ctx);
ASSERT_EQ(100, workspace.getCurrentOffset()); ASSERT_EQ(108, workspace.getCurrentOffset());
ASSERT_EQ(0, workspace.getCurrentSecondaryOffset()); ASSERT_EQ(0, workspace.getCurrentSecondaryOffset());
} }

1
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/imports/converters/ImportClassMapping.java
View File

@ -86,6 +86,7 @@ public class ImportClassMapping {
org.nd4j.linalg.api.ops.impl.image.CropAndResize.class, org.nd4j.linalg.api.ops.impl.image.CropAndResize.class,
org.nd4j.linalg.api.ops.impl.image.ExtractImagePatches.class, org.nd4j.linalg.api.ops.impl.image.ExtractImagePatches.class,
org.nd4j.linalg.api.ops.impl.image.NonMaxSuppression.class, org.nd4j.linalg.api.ops.impl.image.NonMaxSuppression.class,
org.nd4j.linalg.api.ops.impl.image.NonMaxSuppressionV3.class,
org.nd4j.linalg.api.ops.impl.image.ResizeBilinear.class, org.nd4j.linalg.api.ops.impl.image.ResizeBilinear.class,
org.nd4j.linalg.api.ops.impl.image.ResizeNearestNeighbor.class, org.nd4j.linalg.api.ops.impl.image.ResizeNearestNeighbor.class,
org.nd4j.linalg.api.ops.impl.indexaccum.FirstIndex.class, org.nd4j.linalg.api.ops.impl.indexaccum.FirstIndex.class,

2
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/image/NonMaxSuppression.java
View File

@ -53,7 +53,7 @@ public class NonMaxSuppression extends DynamicCustomOp {
@Override @Override
public String[] tensorflowNames() { public String[] tensorflowNames() {
return new String[]{"NonMaxSuppression", "NonMaxSuppressionV2", "NonMaxSuppressionV3"}; return new String[]{"NonMaxSuppression", "NonMaxSuppressionV2"};
} }
@Override @Override

79
nd4j/nd4j-backends/nd4j-api-parent/nd4j-api/src/main/java/org/nd4j/linalg/api/ops/impl/image/NonMaxSuppressionV3.java Normal file
View File

@ -0,0 +1,79 @@
/*******************************************************************************
* 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
******************************************************************************/
package org.nd4j.linalg.api.ops.impl.image;
import lombok.NonNull;
import org.nd4j.autodiff.samediff.SDVariable;
import org.nd4j.autodiff.samediff.SameDiff;
import org.nd4j.imports.NoOpNameFoundException;
import org.nd4j.linalg.api.buffer.DataType;
import org.nd4j.linalg.api.ops.DynamicCustomOp;
import org.nd4j.linalg.api.ops.Op;
import java.util.Collections;
import java.util.List;
/**
* Non max suppression
*
* @author raver119@gmail.com
*/
public class NonMaxSuppressionV3 extends DynamicCustomOp {
public NonMaxSuppressionV3() {}
public NonMaxSuppressionV3(SameDiff sameDiff, @NonNull SDVariable boxes, @NonNull SDVariable scores, @NonNull SDVariable maxOutSize,
@NonNull SDVariable iouThreshold, @NonNull SDVariable scoreThreshold) {
super(null, sameDiff, new SDVariable[]{boxes, scores, maxOutSize, iouThreshold, scoreThreshold}, false);
}
@Override
public String onnxName() {
throw new NoOpNameFoundException("No onnx name found for shape " + opName());
}
@Override
public String tensorflowName() {
return "NonMaxSuppressionV3";
}
@Override
public String[] tensorflowNames() {
return new String[]{"NonMaxSuppressionV3","NonMaxSuppressionV4"};
}
@Override
public String opName() {
return "non_max_suppression_v3";
}
@Override
public Op.Type opType() {
return Op.Type.CUSTOM;
}
@Override
public List<SDVariable> doDiff(List<SDVariable> i_v) {
return Collections.singletonList(sameDiff.zerosLike(arg()));
}
@Override
public List<DataType> calculateOutputDataTypes(List<DataType> inputDataTypes){
//Always 1D integer tensor (indices)
return Collections.singletonList(DataType.INT);
}
}

19
nd4j/nd4j-backends/nd4j-backend-impls/nd4j-native/src/main/java/org/nd4j/nativeblas/Nd4jCpu.java
View File

@ -20827,7 +20827,7 @@ public static final int TAD_THRESHOLD = TAD_THRESHOLD();
// #endif // #endif
/** /**
* image.non_max_suppression op. * image.non_max_suppression ops.
* input: * input:
* 0 - boxes - 2D-tensor with shape (num_boxes, 4) by float type * 0 - boxes - 2D-tensor with shape (num_boxes, 4) by float type
* 1 - scales - 1D-tensor with shape (num_boxes) by float type * 1 - scales - 1D-tensor with shape (num_boxes) by float type
@ -20859,6 +20859,23 @@ public static final int TAD_THRESHOLD = TAD_THRESHOLD();
public native ShapeList calculateOutputShape(ShapeList inputShape, @ByRef Context block); public native ShapeList calculateOutputShape(ShapeList inputShape, @ByRef Context block);
} }
// #endif // #endif
// #if NOT_EXCLUDED(OP_image_non_max_suppression_v3)
@Namespace("nd4j::ops") public static class non_max_suppression_v3 extends DeclarableCustomOp {
static { Loader.load(); }
/** Pointer cast constructor. Invokes {@link Pointer#Pointer(Pointer)}. */
public non_max_suppression_v3(Pointer p) { super(p); }
/** Native array allocator. Access with {@link Pointer#position(long)}. */
public non_max_suppression_v3(long size) { super((Pointer)null); allocateArray(size); }
private native void allocateArray(long size);
@Override public non_max_suppression_v3 position(long position) {
return (non_max_suppression_v3)super.position(position);
}
public non_max_suppression_v3() { super((Pointer)null); allocate(); }
private native void allocate();
public native ShapeList calculateOutputShape(ShapeList inputShape, @ByRef Context block);
}
// #endif
/* /*
* image.non_max_suppression_overlaps op. * image.non_max_suppression_overlaps op.