170 static constexpr int32_t kVALUE = 14;
208 mImpl->setName(name);
220 return mImpl->getName();
239 mImpl->setDimensions(dimensions);
253 return mImpl->getDimensions();
289 mImpl->setType(type);
304 return mImpl->getType();
321 return mImpl->setDynamicRange(min, max);
329 return mImpl->isNetworkInput();
337 return mImpl->isNetworkOutput();
354 mImpl->setBroadcastAcrossBatch(broadcastAcrossBatch);
368 return mImpl->getBroadcastAcrossBatch();
380 return mImpl->getLocation();
399 mImpl->setLocation(location);
411 return mImpl->dynamicRangeIsSet();
419 mImpl->resetDynamicRange();
429 return mImpl->getDynamicRangeMin();
439 return mImpl->getDynamicRangeMax();
461 mImpl->setAllowedFormats(formats);
474 return mImpl->getAllowedFormats();
505 return mImpl->isShapeTensor();
526 return mImpl->isExecutionTensor();
552 mImpl->setDimensionName(index, name);
567 return mImpl->getDimensionName(index);
592 return mLayer->getType();
606 mLayer->setName(name);
616 return mLayer->getName();
624 return mLayer->getNbInputs();
637 return mLayer->getInput(index);
645 return mLayer->getNbOutputs();
655 return mLayer->getOutput(index);
672 return mLayer->setInput(index, tensor);
705 mLayer->setPrecision(dataType);
717 return mLayer->getPrecision();
731 return mLayer->precisionIsSet();
743 mLayer->resetPrecision();
793 mLayer->setOutputType(index, dataType);
808 return mLayer->getOutputType(index);
824 return mLayer->outputTypeIsSet(index);
838 return mLayer->resetOutputType(index);
856 mLayer->setMetadata(metadata);
869 return mLayer->getMetadata();
890 return mLayer->setNbRanks(nbRanks);
902 return mLayer->getNbRanks();
907 apiv::VLayer* mLayer;
1084 static constexpr int32_t kVALUE = 4;
1112 mImpl->setNbOutputMaps(nbOutputMaps);
1122 return mImpl->getNbOutputMaps();
1142 mImpl->setNbGroups(nbGroups);
1152 return mImpl->getNbGroups();
1166 mImpl->setKernelWeights(weights);
1176 return mImpl->getKernelWeights();
1191 mImpl->setBiasWeights(weights);
1201 return mImpl->getBiasWeights();
1218 mImpl->setPrePadding(padding);
1228 return mImpl->getPrePadding();
1245 mImpl->setPostPadding(padding);
1255 return mImpl->getPostPadding();
1269 mImpl->setPaddingMode(paddingMode);
1281 return mImpl->getPaddingMode();
1294 mImpl->setKernelSizeNd(kernelSize);
1304 return mImpl->getKernelSizeNd();
1319 mImpl->setStrideNd(stride);
1329 return mImpl->getStrideNd();
1347 mImpl->setPaddingNd(padding);
1359 return mImpl->getPaddingNd();
1373 mImpl->setDilationNd(dilation);
1383 return mImpl->getDilationNd();
1432 mImpl->setActivationType(type);
1442 return mImpl->getActivationType();
1457 mImpl->setAlpha(alpha);
1471 mImpl->setBeta(beta);
1480 return mImpl->getAlpha();
1489 return mImpl->getBeta();
1519 static constexpr int32_t kVALUE = 3;
1546 mImpl->setPoolingType(type);
1556 return mImpl->getPoolingType();
1571 mImpl->setBlendFactor(blendFactor);
1584 return mImpl->getBlendFactor();
1598 mImpl->setAverageCountExcludesPadding(exclusive);
1609 return mImpl->getAverageCountExcludesPadding();
1627 mImpl->setPrePadding(padding);
1637 return mImpl->getPrePadding();
1655 mImpl->setPostPadding(padding);
1665 return mImpl->getPostPadding();
1678 mImpl->setPaddingMode(paddingMode);
1689 return mImpl->getPaddingMode();
1702 mImpl->setWindowSizeNd(windowSize);
1712 return mImpl->getWindowSizeNd();
1727 mImpl->setStrideNd(stride);
1737 return mImpl->getStrideNd();
1756 mImpl->setPaddingNd(padding);
1768 return mImpl->getPaddingNd();
1799 mImpl->setWindowSize(windowSize);
1809 return mImpl->getWindowSize();
1821 mImpl->setAlpha(alpha);
1831 return mImpl->getAlpha();
1843 mImpl->setBeta(beta);
1853 return mImpl->getBeta();
1875 return mImpl->getK();
1941 mImpl->setMode(mode);
1951 return mImpl->getMode();
1961 mImpl->setShift(shift);
1971 return mImpl->getShift();
1981 mImpl->setScale(scale);
1991 return mImpl->getScale();
2001 mImpl->setPower(power);
2011 return mImpl->getPower();
2026 return mImpl->getChannelAxis();
2047 mImpl->setChannelAxis(channelAxis);
2100 mImpl->setAxes(axes);
2110 return mImpl->getAxes();
2146 mImpl->setAxis(axis);
2156 return mImpl->getAxis();
2183 mImpl->setNbOutputMaps(nbOutputMaps);
2193 return mImpl->getNbOutputMaps();
2213 mImpl->setNbGroups(nbGroups);
2223 return mImpl->getNbGroups();
2237 mImpl->setKernelWeights(weights);
2247 return mImpl->getKernelWeights();
2262 mImpl->setBiasWeights(weights);
2272 return mImpl->getBiasWeights();
2289 mImpl->setPrePadding(padding);
2299 return mImpl->getPrePadding();
2316 mImpl->setPostPadding(padding);
2326 return mImpl->getPostPadding();
2340 mImpl->setPaddingMode(paddingMode);
2352 return mImpl->getPaddingMode();
2367 mImpl->setKernelSizeNd(kernelSize);
2377 return mImpl->getKernelSizeNd();
2394 mImpl->setStrideNd(stride);
2404 return mImpl->getStrideNd();
2422 mImpl->setPaddingNd(padding);
2434 return mImpl->getPaddingNd();
2460 mImpl->setDilationNd(dilation);
2470 return mImpl->getDilationNd();
2518 static constexpr int32_t kVALUE = 14;
2555 return mImpl->setOperation(op);
2567 return mImpl->getOperation();
2688 mImpl->setGatherAxis(axis);
2700 return mImpl->getGatherAxis();
2723 mImpl->setNbElementWiseDims(elementWiseDims);
2733 return mImpl->getNbElementWiseDims();
2743 mImpl->setMode(mode);
2753 return mImpl->getMode();
2782 return mImpl->getPlugin();
2809 return mImpl->getPlugin();
2892 mImpl->setOperation(op);
2902 return mImpl->getOperation();
2978 static constexpr int32_t kVALUE = 5;
2998 mImpl->setOperation(op);
3008 return mImpl->getOperation();
3018 mImpl->setReduceAxes(reduceAxes);
3028 return mImpl->getReduceAxes();
3038 mImpl->setKeepDimensions(keepDimensions);
3048 return mImpl->getKeepDimensions();
3082 mImpl->setPrePaddingNd(padding);
3094 return mImpl->getPrePaddingNd();
3108 mImpl->setPostPaddingNd(padding);
3120 return mImpl->getPostPaddingNd();
3170 mImpl->setFirstTranspose(permutation);
3182 return mImpl->getFirstTranspose();
3210 mImpl->setReshapeDimensions(dimensions);
3223 return mImpl->getReshapeDimensions();
3270 mImpl->setSecondTranspose(permutation);
3282 return mImpl->getSecondTranspose();
3298 return mImpl->setZeroIsPlaceholder(zeroIsPlaceholder);
3311 return mImpl->getZeroIsPlaceholder();
3422 mImpl->setStart(start);
3437 return mImpl->getStart();
3451 return mImpl->setSize(size);
3466 return mImpl->getSize();
3480 mImpl->setStride(stride);
3495 return mImpl->getStride();
3505 mImpl->setMode(mode);
3515 return mImpl->getMode();
3558 mImpl->setAxes(axes);
3573 return mImpl->getAxes();
3643 mImpl->setOperation(op);
3653 return mImpl->getOperation();
3681 return mImpl->getK();
3691 mImpl->setReduceAxes(reduceAxes);
3701 return mImpl->getReduceAxes();
3732 return mImpl->setIndicesType(type);
3744 return mImpl->getIndicesType();
3830 mImpl->setOperation(index, op);
3842 return mImpl->getOperation(index);
3886 return mImpl->setIndicesType(type);
3898 return mImpl->getIndicesType();
3995 mImpl->setToType(toType);
4006 return mImpl->getToType();
4035 mImpl->setWeights(weights);
4045 return mImpl->getWeights();
4057 mImpl->setDimensions(dimensions);
4069 return mImpl->getDimensions();
4115 static constexpr int32_t kVALUE = 3;
4169 static constexpr int32_t kVALUE = 3;
4199 static constexpr int32_t kVALUE = 2;
4235 static constexpr int32_t kVALUE = 4;
4298 return mImpl->setOutputDimensions(dimensions);
4308 return mImpl->getOutputDimensions();
4336 void setScales(
float const* scales, int32_t nbScales)
noexcept
4338 mImpl->setScales(scales, nbScales);
4355 int32_t
getScales(int32_t size,
float* scales)
const noexcept
4357 return mImpl->getScales(size, scales);
4369 mImpl->setResizeMode(interpolationMode);
4379 return mImpl->getResizeMode();
4414 mImpl->setCoordinateTransformation(coordTransform);
4424 return mImpl->getCoordinateTransformation();
4439 mImpl->setSelectorForSinglePixel(selector);
4449 return mImpl->getSelectorForSinglePixel();
4463 mImpl->setNearestRounding(value);
4473 return mImpl->getNearestRounding();
4495 mImpl->setCubicCoeff(A);
4505 return mImpl->getCubicCoeff();
4518 mImpl->setExcludeOutside(excludeFlag);
4528 return mImpl->getExcludeOutside();
4610 return mBoundary->getLoop();
4615 apiv::VLoopBoundaryLayer* mBoundary;
4633 return mBoundary->getConditional();
4638 apiv::VConditionalBoundaryLayer* mBoundary;
4722 return mImpl->setCondition(condition);
4740 return mImpl->addOutput(trueSubgraphOutput, falseSubgraphOutput);
4752 return mImpl->addInput(input);
4767 mImpl->setName(name);
4777 return mImpl->getName();
4847 return mImpl->getLoopOutput();
4864 mImpl->setAxis(axis);
4872 return mImpl->getAxis();
4921 return mImpl->getTripLimit();
4947 mImpl->setAxis(axis);
4955 return mImpl->getAxis();
4969 mImpl->setReverse(reverse);
4979 return mImpl->getReverse();
5008 return mImpl->addRecurrence(initialValue);
5029 return mImpl->addTripLimit(tensor, limit);
5042 return mImpl->addIterator(tensor, axis, reverse);
5055 return mImpl->addLoopOutput(tensor, outputKind, axis);
5070 mImpl->setName(name);
5080 return mImpl->getName();
5135 mImpl->setMessage(message);
5145 return mImpl->getMessage();
5250 mImpl->setDimensions(dimensions);
5265 return mImpl->getDimensions();
5275 mImpl->setOperation(op);
5285 return mImpl->getOperation();
5304 mImpl->setAlpha(alpha);
5319 return mImpl->getAlpha();
5338 mImpl->setBeta(beta);
5353 return mImpl->getBeta();
5414 mImpl->setAlphaInt64(alpha);
5429 return mImpl->getAlphaInt64();
5448 mImpl->setBetaInt64(beta);
5463 return mImpl->getBetaInt64();
5471 return mImpl->isAlphaBetaInt64();
5489 mImpl->setToType(toType);
5501 return mImpl->getToType();
5596 return mImpl->getAxis();
5607 mImpl->setAxis(axis);
5620 return mImpl->setBlockShape(blockShape);
5631 return mImpl->getBlockShape();
5647 mImpl->setToType(toType);
5659 return mImpl->getToType();
5748 return mImpl->getAxis();
5759 mImpl->setAxis(axis);
5776 return mImpl->setBlockShape(blockShape);
5787 return mImpl->getBlockShape();
5803 mImpl->setToType(toType);
5815 return mImpl->getToType();
5870 mImpl->setToType(toType);
5883 return mImpl->getToType();
5896 mImpl->setScaleType(scaleType);
5909 return mImpl->getScaleType();
5922 mImpl->setAxis(axis);
5932 return mImpl->getAxis();
5945 mImpl->setBlockSize(size);
5955 return mImpl->getBlockSize();
5968 mImpl->setBlockShape(blockShape);
5980 return mImpl->getBlockShape();
6036 return mImpl->setEquation(equation);
6046 return mImpl->getEquation();
6145 mImpl->setMode(mode);
6155 return mImpl->getMode();
6165 mImpl->setAxis(axis);
6173 return mImpl->getAxis();
6217 mImpl->setAxis(axis);
6225 return mImpl->getAxis();
6254 mImpl->setInterpolationMode(mode);
6266 return mImpl->getInterpolationMode();
6276 mImpl->setAlignCorners(alignCorners);
6288 return mImpl->getAlignCorners();
6300 return mImpl->setSampleMode(mode);
6312 return mImpl->getSampleMode();
6410 mImpl->setBoundingBoxFormat(fmt);
6422 return mImpl->getBoundingBoxFormat();
6436 mImpl->setTopKBoxLimit(limit);
6446 return mImpl->getTopKBoxLimit();
6481 return mImpl->setIndicesType(type);
6493 return mImpl->getIndicesType();
6526 mImpl->setBatchAxis(batchAxis);
6536 return mImpl->getBatchAxis();
6549 mImpl->setSequenceAxis(sequenceAxis);
6559 return mImpl->getSequenceAxis();
6597 return mImpl->setEpsilon(eps);
6607 return mImpl->getEpsilon();
6617 return mImpl->setAxes(axesMask);
6627 return mImpl->getAxes();
6648 return mImpl->setNbGroups(nbGroups);
6658 return mImpl->getNbGroups();
6686 return mImpl->setComputePrecision(type);
6698 return mImpl->getComputePrecision();
6708 return mImpl->isV2();
6805 static constexpr int32_t kVALUE = 1;
6851 return mImpl->setOperation(op);
6863 return mImpl->getOperation();
6875 mImpl->setExclusive(exclusive);
6887 return mImpl->getExclusive();
6899 mImpl->setReverse(reverse);
6911 return mImpl->getReverse();
6941 static constexpr int32_t kVALUE = 2;
6963 return mBoundary->getAttention();
6968 apiv::VAttentionBoundaryLayer* mBoundary;
7085 return mImpl->setNormalizationOperation(op);
7097 return mImpl->getNormalizationOperation();
7114 return mImpl->setMask(mask);
7126 return mImpl->getMask();
7139 return mImpl->setCausal(isCausal);
7151 return mImpl->getCausal();
7163 return mImpl->setDecomposable(decomposable);
7176 return mImpl->getDecomposable();
7195 return mImpl->setInput(index, input);
7204 return mImpl->getNbInputs();
7216 return mImpl->getInput(index);
7224 return mImpl->getNbOutputs();
7236 return mImpl->getOutput(index);
7253 return mImpl->setName(name);
7265 return mImpl->getName();
7281 return mImpl->setNormalizationQuantizeScale(tensor);
7292 return mImpl->getNormalizationQuantizeScale();
7305 return mImpl->setNormalizationQuantizeToType(type);
7317 return mImpl->getNormalizationQuantizeToType();
7337 return mImpl->setMetadata(metadata);
7350 return mImpl->getMetadata();
7366 return mImpl->setNbRanks(nbRanks);
7378 return mImpl->getNbRanks();
7402 mImpl->setInterleaved(interleaved);
7413 return mImpl->getInterleaved();
7424 return mImpl->setRotaryEmbeddingDim(rotaryEmbeddingDim);
7435 return mImpl->getRotaryEmbeddingDim();
7480 static constexpr int32_t kVALUE = 1;
7530 return mImpl->setCacheMode(cacheMode);
7540 return mImpl->getCacheMode();
7570 static constexpr int32_t kVALUE = 2;
7702 mImpl->setGatedWeights(fcGateWeights, fcUpWeights, fcDownWeights, activationType);
7714 mImpl->setGatedBiases(fcGateBiases, fcUpBiases, fcDownBiases);
7726 mImpl->setActivationType(activationType);
7738 return mImpl->getActivationType();
7761 mImpl->setQuantizationStatic(fcDownActivationScale, dataType);
7790 mImpl->setQuantizationDynamicDblQ(fcDownActivationDblQScale, dataType, blockShape, dynQOutputScaleType);
7805 mImpl->setQuantizationToType(type);
7817 return mImpl->getQuantizationToType();
7833 mImpl->setQuantizationBlockShape(blockShape);
7845 return mImpl->getQuantizationBlockShape();
7857 mImpl->setDynQOutputScaleType(type);
7869 return mImpl->getDynQOutputScaleType();
7890 mImpl->setSwigluParams(limit, alpha, beta);
7904 mImpl->setSwigluParamLimit(limit);
7916 return mImpl->getSwigluParamLimit();
7930 mImpl->setSwigluParamAlpha(alpha);
7942 return mImpl->getSwigluParamAlpha();
7956 mImpl->setSwigluParamBeta(beta);
7968 return mImpl->getSwigluParamBeta();
7985 mImpl->setInput(index, tensor);
8068 return mImpl->addInput(name, type, dimensions);
8082 mImpl->markOutput(tensor);
8100 return mImpl->markDebug(tensor);
8116 return mImpl->unmarkDebug(tensor);
8126 return mImpl->isDebugTensor(tensor);
8148 return mImpl->markUnfusedTensorsAsDebugTensors();
8162 return mImpl->unmarkUnfusedTensorsAsDebugTensors();
8182 return mImpl->addActivation(input, type);
8201 return mImpl->addLRN(input, window, alpha, beta, k);
8227 return mImpl->addScale(input, mode, shift, scale, power);
8240 return mImpl->addSoftMax(input);
8257 return mImpl->addConcatenation(inputs, nbInputs);
8284 return mImpl->addElementWise(input1, input2, op);
8306 return mImpl->addUnary(input, operation);
8320 return mImpl->addShuffle(input);
8337 return mImpl->addOneHot(indices, values, depth, axis);
8349 return mImpl->getNbLayers();
8363 return mImpl->getLayer(index);
8375 return mImpl->getNbInputs();
8391 return mImpl->getInput(index);
8405 return mImpl->getNbOutputs();
8421 return mImpl->getOutput(index);
8448 return mImpl->addReduce(input, operation, reduceAxes, keepDimensions);
8483 return mImpl->addTopK(input, op, k, reduceAxes);
8516 return mImpl->addTopKV2(input, op, k, reduceAxes, indicesType);
8532 return mImpl->addGather(data, indices, axis);
8548 return mImpl->addGatherV2(data, indices, mode);
8567 return mImpl->addRaggedSoftMax(input, bounds);
8589 return mImpl->addMatrixMultiply(input0, op0, input1, op1);
8607 return mImpl->addNonZero(input);
8623 return mImpl->addNonZeroV2(input, indicesType);
8647 return mImpl->addConstant(dimensions, weights);
8661 return mImpl->addIdentity(input);
8676 return mImpl->addCast(input, toType);
8691 mImpl->removeTensor(tensor);
8703 mImpl->unmarkOutput(tensor);
8724 return mImpl->addPluginV2(inputs, nbInputs, plugin);
8741 int32_t nbShapeInputs,
IPluginV3& plugin)
noexcept
8743 return mImpl->addPluginV3(inputs, nbInputs, shapeInputs, nbShapeInputs, plugin);
8762 return mImpl->addSlice(input, start, size, stride);
8786 mImpl->setName(name);
8800 return mImpl->getName();
8816 return mImpl->addShape(input);
8830 return mImpl->hasImplicitBatchDimension();
8840 return mImpl->getFlags();
8852 return mImpl->getFlag(networkDefinitionCreationFlag);
8869 return mImpl->markOutputForShapes(tensor);
8881 return mImpl->unmarkOutputForShapes(tensor);
8899 return mImpl->addParametricReLU(input, slope);
8922 return mImpl->addConvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
8941 return mImpl->addPoolingNd(input, type, windowSize);
8964 return mImpl->addDeconvolutionNd(input, nbOutputMaps, kernelSize, kernelWeights, biasWeights);
9001 return mImpl->addScaleNd(input, mode, shift, scale, power, channelAxis);
9017 return mImpl->addResize(input);
9031 return mImpl->addLoop();
9046 return mImpl->addIfConditional();
9085 return mImpl->addSelect(condition, thenInput, elseInput);
9102 return mImpl->addAssertion(condition, message);
9127 return mImpl->addFill(dimensions, op);
9153 return mImpl->addFillV2(dimensions, op, outputType);
9169 return mImpl->addPaddingNd(input, prePadding, postPadding);
9193 return mImpl->setWeightsName(weights, name);
9212 mImpl->setErrorRecorder(recorder);
9227 return mImpl->getErrorRecorder();
9248 return mImpl->addDequantize(input, scale);
9271 return mImpl->addDequantizeV2(input, scale, outputType);
9291 return mImpl->addScatter(data, indices, updates, mode);
9312 return mImpl->addQuantize(input, scale);
9336 return mImpl->addQuantizeV2(input, scale, outputType);
9364 return mImpl->addDynamicQuantize(input, axis, blockSize, outputType, scaleType);
9388 return mImpl->addDynamicQuantizeV2(input, blockShape, outputType, scaleType);
9403 return mImpl->addEinsum(inputs, nbInputs, equation);
9421 return mImpl->addGridSample(input, grid);
9443 return mImpl->addNMS(boxes, scores, maxOutputBoxesPerClass);
9463 return mImpl->addNMSV2(boxes, scores, maxOutputBoxesPerClass, indicesType);
9480 return mImpl->addReverseSequence(input, sequenceLens);
9512 return mImpl->addNormalization(input, scale, bias, axesMask);
9534 return mImpl->addCumulative(input, axis, operation, exclusive, reverse);
9562 return mImpl->addAttention(query, key, value, normOp, causal);
9586 return mImpl->addRotaryEmbedding(input, cosCache, sinCache, interleaved, rotaryEmbeddingDim);
9621 return mImpl->addKVCacheUpdate(cache, update, writeIndices, cacheMode);
9640 return mImpl->addMoE(hiddenStates, selectedExpertsForTokens, scoresForSelectedExperts);
9667 ReduceOperation reduceOp, int64_t root, int64_t* groups, int64_t groupSize)
noexcept
9669 return mImpl->addDistCollective(input, distCollectiveOp, reduceOp, root, groups, groupSize);
9680 return mImpl->getBuilder();
9693 return mImpl->markWeightsRefittable(name);
9705 return mImpl->unmarkWeightsRefittable(name);
9718 return mImpl->areWeightsMarkedRefittable(name);
9737 return mImpl->addSqueeze(input, axes);
9758 return mImpl->addUnsqueeze(input, axes);
9784 return mImpl->addNormalizationV2(input, scale, bias, axesMask);
9856 virtual
bool getBatch(
void* bindings[],
char const* names[], int32_t nbBindings) noexcept = 0;
9872 virtual
void const* readCalibrationCache(std::
size_t& length) noexcept = 0;
9882 virtual
void writeCalibrationCache(
void const* ptr, std::
size_t length) noexcept = 0;
10048 virtual
double getRegressionCutoff() const noexcept = 0;
10062 virtual
void const* readHistogramCache(std::
size_t& length) noexcept = 0;
10072 virtual
void writeHistogramCache(
void const* ptr, std::
size_t length) noexcept = 0;
10115 return mImpl->getDataType();
10126 return mImpl->getStrides();
10136 return mImpl->getVectorizedDim();
10147 return mImpl->getComponentsPerElement();
10176 return mImpl->getImplementation();
10184 return mImpl->getTactic();
10212 return mImpl->getName();
10224 return mImpl->getDimensions(index, select);
10232 return mImpl->getNbInputs();
10240 return mImpl->getNbOutputs();
10269 return mImpl->getAlgorithmVariant();
10277 return mImpl->getTimingMSec();
10285 return mImpl->getWorkspaceSize();
10299 return mImpl->getAlgorithmIOInfoByIndex(index);
10334 int32_t nbChoices, int32_t* selection)
noexcept = 0;
10347 int32_t nbAlgorithms)
noexcept = 0;
10443 static constexpr int32_t kVALUE = 2;
10641#if ENABLE_FEATURE_DISABLE_RUNTIME_ALLOCATION
10648 kREQUIRE_USER_ALLOCATION = 29,
10661#if ENABLE_FEATURE_DISABLE_RUNTIME_ALLOCATION
10701 static constexpr uint64_t kINVALID_TACTIC_HASH = UINT64_MAX;
10734 return mImpl->serialize();
10758 return mImpl->combine(inputCache, ignoreMismatch);
10768 return mImpl->reset();
10785 int64_t
queryKeys(TimingCacheKey* keyBuffer, int64_t capacity)
const noexcept
10787 return mImpl->queryKeys(keyBuffer, capacity);
10802 TimingCacheValue
query(TimingCacheKey
const& key)
const noexcept
10804 return mImpl->query(key);
10824 bool update(TimingCacheKey
const& key, TimingCacheValue
const& value)
noexcept
10826 return mImpl->update(key, value);
10956 static constexpr int32_t kVALUE = 4;
11008 static constexpr int32_t kVALUE = 3;
11048 static constexpr int32_t kVALUE = 4;
11087 virtual void phaseStart(
char const* phaseName,
char const* parentPhase, int32_t nbSteps)
noexcept = 0;
11160 virtual
void setAvgTimingIterations(int32_t avgTiming) noexcept
11162 mImpl->setAvgTimingIterations(avgTiming);
11174 return mImpl->getAvgTimingIterations();
11187 mImpl->setEngineCapability(capability);
11199 return mImpl->getEngineCapability();
11211 mImpl->setInt8Calibrator(calibrator);
11221 return mImpl->getInt8Calibrator();
11238 mImpl->setFlags(builderFlags);
11250 return mImpl->getFlags();
11262 mImpl->clearFlag(builderFlag);
11274 mImpl->setFlag(builderFlag);
11286 return mImpl->getFlag(builderFlag);
11303 mImpl->setDeviceType(layer, deviceType);
11313 return mImpl->getDeviceType(layer);
11325 return mImpl->isDeviceTypeSet(layer);
11335 mImpl->resetDeviceType(layer);
11345 return mImpl->canRunOnDLA(layer);
11361 mImpl->setDLACore(dlaCore);
11371 return mImpl->getDLACore();
11382 mImpl->setDefaultDeviceType(deviceType);
11392 return mImpl->getDefaultDeviceType();
11414 return mImpl->setProfileStream(stream);
11426 return mImpl->getProfileStream();
11443 return mImpl->addOptimizationProfile(profile);
11456 return mImpl->getNbOptimizationProfiles();
11468 mImpl->setProfilingVerbosity(verbosity);
11481 return mImpl->getProfilingVerbosity();
11493 mImpl->setAlgorithmSelector(selector);
11503 return mImpl->getAlgorithmSelector();
11521 return mImpl->setCalibrationProfile(profile);
11533 return mImpl->getCalibrationProfile();
11552 mImpl->setQuantizationFlags(flags);
11566 return mImpl->getQuantizationFlags();
11580 mImpl->clearQuantizationFlag(flag);
11594 mImpl->setQuantizationFlag(flag);
11608 return mImpl->getQuantizationFlag(flag);
11630 return mImpl->setTacticSources(tacticSources);
11645 return mImpl->getTacticSources();
11665 return mImpl->createTimingCache(blob, size);
11688 return mImpl->setTimingCache(cache, ignoreMismatch);
11698 return mImpl->getTimingCache();
11730 mImpl->setMemoryPoolLimit(pool, poolSize);
11749 return mImpl->getMemoryPoolLimit(pool);
11767 mImpl->setPreviewFeature(feature, enable);
11781 return mImpl->getPreviewFeature(feature);
11814 mImpl->setBuilderOptimizationLevel(level);
11826 return mImpl->getBuilderOptimizationLevel();
11843 mImpl->setHardwareCompatibilityLevel(hardwareCompatibilityLevel);
11856 return mImpl->getHardwareCompatibilityLevel();
11869 mImpl->setPluginsToSerialize(paths, nbPaths);
11882 return mImpl->getPluginToSerialize(index);
11892 return mImpl->getNbPluginsToSerialize();
11921 mImpl->setMaxAuxStreams(nbStreams);
11931 return mImpl->getMaxAuxStreams();
11947 return mImpl->setProgressMonitor(monitor);
11957 return mImpl->getProgressMonitor();
11973 mImpl->setRuntimePlatform(runtimePlatform);
11985 return mImpl->getRuntimePlatform();
11997 mImpl->setMaxNbTactics(maxNbTactics);
12009 return mImpl->getMaxNbTactics();
12025 return mImpl->setTilingOptimizationLevel(level);
12037 return mImpl->getTilingOptimizationLevel();
12053 return mImpl->setL2LimitForTiling(size);
12065 return mImpl->getL2LimitForTiling();
12079 return mImpl->setRemoteAutoTuningConfig(config);
12089 return mImpl->getRemoteAutoTuningConfig();
12166 return mImpl->platformHasFastFp16();
12176 return mImpl->platformHasFastInt8();
12188 return mImpl->getMaxDLABatchSize();
12196 return mImpl->getNbDLACores();
12214 mImpl->setGpuAllocator(allocator);
12228 return mImpl->createBuilderConfig();
12254 return mImpl->createNetworkV2(flags);
12269 return mImpl->createOptimizationProfile();
12288 mImpl->setErrorRecorder(recorder);
12303 return mImpl->getErrorRecorder();
12321 return mImpl->platformHasTf32();
12340 return mImpl->buildSerializedNetwork(network, config);
12362 return mImpl->buildSerializedNetworkToStream(network, config, writer);
12386 return mImpl->buildSerializedNetworkWithKernelText(network, config, kernelText);
12406 return mImpl->buildEngineWithConfig(network, config);
12432 return mImpl->isNetworkSupported(network, config);
12442 return mImpl->getLogger();
12458 return mImpl->setMaxThreads(maxThreads);
12472 return mImpl->getMaxThreads();
12482 return mImpl->getPluginRegistry();
12495extern "C" TENSORRTAPI void* createInferBuilder_INTERNAL(
void* logger, int32_t version)
noexcept;
#define TRT_DEPRECATED_API
Definition: NvInferRuntimeBase.h:44
#define TENSORRTAPI
Definition: NvInferRuntimeBase.h:70
#define NV_TENSORRT_VERSION
Definition: NvInferRuntimeBase.h:102
#define TRT_NODISCARD
A stand-in for [[nodiscard]] and [[nodiscard(REASON)]] that works with older compilers.
Definition: NvInferRuntimeBase.h:57
#define TRT_DEPRECATED
Definition: NvInferRuntimeBase.h:42
#define TRT_DEPRECATED_ENUM
Definition: NvInferRuntimeBase.h:43
Definition: NvInferRuntimeBase.h:219
static constexpr int32_t MAX_DIMS
The maximum rank (number of dimensions) supported for a tensor.
Definition: NvInferRuntimeBase.h:222
An Activation layer in a network definition.
Definition: NvInfer.h:1421
void setBeta(float beta) noexcept
Set the beta parameter (must be finite).
Definition: NvInfer.h:1469
void setActivationType(ActivationType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1430
ActivationType getActivationType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1440
float getAlpha() const noexcept
Get the alpha parameter.
Definition: NvInfer.h:1478
virtual ~IActivationLayer() noexcept=default
float getBeta() const noexcept
Get the beta parameter.
Definition: NvInfer.h:1487
void setAlpha(float alpha) noexcept
Set the alpha parameter (must be finite).
Definition: NvInfer.h:1455
Describes the context and requirements, that could be fulfilled by one or more instances of IAlgorith...
Definition: NvInfer.h:10203
int32_t getNbOutputs() const noexcept
Return number of outputs of the algorithm.
Definition: NvInfer.h:10238
int32_t getNbInputs() const noexcept
Return number of inputs of the algorithm.
Definition: NvInfer.h:10230
char const * getName() const noexcept
Return name of the algorithm node.
Definition: NvInfer.h:10210
virtual ~IAlgorithmContext() noexcept=default
Dims getDimensions(int32_t index, OptProfileSelector select) const noexcept
Get the minimum / optimum / maximum dimensions for input or output tensor.
Definition: NvInfer.h:10222
Describes a variation of execution of a layer. An algorithm is represented by IAlgorithmVariant and t...
Definition: NvInfer.h:10262
std::size_t getWorkspaceSize() const noexcept
The size of the GPU temporary memory in bytes which the algorithm uses at execution time.
Definition: NvInfer.h:10283
float getTimingMSec() const noexcept
The time in milliseconds to execute the algorithm.
Definition: NvInfer.h:10275
IAlgorithmIOInfo const * getAlgorithmIOInfoByIndex(int32_t index) const noexcept
Returns the format of an Algorithm input or output. Algorithm inputs are incrementally numbered first...
Definition: NvInfer.h:10297
virtual ~IAlgorithm() noexcept=default
IAlgorithmVariant const & getAlgorithmVariant() const noexcept
Returns the algorithm variant.
Definition: NvInfer.h:10267
Carries information about input or output of the algorithm. IAlgorithmIOInfo for all the input and ou...
Definition: NvInfer.h:10106
virtual ~IAlgorithmIOInfo() noexcept=default
int64_t getVectorizedDim() const noexcept
Return the index of the vectorized dimension or -1 for non-vectorized formats.
Definition: NvInfer.h:10134
Dims getStrides() const noexcept
Return strides of the input/output tensor of algorithm. For vectorized formats, strides are given in ...
Definition: NvInfer.h:10124
DataType getDataType() const noexcept
Return DataType of the input/output of algorithm.
Definition: NvInfer.h:10113
int64_t getComponentsPerElement() const noexcept
Return the number of components per element. This is always 1 for non-vectorized formats.
Definition: NvInfer.h:10145
provides a unique 128-bit identifier, which along with the input and output information denotes the v...
Definition: NvInfer.h:10169
virtual ~IAlgorithmVariant() noexcept=default
int64_t getTactic() const noexcept
Return tactic of the algorithm.
Definition: NvInfer.h:10182
int64_t getImplementation() const noexcept
Return implementation of the algorithm.
Definition: NvInfer.h:10174
An assertion layer in a network.
Definition: NvInfer.h:5123
void setMessage(char const *message) noexcept
Set the message to print if the assertion fails.
Definition: NvInfer.h:5133
char const * getMessage() const noexcept
Return the assertion message.
Definition: NvInfer.h:5143
virtual ~IAssertionLayer() noexcept=default
This is a base class for Attention boundary layers.
Definition: NvInfer.h:6956
IAttention * getAttention() const noexcept
Get a pointer to the IAttention associated with this boundary layer.
Definition: NvInfer.h:6961
virtual ~IAttentionBoundaryLayer() noexcept=default
Helper for constructing an attention that consumes query, key and value tensors.
Definition: NvInfer.h:7074
ITensor * getMask() noexcept
Get the optional mask in attention.
Definition: NvInfer.h:7124
bool setMetadata(char const *metadata) noexcept
Set the metadata for IAttention.
Definition: NvInfer.h:7335
bool setDecomposable(bool decomposable) noexcept
Set whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7161
bool setName(char const *name) noexcept
Set the name of the attention.
Definition: NvInfer.h:7251
bool getDecomposable() const noexcept
Get whether the attention can be decomposed to use multiple kernels if no fused kernel support found.
Definition: NvInfer.h:7174
ITensor * getInput(int32_t index) const noexcept
Get the IAttention input corresponding to the given index.
Definition: NvInfer.h:7214
ITensor * getOutput(int32_t index) const noexcept
Get the IAttention output corresponding to the given index. IAttention has only one output.
Definition: NvInfer.h:7234
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer. IAttention has one output.
Definition: NvInfer.h:7222
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device attention execution.
Definition: NvInfer.h:7364
int32_t getNbInputs() const noexcept
Get the number of inputs of IAttention. IAttention has three inputs.
Definition: NvInfer.h:7202
bool setCausal(bool isCausal) noexcept
Set whether the attention will run a causal inference. Cannot be used together with setMask().
Definition: NvInfer.h:7137
bool setNormalizationOperation(AttentionNormalizationOp op) noexcept
Set the normalization operation for the attention.
Definition: NvInfer.h:7083
char const * getName() const noexcept
Return the name of the attention.
Definition: NvInfer.h:7263
bool setNormalizationQuantizeToType(DataType type) noexcept
Set the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7303
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:7376
AttentionNormalizationOp getNormalizationOperation() const noexcept
Get the normalization operation for the attention.
Definition: NvInfer.h:7095
bool setNormalizationQuantizeScale(ITensor &tensor) noexcept
Set the quantization scale for the attention normalization output.
Definition: NvInfer.h:7279
char const * getMetadata() const noexcept
Get the metadata of IAttention.
Definition: NvInfer.h:7348
DataType getNormalizationQuantizeToType() const noexcept
Get the datatype the attention normalization is quantized to.
Definition: NvInfer.h:7315
ITensor * getNormalizationQuantizeScale() const noexcept
Get the quantization scale for the attention normalization output.
Definition: NvInfer.h:7290
bool setInput(int32_t index, ITensor &input) noexcept
Append or replace an input of this layer with a specific tensor.
Definition: NvInfer.h:7193
bool setMask(ITensor &mask) noexcept
Set whether a mask will be used for the normalization operation.
Definition: NvInfer.h:7112
bool getCausal() const noexcept
Get whether the attention will run a causal inference.
Definition: NvInfer.h:7149
apiv::VAttention * mImpl
Definition: NvInfer.h:7382
virtual ~IAttention() noexcept=default
This layer represents an output of an IAttention.
Definition: NvInfer.h:7017
virtual ~IAttentionOutputLayer() noexcept=default
Holds properties for configuring a builder to produce an engine.
Definition: NvInfer.h:11148
void setMemoryPoolLimit(MemoryPoolType pool, std::size_t poolSize) noexcept
Set the memory size for the memory pool.
Definition: NvInfer.h:11728
TRT_DEPRECATED void setQuantizationFlags(QuantizationFlags flags) noexcept
Set the quantization flags.
Definition: NvInfer.h:11550
nvinfer1::ITimingCache * createTimingCache(void const *blob, std::size_t size) const noexcept
Create timing cache.
Definition: NvInfer.h:11663
void setPreviewFeature(PreviewFeature feature, bool enable) noexcept
Enable or disable a specific preview feature.
Definition: NvInfer.h:11765
TRT_DEPRECATED void setAlgorithmSelector(IAlgorithmSelector *selector) noexcept
Set Algorithm Selector.
Definition: NvInfer.h:11491
TRT_DEPRECATED void setInt8Calibrator(IInt8Calibrator *calibrator) noexcept
Set Int8 Calibration interface.
Definition: NvInfer.h:11209
bool getPreviewFeature(PreviewFeature feature) const noexcept
Get status of preview feature.
Definition: NvInfer.h:11779
int32_t getBuilderOptimizationLevel() noexcept
Get builder optimization level.
Definition: NvInfer.h:11824
bool setTacticSources(TacticSources tacticSources) noexcept
Set tactic sources.
Definition: NvInfer.h:11628
void setPluginsToSerialize(char const *const *paths, int32_t nbPaths) noexcept
Set the plugin libraries to be serialized with version-compatible engines.
Definition: NvInfer.h:11867
bool setTilingOptimizationLevel(TilingOptimizationLevel level) noexcept
Set the Tiling optimization level.
Definition: NvInfer.h:12023
bool setL2LimitForTiling(int64_t size) noexcept
Set the L2 cache usage limit for Tiling optimization.
Definition: NvInfer.h:12051
TRT_DEPRECATED IInt8Calibrator * getInt8Calibrator() const noexcept
Get Int8 Calibration interface.
Definition: NvInfer.h:11219
std::size_t getMemoryPoolLimit(MemoryPoolType pool) const noexcept
Get the memory size limit of the memory pool.
Definition: NvInfer.h:11747
int32_t getDLACore() const noexcept
Get the DLA core that the engine executes on.
Definition: NvInfer.h:11369
int32_t getNbPluginsToSerialize() const noexcept
Get the number of plugin library paths to be serialized with version-compatible engines.
Definition: NvInfer.h:11890
void setDeviceType(ILayer const *layer, DeviceType deviceType) noexcept
Set the device that this layer must execute on.
Definition: NvInfer.h:11301
void setEngineCapability(EngineCapability capability) noexcept
Configure the builder to target specified EngineCapability flow.
Definition: NvInfer.h:11185
int32_t getMaxAuxStreams() const noexcept
Get the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11929
bool getFlag(BuilderFlag builderFlag) const noexcept
Returns true if the build mode flag is set.
Definition: NvInfer.h:11284
void setMaxNbTactics(int32_t maxNbTactics) noexcept
Set the maximum number of tactics to time when there is a choice of tactics.
Definition: NvInfer.h:11995
TRT_DEPRECATED void clearQuantizationFlag(QuantizationFlag flag) noexcept
clear a quantization flag.
Definition: NvInfer.h:11578
int64_t getL2LimitForTiling() const noexcept
Get the L2 cache usage limit for tiling optimization.
Definition: NvInfer.h:12063
bool setRemoteAutoTuningConfig(char const *config) noexcept
Set a config string for remote auto tuning.
Definition: NvInfer.h:12077
void setProgressMonitor(IProgressMonitor *monitor) noexcept
Sets the progress monitor for building a network.
Definition: NvInfer.h:11945
void setProfilingVerbosity(ProfilingVerbosity verbosity) noexcept
Set verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:11466
int32_t getNbOptimizationProfiles() const noexcept
Get number of optimization profiles.
Definition: NvInfer.h:11454
nvinfer1::ITimingCache const * getTimingCache() const noexcept
Get the pointer to the timing cache from current IBuilderConfig.
Definition: NvInfer.h:11696
void reset() noexcept
Resets the builder configuration to defaults.
Definition: NvInfer.h:11400
bool setTimingCache(ITimingCache const &cache, bool ignoreMismatch) noexcept
Attach a timing cache to IBuilderConfig.
Definition: NvInfer.h:11686
char const * getPluginToSerialize(int32_t index) const noexcept
Get the plugin library path to be serialized with version-compatible engines.
Definition: NvInfer.h:11880
EngineCapability getEngineCapability() const noexcept
Query EngineCapability flow configured for the builder.
Definition: NvInfer.h:11197
RuntimePlatform getRuntimePlatform() const noexcept
Get the target platform for runtime execution.
Definition: NvInfer.h:11983
DeviceType getDefaultDeviceType() const noexcept
Get the default DeviceType which was set by setDefaultDeviceType.
Definition: NvInfer.h:11390
void setRuntimePlatform(RuntimePlatform runtimePlatform) noexcept
Set the target platform for runtime execution.
Definition: NvInfer.h:11971
int32_t getMaxNbTactics() const noexcept
Query the maximum number of tactics timed when there is a choice.
Definition: NvInfer.h:12007
BuilderFlags getFlags() const noexcept
Get the build mode flags for this builder config. Defaults to 0.
Definition: NvInfer.h:11248
void setFlags(BuilderFlags builderFlags) noexcept
Set the build mode flags to turn on builder options for this network.
Definition: NvInfer.h:11236
TacticSources getTacticSources() const noexcept
Get tactic sources.
Definition: NvInfer.h:11643
void resetDeviceType(ILayer const *layer) noexcept
reset the DeviceType for this layer
Definition: NvInfer.h:11333
void setDLACore(int32_t dlaCore) noexcept
Sets the DLA core used by the network. Defaults to -1.
Definition: NvInfer.h:11359
HardwareCompatibilityLevel getHardwareCompatibilityLevel() const noexcept
Get the hardware compatibility level.
Definition: NvInfer.h:11854
char const * getRemoteAutoTuningConfig() const noexcept
Get a config string for remote auto tuning.
Definition: NvInfer.h:12087
TRT_DEPRECATED QuantizationFlags getQuantizationFlags() const noexcept
Get the quantization flags.
Definition: NvInfer.h:11564
void clearFlag(BuilderFlag builderFlag) noexcept
clear a single build mode flag.
Definition: NvInfer.h:11260
int32_t addOptimizationProfile(IOptimizationProfile const *profile) noexcept
Add an optimization profile.
Definition: NvInfer.h:11441
IProgressMonitor * getProgressMonitor() const noexcept
Definition: NvInfer.h:11955
apiv::VBuilderConfig * mImpl
Definition: NvInfer.h:12093
TRT_DEPRECATED IOptimizationProfile const * getCalibrationProfile() noexcept
Get the current calibration profile.
Definition: NvInfer.h:11531
int32_t getAvgTimingIterations() const noexcept
Query the number of averaging iterations.
Definition: NvInfer.h:11172
void setDefaultDeviceType(DeviceType deviceType) noexcept
Sets the default DeviceType to be used by the builder. It ensures that all the layers that can run on...
Definition: NvInfer.h:11380
void setFlag(BuilderFlag builderFlag) noexcept
Set a single build mode flag.
Definition: NvInfer.h:11272
TRT_DEPRECATED bool setCalibrationProfile(IOptimizationProfile const *profile) noexcept
Add a calibration profile.
Definition: NvInfer.h:11519
virtual ~IBuilderConfig() noexcept=default
DeviceType getDeviceType(ILayer const *layer) const noexcept
Get the device that this layer executes on.
Definition: NvInfer.h:11311
bool canRunOnDLA(ILayer const *layer) const noexcept
Checks if a layer can run on DLA.
Definition: NvInfer.h:11343
TRT_DEPRECATED bool getQuantizationFlag(QuantizationFlag flag) const noexcept
Returns true if the quantization flag is set.
Definition: NvInfer.h:11606
cudaStream_t getProfileStream() const noexcept
Get the CUDA stream that is used to profile this network.
Definition: NvInfer.h:11424
void setHardwareCompatibilityLevel(HardwareCompatibilityLevel hardwareCompatibilityLevel) noexcept
Set the hardware compatibility level.
Definition: NvInfer.h:11841
TilingOptimizationLevel getTilingOptimizationLevel() const noexcept
Get the Tiling optimization level.
Definition: NvInfer.h:12035
TRT_DEPRECATED void setQuantizationFlag(QuantizationFlag flag) noexcept
Set a single quantization flag.
Definition: NvInfer.h:11592
void setMaxAuxStreams(int32_t nbStreams) noexcept
Set the maximum number of auxiliary streams that TRT is allowed to use.
Definition: NvInfer.h:11919
ProfilingVerbosity getProfilingVerbosity() const noexcept
Get verbosity level of layer information exposed in NVTX annotations and IEngineInspector.
Definition: NvInfer.h:11479
bool isDeviceTypeSet(ILayer const *layer) const noexcept
whether the DeviceType has been explicitly set for this layer
Definition: NvInfer.h:11323
void setBuilderOptimizationLevel(int32_t level) noexcept
Set builder optimization level.
Definition: NvInfer.h:11812
void setProfileStream(const cudaStream_t stream) noexcept
Set the CUDA stream that is used to profile this network.
Definition: NvInfer.h:11412
TRT_DEPRECATED IAlgorithmSelector * getAlgorithmSelector() const noexcept
Get Algorithm Selector.
Definition: NvInfer.h:11501
Builds an engine from a network definition.
Definition: NvInfer.h:12155
int32_t getMaxDLABatchSize() const noexcept
Get the maximum batch size DLA can support. For any tensor the total volume of index dimensions combi...
Definition: NvInfer.h:12186
int32_t getNbDLACores() const noexcept
Return the number of DLA engines available to this builder.
Definition: NvInfer.h:12194
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:12301
apiv::VBuilder * mImpl
Definition: NvInfer.h:12486
ILogger * getLogger() const noexcept
get the logger with which the builder was created
Definition: NvInfer.h:12440
bool isNetworkSupported(INetworkDefinition const &network, IBuilderConfig const &config) const noexcept
Checks that a network is within the scope of the IBuilderConfig settings.
Definition: NvInfer.h:12430
int32_t getMaxThreads() const noexcept
get the maximum number of threads that can be used by the builder.
Definition: NvInfer.h:12470
IPluginRegistry & getPluginRegistry() noexcept
get the local plugin registry that can be used by the builder.
Definition: NvInfer.h:12480
TRT_DEPRECATED bool platformHasFastInt8() const noexcept
Determine whether the platform has fast native int8.
Definition: NvInfer.h:12174
nvinfer1::IOptimizationProfile * createOptimizationProfile() noexcept
Create a new optimization profile.
Definition: NvInfer.h:12267
void setGpuAllocator(IGpuAllocator *allocator) noexcept
Set the GPU allocator.
Definition: NvInfer.h:12212
nvinfer1::INetworkDefinition * createNetworkV2(NetworkDefinitionCreationFlags flags) noexcept
Create a network definition object.
Definition: NvInfer.h:12252
nvinfer1::IBuilderConfig * createBuilderConfig() noexcept
Create a builder configuration object.
Definition: NvInfer.h:12226
void reset() noexcept
Resets the builder state to default values.
Definition: NvInfer.h:12309
bool setMaxThreads(int32_t maxThreads) noexcept
Set the maximum number of threads.
Definition: NvInfer.h:12456
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:12286
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds and serializes a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:12338
virtual ~IBuilder() noexcept=default
TRT_DEPRECATED bool platformHasTf32() const noexcept
Determine whether the platform has TF32 support.
Definition: NvInfer.h:12319
bool buildSerializedNetworkToStream(INetworkDefinition &network, IBuilderConfig &config, IStreamWriter &writer) noexcept
Builds and serializes a network into stream for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:12359
nvinfer1::ICudaEngine * buildEngineWithConfig(INetworkDefinition &network, IBuilderConfig &config) noexcept
Builds a network for the given INetworkDefinition and IBuilderConfig.
Definition: NvInfer.h:12404
nvinfer1::IHostMemory * buildSerializedNetwork(INetworkDefinition &network, IBuilderConfig &config, IHostMemory *&kernelText) noexcept
Extended form of buildSerializedNetwork that optionally permits getting the kernelText.
Definition: NvInfer.h:12383
A cast layer in a network.
Definition: NvInfer.h:3984
virtual ~ICastLayer() noexcept=default
apiv::VCastLayer * mImpl
Definition: NvInfer.h:4010
DataType getToType() const noexcept
Return cast layer output type.
Definition: NvInfer.h:4004
void setToType(DataType toType) noexcept
Set cast layer output type.
Definition: NvInfer.h:3993
A concatenation layer in a network definition.
Definition: NvInfer.h:2131
void setAxis(int32_t axis) noexcept
Set the axis along which concatenation occurs.
Definition: NvInfer.h:2144
int32_t getAxis() const noexcept
Get the axis along which concatenation occurs.
Definition: NvInfer.h:2154
virtual ~IConcatenationLayer() noexcept=default
This layer represents a condition input to an IIfConditional.
Definition: NvInfer.h:4647
virtual ~IConditionLayer() noexcept=default
Layer that represents a constant value.
Definition: NvInfer.h:4023
void setWeights(Weights weights) noexcept
Set the weights for the layer.
Definition: NvInfer.h:4033
Weights getWeights() const noexcept
Get the weights for the layer.
Definition: NvInfer.h:4043
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions for the layer.
Definition: NvInfer.h:4055
apiv::VConstantLayer * mImpl
Definition: NvInfer.h:4073
virtual ~IConstantLayer() noexcept=default
Dims getDimensions() const noexcept
Get the dimensions for the layer.
Definition: NvInfer.h:4067
A convolution layer in a network definition.
Definition: NvInfer.h:1101
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:1226
Weights getBiasWeights() const noexcept
Get the bias weights for the convolution.
Definition: NvInfer.h:1199
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:1267
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1371
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the convolution.
Definition: NvInfer.h:1357
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the convolution.
Definition: NvInfer.h:1327
Weights getKernelWeights() const noexcept
Get the kernel weights of the convolution.
Definition: NvInfer.h:1174
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the convolution.
Definition: NvInfer.h:1317
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the convolution.
Definition: NvInfer.h:1381
int64_t getNbOutputMaps() const noexcept
Get the number of output maps for the convolution.
Definition: NvInfer.h:1120
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the convolution.
Definition: NvInfer.h:1164
Dims getPostPadding() const noexcept
Get the post-padding.
Definition: NvInfer.h:1253
int64_t getNbGroups() const noexcept
Get the number of groups of the convolution.
Definition: NvInfer.h:1150
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:1279
virtual ~IConvolutionLayer() noexcept=default
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a convolution.
Definition: NvInfer.h:1140
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output maps for the convolution.
Definition: NvInfer.h:1110
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the convolution.
Definition: NvInfer.h:1189
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:1302
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the convolution.
Definition: NvInfer.h:1345
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the convolution.
Definition: NvInfer.h:1216
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the convolution.
Definition: NvInfer.h:1243
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the convolution.
Definition: NvInfer.h:1292
An engine for executing inference on a built network, with functionally unsafe features.
Definition: NvInferRuntime.h:3197
Layer that represents a cumulative operation across a tensor.
Definition: NvInfer.h:6838
bool setOperation(CumulativeOperation op) noexcept
Set the cumulative operation for the layer.
Definition: NvInfer.h:6849
void setReverse(bool reverse) noexcept
Specify whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6897
apiv::VCumulativeLayer * mImpl
Definition: NvInfer.h:6915
bool getExclusive() const noexcept
Get whether it is exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6885
virtual ~ICumulativeLayer() noexcept=default
bool getReverse() const noexcept
Get the boolean that specifies whether the cumulative operation should be applied backward.
Definition: NvInfer.h:6909
void setExclusive(bool exclusive) noexcept
Set whether it is an exclusive accumulation or inclusive accumulation.
Definition: NvInfer.h:6873
CumulativeOperation getOperation() const noexcept
Get the cumulative operation for the layer.
Definition: NvInfer.h:6861
A deconvolution layer in a network definition.
Definition: NvInfer.h:2172
void setBiasWeights(Weights weights) noexcept
Set the bias weights for the deconvolution.
Definition: NvInfer.h:2260
int64_t getNbGroups() const noexcept
Get the number of groups for a deconvolution.
Definition: NvInfer.h:2221
Weights getKernelWeights() const noexcept
Get the kernel weights for the deconvolution.
Definition: NvInfer.h:2245
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding of the deconvolution.
Definition: NvInfer.h:2287
Dims getStrideNd() const noexcept
Get the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2402
Dims getDilationNd() const noexcept
Get the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2468
Weights getBiasWeights() const noexcept
Get the bias weights for the deconvolution.
Definition: NvInfer.h:2270
void setKernelWeights(Weights weights) noexcept
Set the kernel weights for the deconvolution.
Definition: NvInfer.h:2235
int64_t getNbOutputMaps() const noexcept
Get the number of output feature maps for the deconvolution.
Definition: NvInfer.h:2191
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride of the deconvolution.
Definition: NvInfer.h:2392
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:2324
Dims getKernelSizeNd() const noexcept
Get the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2375
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding of the deconvolution.
Definition: NvInfer.h:2314
void setKernelSizeNd(Dims const &kernelSize) noexcept
Set the multi-dimension kernel size of the deconvolution.
Definition: NvInfer.h:2365
virtual ~IDeconvolutionLayer() noexcept=default
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2420
void setNbOutputMaps(int64_t nbOutputMaps) noexcept
Set the number of output feature maps for the deconvolution.
Definition: NvInfer.h:2181
Dims getPaddingNd() const noexcept
Get the multi-dimension padding of the deconvolution.
Definition: NvInfer.h:2432
void setDilationNd(Dims const &dilation) noexcept
Set the multi-dimension dilation of the deconvolution.
Definition: NvInfer.h:2458
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:2338
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups for a deconvolution.
Definition: NvInfer.h:2211
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:2297
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:2350
A Dequantize layer in a network definition.
Definition: NvInfer.h:5736
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5785
void setToType(DataType toType) noexcept
Set the Dequantize layer output type.
Definition: NvInfer.h:5801
virtual ~IDequantizeLayer() noexcept=default
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5746
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5774
DataType getToType() const noexcept
Return the Dequantize layer output type.
Definition: NvInfer.h:5813
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5757
Definition: NvInfer.h:8003
virtual ~IDistCollectiveLayer() noexcept=default
A network layer to perform dynamic quantization.
Definition: NvInfer.h:5841
DataType getScaleType() const noexcept
Return the scale factors data type.
Definition: NvInfer.h:5907
TRT_DEPRECATED void setAxis(int32_t axis) noexcept
Set the axis along which block quantization occurs.
Definition: NvInfer.h:5920
TRT_DEPRECATED void setBlockSize(int32_t size) noexcept
Set the size of the quantization block.
Definition: NvInfer.h:5943
Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5978
void setScaleType(DataType scaleType) noexcept
Set the data type of the scale factors used to quantize the data.
Definition: NvInfer.h:5894
DataType getToType() const noexcept
Return DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5881
TRT_DEPRECATED int32_t getAxis() const noexcept
Get the axis along which blocking occurs.
Definition: NvInfer.h:5930
virtual ~IDynamicQuantizeLayer() noexcept=default
void setToType(DataType toType) noexcept
Set DynamicQuantizeLayer's quantized output type.
Definition: NvInfer.h:5868
void setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5966
TRT_DEPRECATED int32_t getBlockSize() const noexcept
Get the size of the quantization block.
Definition: NvInfer.h:5953
An Einsum layer in a network.
Definition: NvInfer.h:6023
bool setEquation(char const *equation) noexcept
Set the equation. The equation is a comma-separated list of subscript labels, where each label refers...
Definition: NvInfer.h:6034
virtual ~IEinsumLayer() noexcept=default
char const * getEquation() const noexcept
Return the equation.
Definition: NvInfer.h:6044
A elementwise layer in a network definition.
Definition: NvInfer.h:2542
virtual ~IElementWiseLayer() noexcept=default
apiv::VElementWiseLayer * mImpl
Definition: NvInfer.h:2571
ElementWiseOperation getOperation() const noexcept
Get the binary operation for the layer.
Definition: NvInfer.h:2565
void setOperation(ElementWiseOperation op) noexcept
Set the binary operation for the layer.
Definition: NvInfer.h:2553
Generate a tensor according to a specified mode.
Definition: NvInfer.h:5237
bool isAlphaBetaInt64() const noexcept
Return true if alpha/beta have type int64, false if they have type double.
Definition: NvInfer.h:5469
FillOperation getOperation() const noexcept
Get the fill operation for the layer.
Definition: NvInfer.h:5283
void setOperation(FillOperation op) noexcept
Set the fill operation for the layer.
Definition: NvInfer.h:5273
DataType getToType() const noexcept
Get the fill layer output type.
Definition: NvInfer.h:5499
void setAlphaInt64(int64_t alpha) noexcept
Set the alpha parameter with int64 datatype.
Definition: NvInfer.h:5412
void setBetaInt64(int64_t beta) noexcept
Set the beta parameter with int64 datatype.
Definition: NvInfer.h:5446
void setBeta(double beta) noexcept
Set the beta parameter.
Definition: NvInfer.h:5336
int64_t getAlphaInt64() const noexcept
Get the value of alpha parameter with int64 datatype.
Definition: NvInfer.h:5427
int64_t getBetaInt64() const noexcept
Get the value of beta parameter with int64 datatype.
Definition: NvInfer.h:5461
double getAlpha() const noexcept
Get the value of alpha parameter.
Definition: NvInfer.h:5317
void setDimensions(Dims const &dimensions) noexcept
Set the output tensor's dimensions.
Definition: NvInfer.h:5248
void setAlpha(double alpha) noexcept
Set the alpha parameter.
Definition: NvInfer.h:5302
void setToType(DataType toType) noexcept
Set the fill layer output type.
Definition: NvInfer.h:5487
Dims getDimensions() const noexcept
Get the output tensor's dimensions.
Definition: NvInfer.h:5263
double getBeta() const noexcept
Get the value of beta parameter.
Definition: NvInfer.h:5351
virtual ~IFillLayer() noexcept=default
A Gather layer in a network definition. Supports several kinds of gathering.
Definition: NvInfer.h:2675
void setGatherAxis(int32_t axis) noexcept
Set the axis used by GatherMode::kELEMENTS and GatherMode::kDEFAULT The axis must be less than the nu...
Definition: NvInfer.h:2686
void setNbElementWiseDims(int32_t elementWiseDims) noexcept
Set the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2721
apiv::VGatherLayer * mImpl
Definition: NvInfer.h:2757
int32_t getNbElementWiseDims() const noexcept
Get the number of leading dimensions of indices tensor to be handled elementwise.
Definition: NvInfer.h:2731
void setMode(GatherMode mode) noexcept
Set the gather mode.
Definition: NvInfer.h:2741
int32_t getGatherAxis() const noexcept
Get the axis to gather on.
Definition: NvInfer.h:2698
GatherMode getMode() const noexcept
Get the gather mode.
Definition: NvInfer.h:2751
virtual ~IGatherLayer() noexcept=default
A GridSample layer in a network definition.
Definition: NvInfer.h:6245
void setInterpolationMode(InterpolationMode mode) noexcept
Set the grid sample interpolation mode.
Definition: NvInfer.h:6252
bool setSampleMode(SampleMode mode) noexcept
Set the sample mode.
Definition: NvInfer.h:6298
void setAlignCorners(bool alignCorners) noexcept
Set the align corners mode.
Definition: NvInfer.h:6274
apiv::VGridSampleLayer * mImpl
Definition: NvInfer.h:6316
SampleMode getSampleMode() const noexcept
Get the sample mode.
Definition: NvInfer.h:6310
InterpolationMode getInterpolationMode() const noexcept
Get the grid sample interpolation mode.
Definition: NvInfer.h:6264
bool getAlignCorners() const noexcept
Get the align corners mode.
Definition: NvInfer.h:6286
virtual ~IGridSampleLayer() noexcept=default
Class to handle library allocated memory that is accessible to the user.
Definition: NvInferRuntime.h:142
A layer that represents the identity function.
Definition: NvInfer.h:3971
apiv::VIdentityLayer * mImpl
Definition: NvInfer.h:3973
virtual ~IIdentityLayer() noexcept=default
This is a base class for Conditional boundary layers.
Definition: NvInfer.h:4626
IIfConditional * getConditional() const noexcept
Get a pointer to the IIfConditional associated with this boundary layer.
Definition: NvInfer.h:4631
virtual ~IIfConditionalBoundaryLayer() noexcept=default
Helper for constructing conditionally-executed subgraphs.
Definition: NvInfer.h:4709
IIfConditionalInputLayer * addInput(ITensor &input) noexcept
Add an If-conditional input.
Definition: NvInfer.h:4750
char const * getName() const noexcept
Return the name of the conditional.
Definition: NvInfer.h:4775
virtual ~IIfConditional() noexcept=default
IConditionLayer * setCondition(ITensor &condition) noexcept
Set the condition tensor for this If-Conditional construct.
Definition: NvInfer.h:4720
IIfConditionalOutputLayer * addOutput(ITensor &trueSubgraphOutput, ITensor &falseSubgraphOutput) noexcept
Add an If-conditional output.
Definition: NvInfer.h:4738
void setName(char const *name) noexcept
Set the name of the conditional.
Definition: NvInfer.h:4765
This layer represents an output of an IIfConditional.
Definition: NvInfer.h:4664
virtual ~IIfConditionalOutputLayer() noexcept=default
Application-implemented interface for calibration.
Definition: NvInfer.h:9831
virtual TRT_DEPRECATED int32_t getBatchSize() const noexcept=0
Get the batch size used for calibration batches.
A layer to do iterations.
Definition: NvInfer.h:4940
virtual ~IIteratorLayer() noexcept=default
void setReverse(bool reverse) noexcept
Set iteration order to be reverse.
Definition: NvInfer.h:4967
bool getReverse() const noexcept
Check if the iteration order is reverse.
Definition: NvInfer.h:4977
int32_t getAxis() const noexcept
Get axis being iterated over.
Definition: NvInfer.h:4953
void setAxis(int32_t axis) noexcept
Set axis to iterate over.
Definition: NvInfer.h:4945
Layer that represents a KVCacheUpdate operation.
Definition: NvInfer.h:7505
bool setCacheMode(KVCacheMode cacheMode) noexcept
Set the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7528
virtual ~IKVCacheUpdateLayer() noexcept=default
KVCacheMode getCacheMode() const noexcept
Get the mode of the KVCacheUpdate layer.
Definition: NvInfer.h:7538
apiv::VKVCacheUpdateLayer * mImpl
Definition: NvInfer.h:7544
A LRN layer in a network definition.
Definition: NvInfer.h:1786
int64_t getWindowSize() const noexcept
Get the LRN window size.
Definition: NvInfer.h:1807
float getAlpha() const noexcept
Get the LRN alpha value.
Definition: NvInfer.h:1829
void setWindowSize(int64_t windowSize) noexcept
Set the LRN window size.
Definition: NvInfer.h:1797
void setK(float k) noexcept
Set the LRN K value.
Definition: NvInfer.h:1863
void setAlpha(float alpha) noexcept
Set the LRN alpha value.
Definition: NvInfer.h:1819
void setBeta(float beta) noexcept
Set the LRN beta value.
Definition: NvInfer.h:1841
virtual ~ILRNLayer() noexcept=default
float getBeta() const noexcept
Get the LRN beta value.
Definition: NvInfer.h:1851
float getK() const noexcept
Get the LRN K value.
Definition: NvInfer.h:1873
Base class for all layer classes in a network definition.
Definition: NvInfer.h:583
TRT_DEPRECATED void setPrecision(DataType dataType) noexcept
Set the preferred or required computational precision of this layer in a weakly-typed network.
Definition: NvInfer.h:703
TRT_DEPRECATED void setOutputType(int32_t index, DataType dataType) noexcept
Set the output type of this layer in a weakly-typed network.
Definition: NvInfer.h:791
TRT_DEPRECATED bool precisionIsSet() const noexcept
whether the computational precision has been set for this layer
Definition: NvInfer.h:729
void setMetadata(char const *metadata) noexcept
Set the metadata for this layer.
Definition: NvInfer.h:854
TRT_DEPRECATED void resetOutputType(int32_t index) noexcept
reset the output type for this layer
Definition: NvInfer.h:836
void setName(char const *name) noexcept
Set the name of a layer.
Definition: NvInfer.h:604
int32_t getNbInputs() const noexcept
Get the number of inputs of a layer.
Definition: NvInfer.h:622
int32_t getNbRanks() const noexcept
Get the number of ranks for multi-device execution.
Definition: NvInfer.h:900
char const * getMetadata() const noexcept
Get the metadata of the layer.
Definition: NvInfer.h:867
DataType getOutputType(int32_t index) const noexcept
get the output type of this layer
Definition: NvInfer.h:806
DataType getPrecision() const noexcept
get the computational precision of this layer
Definition: NvInfer.h:715
TRT_DEPRECATED bool outputTypeIsSet(int32_t index) const noexcept
whether the output type has been set for this layer
Definition: NvInfer.h:822
char const * getName() const noexcept
Return the name of a layer.
Definition: NvInfer.h:614
int32_t getNbOutputs() const noexcept
Get the number of outputs of a layer.
Definition: NvInfer.h:643
ITensor * getOutput(int32_t index) const noexcept
Get the layer output corresponding to the given index.
Definition: NvInfer.h:653
void setInput(int32_t index, ITensor &tensor) noexcept
Replace an input of this layer with a specific tensor.
Definition: NvInfer.h:670
ITensor * getInput(int32_t index) const noexcept
Get the layer input corresponding to the given index.
Definition: NvInfer.h:635
bool setNbRanks(int32_t nbRanks) noexcept
Set the number of ranks for multi-device execution.
Definition: NvInfer.h:888
LayerType getType() const noexcept
Return the type of a layer.
Definition: NvInfer.h:590
TRT_DEPRECATED void resetPrecision() noexcept
reset the computational precision for this layer
Definition: NvInfer.h:741
virtual ~ILayer() noexcept=default
Application-implemented logging interface for the builder, refitter and runtime.
Definition: NvInferRuntime.h:1588
This is a base class for Loop boundary layers.
Definition: NvInfer.h:4603
virtual ~ILoopBoundaryLayer() noexcept=default
ILoop * getLoop() const noexcept
Get a pointer to ILoop associated with this boundary layer.
Definition: NvInfer.h:4608
Helper for creating a recurrent subgraph.
Definition: NvInfer.h:4998
void setName(char const *name) noexcept
Set the name of the loop.
Definition: NvInfer.h:5068
ITripLimitLayer * addTripLimit(ITensor &tensor, TripLimit limit) noexcept
Add a trip-count limiter, based on the given tensor.
Definition: NvInfer.h:5027
IIteratorLayer * addIterator(ITensor &tensor, int32_t axis=0, bool reverse=false) noexcept
Return layer that subscripts tensor by loop iteration.
Definition: NvInfer.h:5040
ILoopOutputLayer * addLoopOutput(ITensor &tensor, LoopOutput outputKind, int32_t axis=0) noexcept
Make an output for this loop, based on the given tensor.
Definition: NvInfer.h:5053
virtual ~ILoop() noexcept=default
char const * getName() const noexcept
Return the name of the loop.
Definition: NvInfer.h:5078
IRecurrenceLayer * addRecurrence(ITensor &initialValue) noexcept
Create a recurrence layer for this loop with initialValue as its first input.
Definition: NvInfer.h:5006
An ILoopOutputLayer is the sole way to get output from a loop.
Definition: NvInfer.h:4840
virtual ~ILoopOutputLayer() noexcept=default
int32_t getAxis() const noexcept
Get axis being concatenated over.
Definition: NvInfer.h:4870
LoopOutput getLoopOutput() const noexcept
Get which kind a loop output has.
Definition: NvInfer.h:4845
void setAxis(int32_t axis) noexcept
Set where to insert the contenation axis. Ignored if getLoopOutput() is kLAST_VALUE.
Definition: NvInfer.h:4862
Layer that represents a Matrix Multiplication.
Definition: NvInfer.h:3818
apiv::VMatrixMultiplyLayer * mImpl
Definition: NvInfer.h:3846
virtual ~IMatrixMultiplyLayer() noexcept=default
MatrixOperation getOperation(int32_t index) const noexcept
Get the operation for an input tensor.
Definition: NvInfer.h:3840
void setOperation(int32_t index, MatrixOperation op) noexcept
Set the operation for an input tensor.
Definition: NvInfer.h:3828
A MoE layer in a network definition. Mixture of Experts (MoE) is a collection of experts with each ex...
Definition: NvInfer.h:7687
void setSwigluParamLimit(float limit) noexcept
Set the SwiGLU parameter limit.
Definition: NvInfer.h:7902
void setDynQOutputScaleType(DataType type) noexcept
Set the dynamic quantization output scale type.
Definition: NvInfer.h:7855
MoEActType getActivationType() const noexcept
Get the activation type for the MoE layer.
Definition: NvInfer.h:7736
void setQuantizationToType(DataType type) noexcept
Set the data type the mul output is quantized to.
Definition: NvInfer.h:7803
void setQuantizationDynamicDblQ(ITensor &fcDownActivationDblQScale, DataType dataType, Dims const &blockShape, DataType dynQOutputScaleType) noexcept
Configure dynamic quantization (with double quantization) after the mul op. ┌── fcGate ── activation ...
Definition: NvInfer.h:7788
void setQuantizationStatic(ITensor &fcDownActivationScale, DataType dataType) noexcept
Configure static quantization after the mul op. ┌── fcGate ── activation ───┐ │ │ hiddenStates ───┤ ├...
Definition: NvInfer.h:7759
virtual ~IMoELayer() noexcept=default
float getSwigluParamLimit() const noexcept
Get the SwiGLU parameter limit.
Definition: NvInfer.h:7914
DataType getQuantizationToType() const noexcept
Get the data type the mul in MoE layer is quantized to.
Definition: NvInfer.h:7815
DataType getDynQOutputScaleType() const noexcept
Get the dynamic quantization output scale type.
Definition: NvInfer.h:7867
void setActivationType(MoEActType activationType) noexcept
Set the activation type for the MoE layer.
Definition: NvInfer.h:7724
Dims getQuantizationBlockShape() const noexcept
Get the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7843
void setGatedWeights(ITensor &fcGateWeights, ITensor &fcUpWeights, ITensor &fcDownWeights, MoEActType activationType) noexcept
Set the weights of the experts when each expert is a GLU (gated linear unit). In each GLU,...
Definition: NvInfer.h:7700
float getSwigluParamBeta() const noexcept
Get the SwiGLU parameter beta.
Definition: NvInfer.h:7966
void setSwigluParamBeta(float beta) noexcept
Set the SwiGLU parameter beta.
Definition: NvInfer.h:7954
void setGatedBiases(ITensor &fcGateBiases, ITensor &fcUpBiases, ITensor &fcDownBiases) noexcept
Set the biases of the experts when each expert is a GLU (gated linear unit). In each GLU,...
Definition: NvInfer.h:7712
void setSwigluParams(float limit, float alpha, float beta) noexcept
Set the SwiGLU parameters.
Definition: NvInfer.h:7888
void setQuantizationBlockShape(Dims const &blockShape) noexcept
Set the block shape for the quantization of the Mul output.
Definition: NvInfer.h:7831
void setInput(int32_t index, ITensor &tensor) noexcept
Set the input of the MoE layer.
Definition: NvInfer.h:7983
float getSwigluParamAlpha() const noexcept
Get the SwiGLU parameter alpha.
Definition: NvInfer.h:7940
void setSwigluParamAlpha(float alpha) noexcept
Set the SwiGLU parameter alpha.
Definition: NvInfer.h:7928
A non-maximum suppression layer in a network definition.
Definition: NvInfer.h:6397
virtual ~INMSLayer() noexcept=default
void setTopKBoxLimit(int32_t limit) noexcept
Set the TopK box limit parameter for the layer.
Definition: NvInfer.h:6434
void setBoundingBoxFormat(BoundingBoxFormat fmt) noexcept
Set the bounding box format parameter for the layer.
Definition: NvInfer.h:6408
BoundingBoxFormat getBoundingBoxFormat() const noexcept
Get the bounding box format parameter for the layer.
Definition: NvInfer.h:6420
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:6479
apiv::VNMSLayer * mImpl
Definition: NvInfer.h:6497
int32_t getTopKBoxLimit() const noexcept
Get the TopK box limit parameter for the layer.
Definition: NvInfer.h:6444
DataType getIndicesType() const noexcept
Return the NMS layer indices type.
Definition: NvInfer.h:6491
A network definition for input to the builder.
Definition: NvInfer.h:8027
IConcatenationLayer * addConcatenation(ITensor *const *inputs, int32_t nbInputs) noexcept
Add a concatenation layer to the network.
Definition: NvInfer.h:8255
IShuffleLayer * addShuffle(ITensor &input) noexcept
Add a shuffle layer to the network.
Definition: NvInfer.h:8318
void setName(char const *name) noexcept
Sets the name of the network.
Definition: NvInfer.h:8784
ITopKLayer * addTopK(ITensor &input, TopKOperation op, int32_t k, uint32_t reduceAxes, DataType indicesType) noexcept
Add a TopK layer to the network.
Definition: NvInfer.h:8514
bool markDebug(ITensor &tensor) noexcept
Mark a tensor as a debug tensor.
Definition: NvInfer.h:8098
ILRNLayer * addLRN(ITensor &input, int64_t window, float alpha, float beta, float k) noexcept
Add a LRN layer to the network.
Definition: NvInfer.h:8199
ICumulativeLayer * addCumulative(ITensor &input, ITensor &axis, CumulativeOperation operation, bool exclusive, bool reverse) noexcept
Add a cumulative layer to the network.
Definition: NvInfer.h:9532
IAssertionLayer * addAssertion(ITensor &condition, char const *message) noexcept
Add an assertion layer to the network.
Definition: NvInfer.h:9100
TRT_DEPRECATED INonZeroLayer * addNonZero(ITensor &input) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8605
IConvolutionLayer * addConvolutionNd(ITensor &input, int64_t nbOutputMaps, Dims const &kernelSize, Weights kernelWeights, Weights biasWeights) noexcept
Add a multi-dimension convolution layer to the network.
Definition: NvInfer.h:8919
ICastLayer * addCast(ITensor &input, DataType toType) noexcept
Add a cast layer.
Definition: NvInfer.h:8674
IScaleLayer * addScaleNd(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power, int32_t channelAxis) noexcept
Add a multi-dimension scale layer to the network.
Definition: NvInfer.h:8998
char const * getName() const noexcept
Returns the name associated with the network.
Definition: NvInfer.h:8798
IParametricReLULayer * addParametricReLU(ITensor &input, ITensor &slope) noexcept
Add a parametric ReLU layer to the network.
Definition: NvInfer.h:8897
ITensor * getOutput(int32_t index) const noexcept
Get the output tensor specified by the given index.
Definition: NvInfer.h:8419
ITensor * getInput(int32_t index) const noexcept
Get the input tensor specified by the given index.
Definition: NvInfer.h:8389
TRT_DEPRECATED ITopKLayer * addTopK(ITensor &input, TopKOperation op, int32_t k, uint32_t reduceAxes) noexcept
Add a TopK layer to the network.
Definition: NvInfer.h:8481
IDequantizeLayer * addDequantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a dequantization layer to the network.
Definition: NvInfer.h:9269
bool unmarkOutputForShapes(ITensor &tensor) noexcept
Undo markOutputForShapes.
Definition: NvInfer.h:8879
IFillLayer * addFill(Dims const &dimensions, FillOperation op, DataType outputType) noexcept
Add a fill layer to the network.
Definition: NvInfer.h:9151
ILoop * addLoop() noexcept
Add a loop to the network.
Definition: NvInfer.h:9029
bool markUnfusedTensorsAsDebugTensors() noexcept
Mark unfused tensors as debug tensors.
Definition: NvInfer.h:8146
TRT_NODISCARD INormalizationLayer * addNormalizationV2(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9782
IActivationLayer * addActivation(ITensor &input, ActivationType type) noexcept
Add an activation layer to the network.
Definition: NvInfer.h:8180
TRT_DEPRECATED IFillLayer * addFill(Dims const &dimensions, FillOperation op) noexcept
Add a fill layer to the network.
Definition: NvInfer.h:9125
ISliceLayer * addSlice(ITensor &input, Dims const &start, Dims const &size, Dims const &stride) noexcept
Add a slice layer to the network.
Definition: NvInfer.h:8760
virtual ~INetworkDefinition() noexcept=default
TRT_DEPRECATED IQuantizeLayer * addQuantize(ITensor &input, ITensor &scale) noexcept
Add a quantization layer to the network.
Definition: NvInfer.h:9310
virtual IBuilder & getBuilder() const noexcept
Return the builder from which this INetworkDefinition was created.
Definition: NvInfer.h:9678
ILayer * getLayer(int32_t index) const noexcept
Get the layer specified by the given index.
Definition: NvInfer.h:8361
bool isDebugTensor(ITensor const &tensor) const noexcept
Check if a tensor is marked as debug tensor.
Definition: NvInfer.h:8124
bool getFlag(NetworkDefinitionCreationFlag networkDefinitionCreationFlag) const noexcept
Returns true if the network definition creation flag is set.
Definition: NvInfer.h:8850
IIfConditional * addIfConditional() noexcept
Add an if-then-else to the network.
Definition: NvInfer.h:9044
IErrorRecorder * getErrorRecorder() const noexcept
get the ErrorRecorder assigned to this interface.
Definition: NvInfer.h:9225
ISqueezeLayer * addSqueeze(ITensor &input, ITensor &axes) noexcept
Add a squeeze layer to the network.
Definition: NvInfer.h:9735
TRT_DEPRECATED INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9441
IReverseSequenceLayer * addReverseSequence(ITensor &input, ITensor &sequenceLens) noexcept
Add a ReverseSequence layer to the network.
Definition: NvInfer.h:9478
TRT_DEPRECATED IDynamicQuantizeLayer * addDynamicQuantize(ITensor &input, int32_t axis, int32_t blockSize, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9361
int32_t getNbInputs() const noexcept
Get the number of inputs in the network.
Definition: NvInfer.h:8373
NetworkDefinitionCreationFlags getFlags() const noexcept
Get the network definition creation flags for this network definition object. Defaults to 0.
Definition: NvInfer.h:8838
IQuantizeLayer * addQuantize(ITensor &input, ITensor &scale, DataType outputType) noexcept
Add a quantization layer to the network.
Definition: NvInfer.h:9334
IDynamicQuantizeLayer * addDynamicQuantizeV2(ITensor &input, Dims const &blockShape, DataType outputType, DataType scaleType) noexcept
Add a dynamic quantization layer to the network.
Definition: NvInfer.h:9385
IReduceLayer * addReduce(ITensor &input, ReduceOperation operation, uint32_t reduceAxes, bool keepDimensions) noexcept
Add a reduce layer to the network.
Definition: NvInfer.h:8445
IUnaryLayer * addUnary(ITensor &input, UnaryOperation operation) noexcept
Add a unary layer to the network.
Definition: NvInfer.h:8304
IGridSampleLayer * addGridSample(ITensor &input, ITensor &grid) noexcept
Add a GridSample layer to the network.
Definition: NvInfer.h:9419
void removeTensor(ITensor &tensor) noexcept
remove a tensor from the network definition.
Definition: NvInfer.h:8689
bool areWeightsMarkedRefittable(char const *name) const noexcept
Whether the weight has been marked as refittable.
Definition: NvInfer.h:9716
ISelectLayer * addSelect(ITensor &condition, ITensor &thenInput, ITensor &elseInput) noexcept
Add a select layer to the network.
Definition: NvInfer.h:9083
IScatterLayer * addScatter(ITensor &data, ITensor &indices, ITensor &updates, ScatterMode mode) noexcept
Add a Scatter layer to the network with specified mode and axis=0.
Definition: NvInfer.h:9289
TRT_DEPRECATED INormalizationLayer * addNormalization(ITensor &input, ITensor &scale, ITensor &bias, uint32_t axesMask) noexcept
Add a normalization layer to the network.
Definition: NvInfer.h:9510
int32_t getNbLayers() const noexcept
Get the number of layers in the network.
Definition: NvInfer.h:8347
TRT_DEPRECATED bool hasImplicitBatchDimension() const noexcept
Query whether the network was created with an implicit batch dimension.
Definition: NvInfer.h:8828
apiv::VNetworkDefinition * mImpl
Definition: NvInfer.h:9788
IKVCacheUpdateLayer * addKVCacheUpdate(ITensor &cache, ITensor &update, ITensor &writeIndices, KVCacheMode cacheMode) noexcept
Add a KVCacheUpdate layer to the network.
Definition: NvInfer.h:9618
bool markOutputForShapes(ITensor &tensor) noexcept
Enable tensor's value to be computed by IExecutionContext::getShapeBinding.
Definition: NvInfer.h:8867
IOneHotLayer * addOneHot(ITensor &indices, ITensor &values, ITensor &depth, int32_t axis) noexcept
Add a OneHot layer to the network.
Definition: NvInfer.h:8335
IScaleLayer * addScale(ITensor &input, ScaleMode mode, Weights shift, Weights scale, Weights power) noexcept
Add a Scale layer to the network.
Definition: NvInfer.h:8225
IPluginV3Layer * addPluginV3(ITensor *const *inputs, int32_t nbInputs, ITensor *const *shapeInputs, int32_t nbShapeInputs, IPluginV3 &plugin) noexcept
Add a plugin layer implementing the IPluginV3 interface to the network.
Definition: NvInfer.h:8740
void unmarkOutput(ITensor &tensor) noexcept
unmark a tensor as a network output.
Definition: NvInfer.h:8701
IIdentityLayer * addIdentity(ITensor &input) noexcept
Add an identity layer.
Definition: NvInfer.h:8659
IGatherLayer * addGatherV2(ITensor &data, ITensor &indices, GatherMode mode) noexcept
Add gather with specified mode, axis=0 and nbElementWiseDims=0.
Definition: NvInfer.h:8546
INonZeroLayer * addNonZero(ITensor &input, DataType indicesType) noexcept
Add a nonzero layer to the network.
Definition: NvInfer.h:8621
IElementWiseLayer * addElementWise(ITensor &input1, ITensor &input2, ElementWiseOperation op) noexcept
Add an elementwise layer to the network.
Definition: NvInfer.h:8282
IConstantLayer * addConstant(Dims const &dimensions, Weights weights) noexcept
Add a constant layer to the network.
Definition: NvInfer.h:8645
void setErrorRecorder(IErrorRecorder *recorder) noexcept
Set the ErrorRecorder for this interface.
Definition: NvInfer.h:9210
IPoolingLayer * addPoolingNd(ITensor &input, PoolingType type, Dims const &windowSize) noexcept
Add a multi-dimension pooling layer to the network.
Definition: NvInfer.h:8939
INMSLayer * addNMS(ITensor &boxes, ITensor &scores, ITensor &maxOutputBoxesPerClass, DataType indicesType) noexcept
Add a non-maximum suppression layer to the network.
Definition: NvInfer.h:9461
IRaggedSoftMaxLayer * addRaggedSoftMax(ITensor &input, ITensor &bounds) noexcept
Add a RaggedSoftMax layer to the network.
Definition: NvInfer.h:8565
IShapeLayer * addShape(ITensor &input) noexcept
Add a shape layer to the network.
Definition: NvInfer.h:8814
IGatherLayer * addGather(ITensor &data, ITensor &indices, int32_t axis) noexcept
Add gather with mode GatherMode::kDEFAULT and specified axis and nbElementWiseDims=0.
Definition: NvInfer.h:8530
IAttention * addAttention(ITensor &query, ITensor &key, ITensor &value, AttentionNormalizationOp normOp, bool causal) noexcept
Add an attention to the network.
Definition: NvInfer.h:9559
bool unmarkWeightsRefittable(char const *name) noexcept
Unmark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9703
bool markWeightsRefittable(char const *name) noexcept
Mark weights as refittable when the builder flag kREFIT_INDIVIDUAL is set.
Definition: NvInfer.h:9691
IRotaryEmbeddingLayer * addRotaryEmbedding(ITensor &input, ITensor &cosCache, ITensor &sinCache, bool interleaved, int32_t rotaryEmbeddingDim) noexcept
Add a Rotary Position Embedding (RoPE) layer to the network.
Definition: NvInfer.h:9584
IDeconvolutionLayer * addDeconvolutionNd(ITensor &input, int64_t nbOutputMaps, Dims kernelSize, Weights kernelWeights, Weights biasWeights) noexcept
Add a multi-dimension deconvolution layer to the network.
Definition: NvInfer.h:8961
IResizeLayer * addResize(ITensor &input) noexcept
Add a resize layer to the network.
Definition: NvInfer.h:9015
IUnsqueezeLayer * addUnsqueeze(ITensor &input, ITensor &axes) noexcept
Add an unsqueeze layer to the network.
Definition: NvInfer.h:9756
IMatrixMultiplyLayer * addMatrixMultiply(ITensor &input0, MatrixOperation op0, ITensor &input1, MatrixOperation op1) noexcept
Add a MatrixMultiply layer to the network.
Definition: NvInfer.h:8586
ISoftMaxLayer * addSoftMax(ITensor &input) noexcept
Add a SoftMax layer to the network.
Definition: NvInfer.h:8238
bool unmarkDebug(ITensor &tensor) noexcept
Unmark a tensor as a debug tensor.
Definition: NvInfer.h:8114
IEinsumLayer * addEinsum(ITensor *const *inputs, int32_t nbInputs, char const *equation) noexcept
Add an Einsum layer to the network.
Definition: NvInfer.h:9401
void markOutput(ITensor &tensor) noexcept
Mark a tensor as a network output.
Definition: NvInfer.h:8080
TRT_DEPRECATED IPluginV2Layer * addPluginV2(ITensor *const *inputs, int32_t nbInputs, IPluginV2 &plugin) noexcept
Add a plugin layer to the network using the IPluginV2 interface.
Definition: NvInfer.h:8722
IPaddingLayer * addPaddingNd(ITensor &input, Dims const &prePadding, Dims const &postPadding) noexcept
Add a padding layer to the network. Only 2D padding is currently supported.
Definition: NvInfer.h:9167
TRT_DEPRECATED IDequantizeLayer * addDequantize(ITensor &input, ITensor &scale) noexcept
Add a dequantization layer to the network.
Definition: NvInfer.h:9246
int32_t getNbOutputs() const noexcept
Get the number of outputs in the network.
Definition: NvInfer.h:8403
bool setWeightsName(Weights weights, char const *name) noexcept
Associate a name with all current uses of the given weights.
Definition: NvInfer.h:9191
TRT_NODISCARD IDistCollectiveLayer * addDistCollective(ITensor &input, CollectiveOperation distCollectiveOp, ReduceOperation reduceOp, int64_t root, int64_t *groups, int64_t groupSize) noexcept
Add a DistCollective layer to the network.
Definition: NvInfer.h:9666
IMoELayer * addMoE(ITensor &hiddenStates, ITensor &selectedExpertsForTokens, ITensor &scoresForSelectedExperts) noexcept
Add a MoE (Mixture of Experts) layer to the network.
Definition: NvInfer.h:9638
bool unmarkUnfusedTensorsAsDebugTensors() noexcept
Undo the marking of unfused tensors as debug tensors.
Definition: NvInfer.h:8160
Forward declaration of IEngineInspector for use by other interfaces.
Definition: NvInferRuntime.h:51
Definition: NvInfer.h:3872
DataType getIndicesType() const noexcept
Return the NonZero layer indices type.
Definition: NvInfer.h:3896
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3884
virtual ~INonZeroLayer() noexcept=default
A normalization layer in a network definition.
Definition: NvInfer.h:6586
float getEpsilon() const noexcept
Get the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6605
TRT_DEPRECATED void setComputePrecision(DataType type) noexcept
Set the compute precision of this layer.
Definition: NvInfer.h:6684
uint32_t getAxes() const noexcept
Get the axes value used for the normalization calculation.
Definition: NvInfer.h:6625
virtual ~INormalizationLayer() noexcept=default
void setEpsilon(float eps) noexcept
Set the epsilon value used for the normalization calculation.
Definition: NvInfer.h:6595
TRT_NODISCARD bool isV2() const noexcept
Returns true if this layer was created through addNormalizationV2().
Definition: NvInfer.h:6706
apiv::VNormalizationLayer * mImpl
Definition: NvInfer.h:6712
int64_t getNbGroups() const noexcept
Get the number of groups used to split the channels for the normalization calculation.
Definition: NvInfer.h:6656
void setAxes(uint32_t axesMask) noexcept
Set the reduction axes for the normalization calculation.
Definition: NvInfer.h:6615
void setNbGroups(int64_t nbGroups) noexcept
Set the number of groups used to split the channels in the normalization calculation.
Definition: NvInfer.h:6646
TRT_DEPRECATED DataType getComputePrecision() const noexcept
Get the compute precision of this layer.
Definition: NvInfer.h:6696
A OneHot layer in a network definition.
Definition: NvInfer.h:6208
virtual ~IOneHotLayer() noexcept=default
apiv::VOneHotLayer * mImpl
Definition: NvInfer.h:6229
void setAxis(int32_t axis) noexcept
Set the axis parameter.
Definition: NvInfer.h:6215
int32_t getAxis() const noexcept
Get the value of the axis parameter.
Definition: NvInfer.h:6223
Optimization profile for dynamic input dimensions and shape tensors.
Definition: NvInferRuntime.h:2675
Layer that represents a padding operation.
Definition: NvInfer.h:3069
Dims getPostPaddingNd() const noexcept
Get the padding that is applied at the end of the tensor.
Definition: NvInfer.h:3118
void setPrePaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the start of the tensor.
Definition: NvInfer.h:3080
virtual ~IPaddingLayer() noexcept=default
void setPostPaddingNd(Dims const &padding) noexcept
Set the padding that is applied at the end of the tensor.
Definition: NvInfer.h:3106
Dims getPrePaddingNd() const noexcept
Get the padding that is applied at the start of the tensor.
Definition: NvInfer.h:3092
apiv::VPaddingLayer * mImpl
Definition: NvInfer.h:3124
Layer that represents a parametric ReLU operation.
Definition: NvInfer.h:4087
apiv::VParametricReLULayer * mImpl
Definition: NvInfer.h:4089
virtual ~IParametricReLULayer() noexcept=default
Single registration point for all plugins in an application. It is used to find plugin implementation...
Definition: NvInferRuntimeCommon.h:56
Plugin class for user-implemented layers.
Definition: NvInferRuntimePlugin.h:139
Layer type for pluginV2.
Definition: NvInfer.h:2773
virtual ~IPluginV2Layer() noexcept=default
apiv::VPluginV2Layer * mImpl
Definition: NvInfer.h:2786
IPluginV2 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2780
Layer type for V3 plugins.
Definition: NvInfer.h:2800
virtual ~IPluginV3Layer() noexcept=default
IPluginV3 & getPlugin() noexcept
Get the plugin for the layer.
Definition: NvInfer.h:2807
apiv::VPluginV3Layer * mImpl
Definition: NvInfer.h:2813
A Pooling layer in a network definition.
Definition: NvInfer.h:1535
PoolingType getPoolingType() const noexcept
Get the type of activation to be performed.
Definition: NvInfer.h:1554
PaddingMode getPaddingMode() const noexcept
Get the padding mode.
Definition: NvInfer.h:1687
Dims getPostPadding() const noexcept
Get the padding.
Definition: NvInfer.h:1663
bool getAverageCountExcludesPadding() const noexcept
Get whether average pooling uses as a denominator the overlap area between the window and the unpadde...
Definition: NvInfer.h:1607
Dims getPrePadding() const noexcept
Get the pre-padding.
Definition: NvInfer.h:1635
void setPoolingType(PoolingType type) noexcept
Set the type of activation to be performed.
Definition: NvInfer.h:1544
void setWindowSizeNd(Dims const &windowSize) noexcept
Set the multi-dimension window size for pooling.
Definition: NvInfer.h:1700
void setPaddingMode(PaddingMode paddingMode) noexcept
Set the padding mode.
Definition: NvInfer.h:1676
Dims getWindowSizeNd() const noexcept
Get the multi-dimension window size for pooling.
Definition: NvInfer.h:1710
void setAverageCountExcludesPadding(bool exclusive) noexcept
Set whether average pooling uses as a denominator the overlap area between the window and the unpadde...
Definition: NvInfer.h:1596
void setPaddingNd(Dims const &padding) noexcept
Set the multi-dimension padding for pooling.
Definition: NvInfer.h:1754
float getBlendFactor() const noexcept
Get the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1582
void setStrideNd(Dims const &stride) noexcept
Set the multi-dimension stride for pooling.
Definition: NvInfer.h:1725
Dims getStrideNd() const noexcept
Get the multi-dimension stride for pooling.
Definition: NvInfer.h:1735
virtual ~IPoolingLayer() noexcept=default
Dims getPaddingNd() const noexcept
Get the multi-dimension padding for pooling.
Definition: NvInfer.h:1766
void setPostPadding(Dims const &padding) noexcept
Set the multi-dimension post-padding for pooling.
Definition: NvInfer.h:1653
void setPrePadding(Dims const &padding) noexcept
Set the multi-dimension pre-padding for pooling.
Definition: NvInfer.h:1625
void setBlendFactor(float blendFactor) noexcept
Set the blending factor for the max_average_blend mode: max_average_blendPool = (1-blendFactor)*maxPo...
Definition: NvInfer.h:1569
A Quantize layer in a network definition.
Definition: NvInfer.h:5584
void setToType(DataType toType) noexcept
Set the Quantize layer output type.
Definition: NvInfer.h:5645
bool setBlockShape(Dims const &blockShape) noexcept
Set the shape of the quantization block.
Definition: NvInfer.h:5618
void setAxis(int32_t axis) noexcept
Set the quantization axis.
Definition: NvInfer.h:5605
TRT_NODISCARD Dims getBlockShape() const noexcept
Get the shape of the quantization block.
Definition: NvInfer.h:5629
int32_t getAxis() const noexcept
Get the quantization axis.
Definition: NvInfer.h:5594
virtual ~IQuantizeLayer() noexcept=default
DataType getToType() const noexcept
Return the Quantize layer output type.
Definition: NvInfer.h:5657
A RaggedSoftmax layer in a network definition.
Definition: NvInfer.h:3921
apiv::VRaggedSoftMaxLayer * mImpl
Definition: NvInfer.h:3923
virtual ~IRaggedSoftMaxLayer() noexcept=default
A recurrence layer in a network definition.
Definition: NvInfer.h:4793
virtual ~IRecurrenceLayer() noexcept=default
Layer that represents a reduction across a non-bool tensor.
Definition: NvInfer.h:2989
void setKeepDimensions(bool keepDimensions) noexcept
Set the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:3036
void setOperation(ReduceOperation op) noexcept
Set the reduce operation for the layer.
Definition: NvInfer.h:2996
ReduceOperation getOperation() const noexcept
Get the reduce operation for the layer.
Definition: NvInfer.h:3006
virtual ~IReduceLayer() noexcept=default
uint32_t getReduceAxes() const noexcept
Get the axes over which to reduce for the layer.
Definition: NvInfer.h:3026
void setReduceAxes(uint32_t reduceAxes) noexcept
Set the axes over which to reduce.
Definition: NvInfer.h:3016
apiv::VReduceLayer * mImpl
Definition: NvInfer.h:3052
bool getKeepDimensions() const noexcept
Get the boolean that specifies whether or not to keep the reduced dimensions for the layer.
Definition: NvInfer.h:3046
A resize layer in a network definition.
Definition: NvInfer.h:4276
void setSelectorForSinglePixel(ResizeSelector selector) noexcept
Set coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4437
void setNearestRounding(ResizeRoundMode value) noexcept
Set rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4461
virtual ~IResizeLayer() noexcept=default
int32_t getScales(int32_t size, float *scales) const noexcept
Copies resize scales to scales[0, ..., nbScales-1], where nbScales is the number of scales that were ...
Definition: NvInfer.h:4355
void setOutputDimensions(Dims const &dimensions) noexcept
Set the output dimensions.
Definition: NvInfer.h:4296
void setCubicCoeff(float A) noexcept
Set the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4493
void setScales(float const *scales, int32_t nbScales) noexcept
Set the resize scales.
Definition: NvInfer.h:4336
float getCubicCoeff() const noexcept
Get the coefficient 'A' used in cubic interpolation.
Definition: NvInfer.h:4503
ResizeSelector getSelectorForSinglePixel() const noexcept
Get the coordinate selector function when resized to single pixel.
Definition: NvInfer.h:4447
InterpolationMode getResizeMode() const noexcept
Get resize mode for an input tensor.
Definition: NvInfer.h:4377
void setCoordinateTransformation(ResizeCoordinateTransformation coordTransform) noexcept
Set coordinate transformation function.
Definition: NvInfer.h:4412
void setExcludeOutside(bool excludeFlag) noexcept
Set the state for excluding outside pixels.
Definition: NvInfer.h:4516
void setResizeMode(InterpolationMode interpolationMode) noexcept
Set resize mode for an input tensor.
Definition: NvInfer.h:4367
Dims getOutputDimensions() const noexcept
Get the output dimensions.
Definition: NvInfer.h:4306
ResizeRoundMode getNearestRounding() const noexcept
Get rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4471
bool getExcludeOutside() const noexcept
Get the state for excluding outside pixels.
Definition: NvInfer.h:4526
ResizeCoordinateTransformation getCoordinateTransformation() const noexcept
Get coordinate transformation function.
Definition: NvInfer.h:4422
A ReverseSequence layer in a network definition.
Definition: NvInfer.h:6514
void setSequenceAxis(int32_t sequenceAxis) noexcept
Set the sequence axis. Default is 0.
Definition: NvInfer.h:6547
int32_t getBatchAxis() const noexcept
Return the batch axis. Return 1 if no batch axis was set.
Definition: NvInfer.h:6534
apiv::VReverseSequenceLayer * mImpl
Definition: NvInfer.h:6563
int32_t getSequenceAxis() const noexcept
Return the sequence axis. Return 0 if no sequence axis was set.
Definition: NvInfer.h:6557
void setBatchAxis(int32_t batchAxis) noexcept
Set the batch axis. Default is 1.
Definition: NvInfer.h:6524
virtual ~IReverseSequenceLayer() noexcept=default
Layer that implements Rotary Position Embedding (RoPE) (https://arxiv.org/abs/2104....
Definition: NvInfer.h:7393
TRT_NODISCARD int32_t getRotaryEmbeddingDim() const noexcept
Get the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7433
virtual ~IRotaryEmbeddingLayer() noexcept=default
void setInterleaved(bool interleaved) noexcept
Set whether the input is in interleaved format, i.e., whether the 2-d vectors rotated are taken from ...
Definition: NvInfer.h:7400
TRT_NODISCARD bool setRotaryEmbeddingDim(int32_t rotaryEmbeddingDim) noexcept
Set the number of hidden dimensions participating in RoPE. The default value is 0,...
Definition: NvInfer.h:7422
apiv::VRotaryEmbeddingLayer * mImpl
Definition: NvInfer.h:7456
TRT_NODISCARD bool getInterleaved() const noexcept
Get whether the input is in interleaved format. The default value is false.
Definition: NvInfer.h:7411
A Scale layer in a network definition.
Definition: NvInfer.h:1932
Weights getScale() const noexcept
Get the scale value.
Definition: NvInfer.h:1989
Weights getPower() const noexcept
Get the power value.
Definition: NvInfer.h:2009
void setScale(Weights scale) noexcept
Set the scale value.
Definition: NvInfer.h:1979
void setPower(Weights power) noexcept
Set the power value.
Definition: NvInfer.h:1999
ScaleMode getMode() const noexcept
Get the scale mode.
Definition: NvInfer.h:1949
void setShift(Weights shift) noexcept
Set the shift value.
Definition: NvInfer.h:1959
void setChannelAxis(int32_t channelAxis) noexcept
Set the channel axis.
Definition: NvInfer.h:2045
Weights getShift() const noexcept
Get the shift value.
Definition: NvInfer.h:1969
virtual ~IScaleLayer() noexcept=default
void setMode(ScaleMode mode) noexcept
Set the scale mode.
Definition: NvInfer.h:1939
int32_t getChannelAxis() const noexcept
Get the channel axis.
Definition: NvInfer.h:2024
A scatter layer in a network definition. Supports several kinds of scattering.
Definition: NvInfer.h:6136
void setMode(ScatterMode mode) noexcept
Set the scatter mode.
Definition: NvInfer.h:6143
apiv::VScatterLayer * mImpl
Definition: NvInfer.h:6177
void setAxis(int32_t axis) noexcept
Set the axis used by ScatterMode::kELEMENTS.
Definition: NvInfer.h:6163
int32_t getAxis() const noexcept
Get the axis.
Definition: NvInfer.h:6171
ScatterMode getMode() const noexcept
Get the scatter mode.
Definition: NvInfer.h:6153
virtual ~IScatterLayer() noexcept=default
Select elements from two data tensors based on a condition tensor.
Definition: NvInfer.h:5101
virtual ~ISelectLayer() noexcept=default
Layer type for getting shape of a tensor.
Definition: NvInfer.h:3594
virtual ~IShapeLayer() noexcept=default
apiv::VShapeLayer * mImpl
Definition: NvInfer.h:3596
Layer type for shuffling data.
Definition: NvInfer.h:3157
apiv::VShuffleLayer * mImpl
Definition: NvInfer.h:3315
void setFirstTranspose(Permutation permutation) noexcept
Set the permutation applied by the first transpose operation.
Definition: NvInfer.h:3168
void setSecondTranspose(Permutation permutation) noexcept
Set the permutation applied by the second transpose operation.
Definition: NvInfer.h:3268
Dims getReshapeDimensions() const noexcept
Get the reshaped dimensions.
Definition: NvInfer.h:3221
void setReshapeDimensions(Dims const &dimensions) noexcept
Set the reshaped dimensions.
Definition: NvInfer.h:3208
Permutation getFirstTranspose() const noexcept
Get the permutation applied by the first transpose operation.
Definition: NvInfer.h:3180
virtual ~IShuffleLayer() noexcept=default
Permutation getSecondTranspose() const noexcept
Get the permutation applied by the second transpose operation.
Definition: NvInfer.h:3280
bool getZeroIsPlaceholder() const noexcept
Get meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3309
void setZeroIsPlaceholder(bool zeroIsPlaceholder) noexcept
Set meaning of 0 in reshape dimensions.
Definition: NvInfer.h:3296
Slices an input tensor into an output tensor based on the offset and strides.
Definition: NvInfer.h:3409
void setStride(Dims const &stride) noexcept
Set the stride for computing the output slice data.
Definition: NvInfer.h:3478
apiv::VSliceLayer * mImpl
Definition: NvInfer.h:3577
virtual ~ISliceLayer() noexcept=default
void setSize(Dims const &size) noexcept
Set the dimensions of the output slice.
Definition: NvInfer.h:3449
void setAxes(Dims const &axes) noexcept
Set the axes for this ISliceLayer.
Definition: NvInfer.h:3556
void setStart(Dims const &start) noexcept
Set the start offset that the slice layer uses to create the output slice.
Definition: NvInfer.h:3420
Dims getStart() const noexcept
Get the start offset for the slice layer.
Definition: NvInfer.h:3435
void setMode(SampleMode mode) noexcept
Set the slice mode.
Definition: NvInfer.h:3503
Dims getSize() const noexcept
Get dimensions of the output slice.
Definition: NvInfer.h:3464
SampleMode getMode() const noexcept
Get the slice mode.
Definition: NvInfer.h:3513
Dims getStride() const noexcept
Get the stride for the output slice.
Definition: NvInfer.h:3493
Dims getAxes() const noexcept
Get the axes for this ISliceLayer.
Definition: NvInfer.h:3571
A Softmax layer in a network definition.
Definition: NvInfer.h:2076
void setAxes(uint32_t axes) noexcept
Set the axis along which softmax is computed. Currently, only one axis can be set.
Definition: NvInfer.h:2098
uint32_t getAxes() const noexcept
Get the axis along which softmax occurs.
Definition: NvInfer.h:2108
virtual ~ISoftMaxLayer() noexcept=default
Layer that represents a squeeze operation, removing unit dimensions of the first input tensor on a se...
Definition: NvInfer.h:6726
virtual ~ISqueezeLayer() noexcept=default
apiv::VSqueezeLayer * mImpl
Definition: NvInfer.h:6743
A tensor in a network definition.
Definition: NvInfer.h:189
void setAllowedFormats(TensorFormats formats) noexcept
Set allowed formats for an input or output tensor. By default all formats are allowed....
Definition: NvInfer.h:459
TensorLocation getLocation() const noexcept
Get the storage location of a tensor.
Definition: NvInfer.h:378
void setDimensions(Dims const &dimensions) noexcept
Set the dimensions of a tensor.
Definition: NvInfer.h:237
void resetDynamicRange() noexcept
Undo effect of setDynamicRange.
Definition: NvInfer.h:417
void setName(char const *name) noexcept
Set the tensor name.
Definition: NvInfer.h:206
bool isExecutionTensor() const noexcept
Whether the tensor is an execution tensor.
Definition: NvInfer.h:524
TRT_DEPRECATED bool dynamicRangeIsSet() const noexcept
Query whether dynamic range is set.
Definition: NvInfer.h:409
char const * getName() const noexcept
Get the tensor name.
Definition: NvInfer.h:218
bool isShapeTensor() const noexcept
Whether the tensor is a shape tensor.
Definition: NvInfer.h:503
float getDynamicRangeMax() const noexcept
Get maximum of dynamic range.
Definition: NvInfer.h:437
bool isNetworkInput() const noexcept
Whether the tensor is a network input.
Definition: NvInfer.h:327
TRT_DEPRECATED void setBroadcastAcrossBatch(bool broadcastAcrossBatch) noexcept
Set whether to enable broadcast of tensor across the implicit batch dimension.
Definition: NvInfer.h:352
TRT_DEPRECATED bool setDynamicRange(float min, float max) noexcept
Set dynamic range for the tensor.
Definition: NvInfer.h:319
TRT_DEPRECATED void setType(DataType type) noexcept
Set the data type of a tensor.
Definition: NvInfer.h:287
TRT_DEPRECATED bool getBroadcastAcrossBatch() const noexcept
Check if tensor is broadcast across the implicit batch dimension.
Definition: NvInfer.h:366
bool isNetworkOutput() const noexcept
Whether the tensor is a network output.
Definition: NvInfer.h:335
DataType getType() const noexcept
Get the data type of a tensor.
Definition: NvInfer.h:302
apiv::VTensor * mImpl
Definition: NvInfer.h:571
float getDynamicRangeMin() const noexcept
Get minimum of dynamic range.
Definition: NvInfer.h:427
virtual ~ITensor() noexcept=default
void setDimensionName(int32_t index, char const *name) noexcept
Name a dimension of an input tensor.
Definition: NvInfer.h:550
char const * getDimensionName(int32_t index) const noexcept
Get the name of an input dimension.
Definition: NvInfer.h:565
TRT_DEPRECATED void setLocation(TensorLocation location) noexcept
Set the storage location of a tensor.
Definition: NvInfer.h:397
Dims getDimensions() const noexcept
Get the dimensions of a tensor.
Definition: NvInfer.h:251
TensorFormats getAllowedFormats() const noexcept
Get a bitmask of TensorFormat values that the tensor supports. For a shape tensor,...
Definition: NvInfer.h:472
Class to handle tactic timing info collected from builder.
Definition: NvInfer.h:10719
int64_t queryKeys(TimingCacheKey *keyBuffer, int64_t capacity) const noexcept
Query cache keys from Timing Cache.
Definition: NvInfer.h:10785
bool combine(ITimingCache const &inputCache, bool ignoreMismatch) noexcept
Combine input timing cache into local instance.
Definition: NvInfer.h:10756
TimingCacheValue query(TimingCacheKey const &key) const noexcept
Query value in a cache entry.
Definition: NvInfer.h:10802
virtual ~ITimingCache() noexcept=default
bool update(TimingCacheKey const &key, TimingCacheValue const &value) noexcept
Update values in a cache entry.
Definition: NvInfer.h:10824
apiv::VTimingCache * mImpl
Definition: NvInfer.h:10830
bool reset() noexcept
Empty the timing cache.
Definition: NvInfer.h:10766
Layer that represents a TopK reduction.
Definition: NvInfer.h:3634
void setK(int32_t k) noexcept
Set the static k value for the layer.
Definition: NvInfer.h:3665
void setReduceAxes(uint32_t reduceAxes) noexcept
Set which axes to reduce for the layer.
Definition: NvInfer.h:3689
TopKOperation getOperation() const noexcept
Get the operation for the layer.
Definition: NvInfer.h:3651
apiv::VTopKLayer * mImpl
Definition: NvInfer.h:3748
void setOperation(TopKOperation op) noexcept
Set the operation for the layer.
Definition: NvInfer.h:3641
bool setIndicesType(DataType type) noexcept
Set the indices type for the layer.
Definition: NvInfer.h:3730
int32_t getK() const noexcept
Get the k value for the layer.
Definition: NvInfer.h:3679
uint32_t getReduceAxes() const noexcept
Get the axes to reduce for the layer.
Definition: NvInfer.h:3699
virtual ~ITopKLayer() noexcept=default
DataType getIndicesType() const noexcept
Return the TopK layer indices type.
Definition: NvInfer.h:3742
A layer that represents a trip-count limiter.
Definition: NvInfer.h:4914
TripLimit getTripLimit() const noexcept
Get a trip limiter type.
Definition: NvInfer.h:4919
virtual ~ITripLimitLayer() noexcept=default
Layer that represents an unary operation.
Definition: NvInfer.h:2881
void setOperation(UnaryOperation op) noexcept
Set the unary operation for the layer.
Definition: NvInfer.h:2890
apiv::VUnaryLayer * mImpl
Definition: NvInfer.h:2906
UnaryOperation getOperation() const noexcept
Get the unary operation for the layer.
Definition: NvInfer.h:2900
virtual ~IUnaryLayer() noexcept=default
Layer that represents an unsqueeze operation, which reshapes the first input tensor by inserting unit...
Definition: NvInfer.h:6756
virtual ~IUnsqueezeLayer() noexcept=default
apiv::VUnsqueezeLayer * mImpl
Definition: NvInfer.h:6774
An Interface class for version control.
Definition: NvInferRuntimeBase.h:279
Version information associated with a TRT interface.
Definition: NvInferRuntimeBase.h:244
An array of weights used as a layer parameter.
Definition: NvInferRuntime.h:124
Definition: NvInfer.h:10310
virtual int32_t selectAlgorithms(IAlgorithmContext const &context, IAlgorithm const *const *choices, int32_t nbChoices, int32_t *selection) noexcept=0
Select Algorithms for a layer from the given list of algorithm choices.
virtual void reportAlgorithms(IAlgorithmContext const *const *algoContexts, IAlgorithm const *const *algoChoices, int32_t nbAlgorithms) noexcept=0
Called by TensorRT to report choices it made.
InterfaceInfo getInterfaceInfo() const noexcept override
Return version information associated with this interface. Applications must not override this method...
Definition: NvInfer.h:10315
virtual ~IAlgorithmSelector() noexcept=default
Definition: NvInferRuntimeBase.h:416
Definition: NvInferRuntime.h:1656
Definition: NvInfer.h:9937
~IInt8EntropyCalibrator2() noexcept override=default
CalibrationAlgoType getAlgorithm() noexcept override
Definition: NvInfer.h:9950
InterfaceInfo getInterfaceInfo() const noexcept override
Return version information associated with this interface. Applications must not override this method...
Definition: NvInfer.h:9942
Definition: NvInfer.h:9897
CalibrationAlgoType getAlgorithm() noexcept override
Definition: NvInfer.h:9910
InterfaceInfo getInterfaceInfo() const noexcept override
Return version information associated with this interface. Applications must not override this method...
Definition: NvInfer.h:9902
~IInt8EntropyCalibrator() noexcept override=default
Definition: NvInfer.h:10016
CalibrationAlgoType getAlgorithm() noexcept override
Definition: NvInfer.h:10029
InterfaceInfo getInterfaceInfo() const noexcept override
Return version information associated with this interface. Applications must not override this method...
Definition: NvInfer.h:10021
virtual double getQuantile() const noexcept=0
The quantile (between 0 and 1) that will be used to select the region maximum when the quantile metho...
Definition: NvInfer.h:9977
~IInt8MinMaxCalibrator() noexcept override=default
CalibrationAlgoType getAlgorithm() noexcept override
Definition: NvInfer.h:9990
InterfaceInfo getInterfaceInfo() const noexcept override
Return version information associated with this interface. Applications must not override this method...
Definition: NvInfer.h:9982
Definition: NvInferPluginBase.h:206
Definition: NvInfer.h:11055
virtual bool stepComplete(char const *phaseName, int32_t step) noexcept=0
Signal that a step of an optimizer phase has finished.
virtual ~IProgressMonitor() noexcept=default
IProgressMonitor()=default
virtual void phaseFinish(char const *phaseName) noexcept=0
Signal that a phase of the optimizer has finished.
virtual void phaseStart(char const *phaseName, char const *parentPhase, int32_t nbSteps) noexcept=0
Signal that a phase of the optimizer has started.
Definition: NvInferRuntime.h:666
IBuilder * createInferBuilder(ILogger &logger) noexcept
Create an instance of an IBuilder class.
Definition: NvInfer.h:12509
The TensorRT API version 1 namespace.
Definition: NvInferSafePlugin.h:33
uint32_t TacticSources
Represents a collection of one or more TacticSource values combine using bitwise-OR operations.
Definition: NvInferRuntime.h:2961
ResizeSelector
The coordinate selector when resize to single pixel output.
Definition: NvInfer.h:4181
@ kFORMULA
Use formula to map the original index.
@ kUPPER
Select the upper left pixel.
EngineCapability
List of supported engine capability flows.
Definition: NvInferRuntime.h:76
MemoryPoolType
The type for memory pools used by TensorRT.
Definition: NvInfer.h:10841
TENSORRTAPI bool setInternalLibraryPath(AsciiChar const *path) noexcept
Set a custom directory path for loading internal TensorRT libraries when building engines.
ScaleMode
Controls how shift, scale and power are applied in a Scale layer.
Definition: NvInfer.h:1889
@ kUNIFORM
Identical coefficients across all elements of the tensor.
@ kCHANNEL
Per-channel coefficients.
RuntimePlatform
Describes the intended runtime platform (operating system and CPU architecture) for the execution of ...
Definition: NvInfer.h:10421
uint32_t QuantizationFlags
Represents one or more QuantizationFlag values using binary OR operations.
Definition: NvInfer.h:10373
HardwareCompatibilityLevel
Describes requirements of compatibility with GPU architectures other than that of the GPU on which th...
Definition: NvInfer.h:10969
@ kSAME_COMPUTE_CAPABILITY
CumulativeOperation
Enumerates the cumulative operations that may be performed by a Cumulative layer.
Definition: NvInfer.h:6790
BoundingBoxFormat
Representation of bounding box data used for the Boxes input tensor in INMSLayer.
Definition: NvInfer.h:6328
@ kCENTER_SIZES
(x_center, y_center, width, height) where (x_center, y_center) is the center point of the box
@ kCORNER_PAIRS
(x1, y1, x2, y2) where (x1, y1) and (x2, y2) are any pair of diagonal corners
constexpr int32_t EnumMax< BuilderFlag >() noexcept
Definition: NvInfer.h:10659
constexpr int32_t EnumMax< LayerType >() noexcept
Definition: NvInfer.h:124
constexpr int32_t EnumMax< CalibrationAlgoType >() noexcept
Definition: NvInfer.h:9812
UnaryOperation
Enumerates the unary operations that may be performed by a Unary layer.
Definition: NvInfer.h:2834
@ kISINF
Return true if input value equals +/- infinity for floating-point data type.
@ kCOSH
Hyperbolic cosine.
@ kACOSH
Inverse hyperbolic cosine.
@ kERF
Gauss error function.
@ kISNAN
Return true if input value is a NaN for floating-point data type.
@ kROUND
Round to nearest even for floating-point data type.
@ kATANH
Inverse hyperbolic tangent.
@ kASINH
Inverse hyperbolic sine.
@ kSIGN
Sign, If input > 0, output 1; if input < 0, output -1; if input == 0, output 0.
constexpr int32_t EnumMax< ReduceOperation >() noexcept
Definition: NvInfer.h:2949
constexpr int32_t EnumMax< TripLimit >() noexcept
Definition: NvInfer.h:4582
ActivationType
Enumerates the types of activation to perform in an activation layer.
Definition: NvInfer.h:143
@ kSELU
Selu activation: x>0 ? beta * x : beta * (alpha*exp(x) - alpha)
@ kSCALED_TANH
Scaled tanh activation: alpha*tanh(beta*x)
@ kRELU
Rectified linear activation.
@ kELU
Elu activation: x>=0 ? x : alpha * (exp(x) - 1).
@ kLEAKY_RELU
LeakyRelu activation: x>=0 ? x : alpha * x.
@ kSOFTSIGN
Softsign activation: x / (1+|x|)
@ kHARD_SIGMOID
Hard sigmoid activation: max(0, min(1, alpha*x+beta))
@ kTHRESHOLDED_RELU
Thresholded ReLU activation: x>alpha ? x : 0.
@ kSIGMOID
Sigmoid activation.
@ kCLIP
Clip activation: max(alpha, min(beta, x))
@ kGELU_TANH
GELU tanh activation: 0.5 * x * (1 + tanh(sqrt(2/pi) * (0.044715F * pow(x, 3) + x)))
@ kGELU_ERF
GELU erf activation: 0.5 * x * (1 + erf(sqrt(0.5) * x))
@ kSOFTPLUS
Parametric softplus activation: alpha*log(exp(beta*x)+1)
FillOperation
Enumerates the tensor fill operations that may performed by a fill layer.
Definition: NvInfer.h:5162
ResizeRoundMode
The rounding mode for nearest neighbor resize.
Definition: NvInfer.h:4211
@ kHALF_DOWN
Round half down.
char_t AsciiChar
Definition: NvInferRuntimeBase.h:116
PaddingMode
Enumerates the modes of padding to perform in convolution, deconvolution and pooling layer,...
Definition: NvInfer.h:1067
@ kSAME_LOWER
Use SAME padding, with prePadding >= postPadding.
@ kEXPLICIT_ROUND_DOWN
Use explicit padding, rounding output size down.
@ kEXPLICIT_ROUND_UP
Use explicit padding, rounding output size up.
@ kSAME_UPPER
Use SAME padding, with prePadding <= postPadding.
TripLimit
Enum that describes kinds of trip limits.
Definition: NvInfer.h:4570
@ kWHILE
Tensor is a scalar of type kBOOL. Loop terminates when value is false.
@ kCOUNT
Tensor is a scalar of type kINT32 or kINT64 that contains the trip count.
uint32_t NetworkDefinitionCreationFlags
Represents one or more NetworkDefinitionCreationFlag flags using binary OR operations....
Definition: NvInfer.h:12103
PreviewFeature
Define preview features.
Definition: NvInfer.h:10916
@ kRUNTIME_ACTIVATION_RESIZE_10_10
@ kMULTIDEVICE_RUNTIME_10_16
@ kALIASED_PLUGIN_IO_10_03
TilingOptimizationLevel
Define the optimization levels for Tiling.
Definition: NvInfer.h:11022
@ kFAST
Use a fast algorithm and heuristic based strategy. Slightly increases engine build time.
@ kFULL
Increase search space even wider. Significantly increases engine build time.
constexpr int32_t EnumMax< GatherMode >() noexcept
Definition: NvInfer.h:2593
DataType
The type of weights and tensors. The datatypes other than kBOOL, kINT32, and kINT64 are "activation d...
Definition: NvInferRuntimeBase.h:146
uint32_t BuilderFlags
Represents one or more BuilderFlag values using binary OR operations, e.g., 1U << BuilderFlag::kFP16 ...
Definition: NvInfer.h:10453
DeviceType
The device that this layer/network will execute on.
Definition: NvInferRuntime.h:1350
constexpr int32_t EnumMax< ScaleMode >() noexcept
Definition: NvInfer.h:1901
CalibrationAlgoType
Version of calibration algorithm to use.
Definition: NvInfer.h:9799
@ kENTROPY_CALIBRATION_2
Entropy calibration.
@ kLEGACY_CALIBRATION
Legacy calibration.
@ kENTROPY_CALIBRATION
Legacy entropy calibration.
@ kMINMAX_CALIBRATION
Minmax calibration.
LayerType
The type values of layer classes.
Definition: NvInfer.h:58
@ kGRID_SAMPLE
Grid sample layer.
@ kRAGGED_SOFTMAX
Ragged softmax layer.
@ kDECONVOLUTION
Deconvolution layer.
@ kASSERTION
Assertion layer.
@ kATTENTION_INPUT
Attention Input.
@ kMATRIX_MULTIPLY
Matrix multiply layer.
@ kCONDITION
Condition layer.
@ kCUMULATIVE
Cumulative layer.
@ kCONDITIONAL_INPUT
Conditional Input layer.
@ kIDENTITY
Identity layer.
@ kNORMALIZATION
Normalization layer.
@ kQUANTIZE
Quantize layer.
@ kCONVOLUTION
Convolution layer.
@ kPARAMETRIC_RELU
Parametric ReLU layer.
@ kATTENTION_OUTPUT
Attention Output.
@ kUNSQUEEZE
Unsqueeze Layer.
@ kCONCATENATION
Concatenation layer.
@ kREVERSE_SEQUENCE
Reverse sequence layer.
@ kROTARY_EMBEDDING
Rotary Embedding layer.
@ kRECURRENCE
Loop Recurrence layer.
@ kDEQUANTIZE
Dequantize layer.
@ kPLUGIN_V3
PluginV3 layer.
@ kITERATOR
Loop Iterator layer.
@ kTRIP_LIMIT
Loop Trip limit layer.
@ kDYNAMIC_QUANTIZE
Dynamic Quantize layer.
@ kUNARY
UnaryOp operation Layer.
@ kACTIVATION
Activation layer.
@ kELEMENTWISE
Elementwise layer.
@ kPLUGIN_V2
PluginV2 layer.
@ kLOOP_OUTPUT
Loop output layer.
@ kCONDITIONAL_OUTPUT
Conditional Output layer.
@ kCONSTANT
Constant layer.
@ kNON_ZERO
NonZero layer.
@ kKVCACHE_UPDATE
KV Cache Update layer.
@ kDIST_COLLECTIVE
DistCollective layer.
constexpr int32_t EnumMax< QuantizationFlag >() noexcept
Definition: NvInfer.h:10398
SampleMode
Controls how ISliceLayer and IGridSample handle out-of-bounds coordinates.
Definition: NvInfer.h:3325
@ kCLAMP
Out of bounds indices are clamped to bounds.
@ kSTRICT_BOUNDS
Fail with error when the coordinates are out of bounds.
@ kWRAP
Coordinates wrap around periodically.
GatherMode
Control form of IGatherLayer.
Definition: NvInfer.h:2581
@ kDEFAULT
Similar to ONNX Gather.
@ kELEMENT
Similar to ONNX GatherElements.
@ kND
Similar to ONNX GatherND.
MoEActType
Enumerates the activation type for the MoE layer.
Definition: NvInfer.h:7554
uint32_t TensorFormats
It is capable of representing one or more TensorFormat by binary OR operations, e....
Definition: NvInfer.h:135
ProfilingVerbosity
List of verbosity levels of layer information exposed in NVTX annotations and in IEngineInspector.
Definition: NvInferRuntime.h:2973
NetworkDefinitionCreationFlag
List of immutable network properties expressed at network creation time. NetworkDefinitionCreationFla...
Definition: NvInfer.h:12114
@ kPREFER_JIT_PYTHON_PLUGINS
@ kPREFER_AOT_PYTHON_PLUGINS
ElementWiseOperation
Enumerates the binary operations that may be performed by an ElementWise layer.
Definition: NvInfer.h:2491
@ kSUB
Subtract the second element from the first.
@ kSUM
Sum of the two elements.
@ kPROD
Product of the two elements.
@ kFLOOR_DIV
Floor division of the first element by the second.
@ kEQUAL
Check if two elements are equal.
@ kAND
Logical AND of two elements.
@ kOR
Logical OR of two elements.
@ kMIN
Minimum of the two elements.
@ kPOW
The first element to the power of the second element.
@ kLESS
Check if element in first tensor is less than corresponding element in second tensor.
@ kGREATER
Check if element in first tensor is greater than corresponding element in second tensor.
@ kXOR
Logical XOR of two elements.
@ kDIV
Divide the first element by the second.
QuantizationFlag
List of valid flags for quantizing the network to int8.
Definition: NvInfer.h:10385
@ kCALIBRATE_BEFORE_FUSION
CollectiveOperation
Enumerates the collective operations that may be performed by a DistCollective layer.
Definition: NvInfer.h:2962
@ kREDUCE_SCATTER
Reduce scatter.
constexpr int32_t EnumMax< SampleMode >() noexcept
Definition: NvInfer.h:3341
InterpolationMode
Enumerates various modes of interpolation.
Definition: NvInfer.h:4099
@ kNEAREST
ND (0 < N <= 8) nearest neighbor resizing.
@ kCUBIC
Supports bicubic (2D) interpolation.
@ kLINEAR
Supports linear (1D), bilinear (2D), and trilinear (3D) interpolation.
BuilderFlag
List of valid modes that the builder can enable when creating an engine from a network definition.
Definition: NvInfer.h:10463
@ kWEIGHT_STREAMING
Enable weight streaming for the current engine.
@ kGPU_FALLBACK
Enable layers marked to execute on GPU if layer cannot execute on DLA.
@ kSPARSE_WEIGHTS
Allow the builder to examine weights and use optimized functions when weights have suitable sparsity.
@ kERROR_ON_TIMING_CACHE_MISS
@ kEDITABLE_TIMING_CACHE
Enable editable timing cache.
@ kPREFER_PRECISION_CONSTRAINTS
@ kDISTRIBUTIVE_INDEPENDENCE
@ kSTRIP_PLAN
Strip the refittable weights from the engine plan file.
@ kMONITOR_MEMORY
Enable memory monitor during build time.
@ kDISABLE_TIMING_CACHE
Disable reuse of timing information across identical layers.
@ kDISABLE_COMPILATION_CACHE
@ kREFIT
Enable building a refittable engine.
@ kOBEY_PRECISION_CONSTRAINTS
@ kREJECT_EMPTY_ALGORITHMS
constexpr int32_t EnumMax< TopKOperation >() noexcept
Definition: NvInfer.h:3617
TENSORRTAPI nvinfer1::IPluginRegistry * getBuilderPluginRegistry(nvinfer1::EngineCapability capability) noexcept
Return the plugin registry for building a Standard engine, or nullptr if no registry exists.
constexpr int32_t EnumMax< MemoryPoolType >() noexcept
Definition: NvInfer.h:10902
TopKOperation
Enumerates the operations that may be performed by a TopK layer.
Definition: NvInfer.h:3606
ReduceOperation
Enumerates the reduce operations that may be performed by a Reduce layer.
Definition: NvInfer.h:2934
@ kAVG
Average of the elements.
constexpr int32_t EnumMax< LoopOutput >() noexcept
Definition: NvInfer.h:4559
constexpr int32_t EnumMax< NetworkDefinitionCreationFlag >() noexcept
Definition: NvInfer.h:12142
TRT_DEPRECATED_API nvinfer1::safe::IPluginRegistry * getBuilderSafePluginRegistry(nvinfer1::EngineCapability capability) noexcept
Return the plugin registry for building a Safety engine, or nullptr if no registry exists.
ScatterMode
Control form of IScatterLayer.
Definition: NvInfer.h:6062
MatrixOperation
Enumerates the operations that may be performed on a tensor by IMatrixMultiplyLayer before multiplica...
Definition: NvInfer.h:3759
@ kTRANSPOSE
Like kNONE, but transpose the matrix dimensions.
ResizeCoordinateTransformation
The resize coordinate transformation function.
Definition: NvInfer.h:4127
constexpr int32_t EnumMax< UnaryOperation >() noexcept
Definition: NvInfer.h:2868
LoopOutput
Enum that describes kinds of loop outputs.
Definition: NvInfer.h:4542
@ kLAST_VALUE
Output value is value of tensor for last iteration.
@ kCONCATENATE
Output value is concatenation of values of tensor for each iteration, in forward order.
@ kREVERSE
Output value is concatenation of values of tensor for each iteration, in reverse order.
constexpr int32_t EnumMax< BoundingBoxFormat >() noexcept
Definition: NvInfer.h:6341
constexpr int32_t EnumMax< MatrixOperation >() noexcept
Definition: NvInfer.h:3787
KVCacheMode
Enumerates the KVCache modes that may be performed by a KVCacheUpdate layer.
Definition: NvInfer.h:7466
PoolingType
The type of pooling to perform in a pooling layer.
Definition: NvInfer.h:1503
@ kAVERAGE
Average over elements. If the tensor is padded, the count includes the padding.
@ kMAX
Maximum over elements.
@ kMAX_AVERAGE_BLEND
Blending between max and average pooling: (1-blendFactor)*maxPool + blendFactor*avgPool.
v_1_0::IProgressMonitor IProgressMonitor
Definition: NvInfer.h:11138
constexpr int32_t EnumMax< FillOperation >() noexcept
Definition: NvInfer.h:5196
TensorLocation
The location for tensor data storage, device or host.
Definition: NvInferRuntime.h:204
OptProfileSelector
When setting or querying optimization profile parameters (such as shape tensor inputs or dynamic dime...
Definition: NvInferRuntime.h:2635
AttentionNormalizationOp
Enumerates the operations that may be performed by the normalization in the attention subgraph.
Definition: NvInfer.h:6925
constexpr int32_t EnumMax< ScatterMode >() noexcept
Definition: NvInfer.h:6073
Represents a permutation of dimensions.
Definition: NvInfer.h:3134
Declaration of EnumMaxImpl struct to store maximum number of elements in an enumeration type.
Definition: NvInferRuntimeBase.h:129
The key to retrieve timing cache entries.
Definition: NvInfer.h:10683
Definition: NvInfer.h:10695
uint64_t tacticHash
Hash of the selected tactic.
Definition: NvInfer.h:10697
float timingMSec
Timing of this tactic in milliseconds. Negative numbers and NaN are invalid values.
Definition: NvInfer.h:10699