aws-neuron/Qwen3-1.7B-TP4-BS4-SEQ2048
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: aws-neuron/Qwen3-1.7B-TP4-BS4-SEQ2048
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-30 21:56:23 +08:00
commit fd57cdfd17
114 changed files with 237628 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
token_generation_model/_tp0_bk4/command.txt Normal file
View File

@@ -0,0 +1 @@
neuronx-cc compile --framework=XLA model.MODULE_b4b44a47076a167bd9a5+795ee7cc.hlo_module.pb --output model.MODULE_b4b44a47076a167bd9a5+795ee7cc.neff --target=trn2 --auto-cast=none --model-type=transformer '--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=1 --vectorize-strided-dma ' --lnc=2 -O2 --internal-hlo2tensorizer-options=--verify-hlo=true --logfile=log-neuron-cc.txt --verbose=35

1
token_generation_model/_tp0_bk4/compile_flags.MODULE_b4b44a47076a167bd9a5+795ee7cc.json Normal file
View File

@@ -0,0 +1 @@
["--target=trn2", "--auto-cast=none", "--model-type=transformer", "--tensorizer-options=--enable-ccop-compute-overlap --cc-pipeline-tiling-factor=1 --vectorize-strided-dma ", "--lnc=2", "-O2", "--internal-hlo2tensorizer-options=--verify-hlo=true", "--logfile=/models/Qwen3-1.7B-compiled/token_generation_model/_tp0_bk4/log-neuron-cc.txt"]

590
token_generation_model/_tp0_bk4/global_metric_store.json Normal file
View File

@@ -0,0 +1,590 @@
{
"Average": {
"tensorizer": {
"StaticProfiler::AverageFractalPeUtilization": 98.06681823730469,
"StaticProfiler::AveragePartitionUtilization": 88.63555145263672,
"StaticProfiler::AveragePeUtilization": 75.70149230957031,
"StaticProfiler::LocalizationEfficiency": 131.45083618164063,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 134.9540252685547,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 0,
"TilingProfiler::AveragePeUtilizationAfterTiling": 0
}
},
"Count": {
"tensorizer": {
"StaticProfiler::AverageFractalPeUtilization": 1,
"StaticProfiler::AveragePartitionUtilization": 1,
"StaticProfiler::AveragePeUtilization": 1,
"StaticProfiler::LocalizationEfficiency": 1,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 1,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 1,
"TilingProfiler::AveragePeUtilizationAfterTiling": 1
}
},
"Sum": {
"compiletime": {
"AGOrderingAnalysisPass": 5.95287561416626,
"AffinePredicateResolution": 0.49277400970458984,
"AliasDependencyElimination": 0.01732039451599121,
"AliasDependencyInduction": 18.183685302734375,
"AliasDependencyReset": 18.999177932739258,
"BFComputeCutting": 0.3126344680786133,
"BirCodeGenLoop": 1.8802330493927002,
"CCOpFusion": 0.8809163570404053,
"CanonicalizeConv": 0.0,
"CanonicalizeDAGForPGTiling": 0.17032384872436523,
"CanonicalizeForTensorizer": 0.0005619999719783664,
"CanonicalizeIR": 0.7969982624053955,
"Canonicalizer": 0.011158999986946583,
"CoalesceCCOp": 0.6349403858184814,
"CommuteConcat": 0.02859354019165039,
"DMALocalityOpt": 0.03904151916503906,
"DMAProfiler": 0.14334964752197266,
"DMATilingProfiler": 0.08906126022338867,
"DataLocalityOpt": 4.2657470703125,
"DataStreaming": 0.1790633201599121,
"DeConcat": 0.06563258171081543,
"DeadCodeElimination": 0.0297544002532959,
"DeadStoreElimination": 1.2178454399108887,
"DelinearIndices": 0.35585713386535645,
"Delinearization": 0.6069140434265137,
"DelinearizeSPMD": 0.6448159217834473,
"DoNothing": 0.0005474090576171875,
"DramToDramTranspose": 0.3358180522918701,
"DumpGraphAndMetadata": 0.17146539688110352,
"EliminateDivs": 1.3107364177703857,
"ExpandBatchNorm": 0.48049426078796387,
"ExpandISAMacro": 0.08499646186828613,
"FactorizeBlkDims": 0.6006636619567871,
"FactorizeThreadAxesInFreeDims": 0.09288716316223145,
"FlattenMacroLoop": 0.0829305648803711,
"GenericAccessSimplifier": 0.026407241821289063,
"HoistCompute": 9.699999645818025e-05,
"IdentifyCrossPassTensors": 0.00019700000120792538,
"InferInitValue": 1.2918815612792969,
"InferIntrinsicOnCC": 0.31558942794799805,
"InferNeuronTensor": 1.773860216140747,
"InferNonlocalTensors": 6.451204299926758,
"InferPSumTensor": 1.442307472229004,
"InferShardAxis": 10.398513793945313,
"InferSharedMemLoc": 0.12981677055358887,
"InlineNativeKernels": 0.051079511642456055,
"InsertCoreBarrier": 0.3666720390319824,
"InsertIOTransposes": 1.088529348373413,
"InsertImplicitShardAxisBeforeISel": 0.4208717346191406,
"InsertLocalTransposes": 0.6978082656860352,
"InsertOffloadedTransposes": 0.13763642311096191,
"LICM": 0.11658525466918945,
"LateLegalizeInst": 0.5649778842926025,
"LateLegalizePostSplit": 0.10899925231933594,
"LateLowerReshapeOp": 0.03313422203063965,
"LateLowerTensorOp": 6.405031204223633,
"LateNeuronInstComb": 1.0954866409301758,
"LayoutPreprocessing": 1.1262352466583252,
"LayoutPreprocessingAndAnalysis": 1.4945013523101807,
"LayoutRequirementAnalysis": 0.36240172386169434,
"LegalizeCCOpLayout": 1.2990221977233887,
"LegalizeOpLevelAlias": 1.2160041332244873,
"LegalizePartitionReduce": 0.0940859317779541,
"LegalizeSundaAccess": 3.4260854721069336,
"LegalizeSundaMacro": 0.6391239166259766,
"LegalizeType": 0.1378321647644043,
"LocalLayoutOpt": 0.6706066131591797,
"LoopFusion": 0.28072071075439453,
"LoopSplitting": 0.07517576217651367,
"LowerBroadcast": 0.06826472282409668,
"LowerCCOpBlockAxis": 0.20669817924499512,
"LowerComplexBroadcast": 0.1663343906402588,
"LowerIntrinsics": 0.9762301445007324,
"LowerShardAxis": 0.24805045127868652,
"LowerTensorOp": 3.8965094089508057,
"LowerToSendRecv": 0.18735170364379883,
"LowerTranspose": 0.4830780029296875,
"MacroGeneration": 2.835282802581787,
"MaskPropagation": 0.1340806484222412,
"MemcastMotion": 0.00024199999461416155,
"MemcpyElimination": 28.533254623413086,
"MutateDataType": 0.05120110511779785,
"NeuronAliasDependencyInduction": 0.05350494384765625,
"NeuronAliasDependencyReset": 0.07738471031188965,
"NeuronInstComb": 0.4167485237121582,
"NeuronLICM": 0.2737100124359131,
"NeuronLoopFusion": 1.1017746925354004,
"NeuronLoopInterchange": 0.05889534950256348,
"NeuronSimplifier": 0.5097112655639648,
"NeuronSimplifyPredicates": 0.2256777286529541,
"NeuronValueNumbering": 0.1252288818359375,
"OptimizeAliasedCopyChain": 0.31018686294555664,
"OptimizeNKIKernels": 1.1799430847167969,
"PAGLayoutOpt": 17.110271453857422,
"PComputeCutting": 1.0609712600708008,
"PGLayoutTilingPipeline": 50.19084930419922,
"PGTiling": 11.0619478225708,
"PadElimination": 0.012235879898071289,
"ParAxesAnnotation": 16.403249740600586,
"PartialLoopFusion": 1.436572551727295,
"PartialSimdFusion": 0.8934595584869385,
"PenguinizeFunctions": 0.0003769999893847853,
"PerfectLoopNest": 0.07143855094909668,
"PruneFunctions": 0.00010900000052060932,
"RecognizeOpIdiom": 0.13362383842468262,
"Recompute": 0.0345001220703125,
"RelaxPredicates": 0.11409568786621094,
"Rematerialization": 0.22487568855285645,
"RemoveOptimizationBarriers": 0.00021300000662449747,
"RemoveShardedPartitionAxes": 0.8885042667388916,
"ReshapeWeights": 0.02379584312438965,
"ResolveAccessConflict": 0.46589231491088867,
"ResolveComplicatePredicates": 0.5794906616210938,
"RewriteReplicationMatmul": 0.040407657623291016,
"RewriteWeights": 0.08104610443115234,
"SFKVectorizer": 11.748018264770508,
"ScatterMotion": 0.004219000227749348,
"ShardingPropagationAnalysis": 0.7053203582763672,
"SimpleAllReduceTiling": 0.22031855583190918,
"Simplifier": 0.12051248550415039,
"SimplifyMacroPredicates": 0.27616286277770996,
"SimplifyNeuronTensor": 0.460674524307251,
"SimplifySlice": 0.026907682418823242,
"SimplifyTensor": 0.25261998176574707,
"SpillPSum": 0.7110695838928223,
"SplitAPUnionSets": 0.49262213706970215,
"SplitAccGrp": 0.04794502258300781,
"StaticProfiler": 0.13262128829956055,
"StaticTransposeLocalTensor": 0.6367781162261963,
"SundaISel": 1.6440446376800537,
"TCTransform": 0.030546188354492188,
"TensorInitialization": 0.24484515190124512,
"TensorOpSimplifier": 3.413067102432251,
"TensorOpTransform": 18.72884178161621,
"TensorizerLegalizationPass": 0.0004299999854993075,
"TileCCOps": 0.18365025520324707,
"TilingProfiler": 0.8507044315338135,
"TransformConvOp": 1.1984179019927979,
"TritiumFusion": 1.2713313102722168,
"ValueNumbering": 0.08294677734375,
"VectorizeDMA": 2.601423501968384,
"VectorizeMatMult": 0.10021328926086426,
"VerifySupportedOps": 0.0005520000122487545,
"WeightCoalescing": 0.06738448143005371,
"ZeroSizeTensorElimination": 0.00041866302490234375,
"algsimp": 0.0031190000008791685,
"batchnorm_expander": 0.0024679999332875013,
"boundary-marker-removal": 0.0005530000198632479,
"call-inliner": 0.00031400000443682075,
"canonicalize-boundary-marker": 0.000981000019237399,
"collective-stream-id-checker": 0.0006099999882280827,
"comparison-expander": 0.00041000000783242285,
"computation-deduplicator": 0.0009740000241436064,
"config-lowering": 0.00040600000647827983,
"constant_folding": 0.00021100000594742596,
"cse": 0.0007169999880716205,
"dce": 6.500000017695129e-05,
"dynamic-slice-transpose": 0.0003060000017285347,
"eliminate-redundant-compare": 0.00017600000137463212,
"emit-offloaded-dropout": 0.00033400001120753586,
"flatten-call-graph": 0.0003760000108741224,
"fuse-send-recv": 0.011651000007987022,
"hilo-conditional-to-select": 0.00012700000661425292,
"hilo::LegalizeAlias": 0.0029070000164210796,
"hilo::NeuronInstCombine": 0.0015899999998509884,
"hilo::NeuronOpFusion": 0.000295000005280599,
"hilo::ReplaceTokenTypeWithU8Pass": 0.00023499999952036887,
"hilo::ScheduleFusion": 4.600000102072954e-05,
"hilo::SixtyFourHack": 0.0004560000088531524,
"hilo::VerifyAliasing": 0.00018600000475998968,
"hlo-mac-count": 0.02524399943649769,
"io-con-pipe-begin": 0.00023999999393709004,
"io-con-pipe-end": 9.999999974752427e-07,
"io-layout-normalization": 0.08203800022602081,
"legalize-ccops-for-tensorizer": 0.00014600000577047467,
"legalize-compare": 0.0009549999958835542,
"lower-argminmax-custom-call": 0.0003000000142492354,
"map-inline": 0.0008740000193938613,
"metadata-naming": 0.06593800336122513,
"mlir::detail::OpToOpPassAdaptor": 0.00038400001358240843,
"mlir::hlo::MhloToPyPenguin": 0.5922709703445435,
"mlir::mhlo::LowerComplexExtraPass": 0.005900000222027302,
"mlir::mhlo::LowerComplexPass": 0.004220999777317047,
"native-to-custom-softmax": 0.0007440000190399587,
"native-to-custom-softmax-dx": 0.0006929999799467623,
"neuron-hlo-verifier": 0.3141849935054779,
"operand_upcaster": 0.0021929999347776175,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.3108839988708496,
"pre-hlo-begin": 1.9999999494757503e-05,
"pre-hlo-end": 9.999999974752427e-07,
"replace-minimum-constant": 0.0002690000110305846,
"reshape-mover": 0.00011000000085914508,
"simplify-concat": 0.06884700059890747,
"simplify-while-loops": 8.499999967170879e-05,
"transform-variadic-reduce": 0.0008679999737069011,
"tuple-simplifier": 0.00019299999985378236,
"unpack-nested-aws-ntwsr": 0.00036700000055134296,
"unroll-while-loop": 1.1000000085914508e-05
},
"hilo": {
"HloMacCount": 1972166656.0,
"Traffic": 1252647040.0
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 63618,
"StaticProfiler::AifUb": 27.032575607299805,
"StaticProfiler::ArithmeticIntensityTensorizer": 35.5345458984375,
"StaticProfiler::AverageDmaLength": 2744.118896484375,
"StaticProfiler::DDRTransferBytes": 1130475120,
"StaticProfiler::InternalTransferBytes": 288112224,
"StaticProfiler::LoadExpanded": 338844,
"StaticProfiler::StoreExpanded": 4273,
"StaticProfiler::TotalDMAExpanded": 343117,
"StaticProfiler::TotalDynamicInstancesCount": 81003,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 66628,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 123,
"TilingProfiler::MatMultInstructionsAfterTiling": 40384,
"TilingProfiler::NumPfTransposes": 292,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 142,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 14486,
"TilingProfiler::PfTransposeInstructionsForIo": 11952,
"TilingProfiler::PfTransposeInstructionsForLocal": 872,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 1662,
"TilingProfiler::ReduceInstructionsAfterTiling": 563,
"TilingProfiler::SimdInstructionsAfterTiling": 3485,
"TilingProfiler::TotalInstructionsAfterTiling": 0,
"TransformConvOp::Conv1d_depthwise_bf01_oi01_bf01": 0,
"TransformConvOp::Conv2d_dw_fb01_io01_01bf_rep_nhwc_Pcinh": 0,
"TransformConvOp::Conv2d_pbp_0f1b_0i1o_01fb_experimental_1": 0,
"TransformConvOp::Conv2d_pbp_fb01_io01_01bf_experimental_1": 0,
"TransformConvOp::conv2d_column_packing": 0,
"TransformConvOp::conv2d_column_packing_1": 0,
"TransformConvOp::conv2d_column_packing_io10": 0,
"TransformConvOp::conv2d_depthwise_f01b_o01i_bf01": 0
}
},
"all": {
"compiletime": {
"CanonicalizeConv": 0.0,
"CanonicalizeForTensorizer": 0.0005619999719783664,
"Canonicalizer": 0.011158999986946583,
"HoistCompute": 9.699999645818025e-05,
"IdentifyCrossPassTensors": 0.00019700000120792538,
"MemcastMotion": 0.00024199999461416155,
"PenguinizeFunctions": 0.0003769999893847853,
"PruneFunctions": 0.00010900000052060932,
"RemoveOptimizationBarriers": 0.00021300000662449747,
"ScatterMotion": 0.004219000227749348,
"TensorizerLegalizationPass": 0.0004299999854993075,
"VerifySupportedOps": 0.0005520000122487545,
"algsimp": 0.0031190000008791685,
"batchnorm_expander": 0.0024679999332875013,
"boundary-marker-removal": 0.0005530000198632479,
"call-inliner": 0.00031400000443682075,
"canonicalize-boundary-marker": 0.000981000019237399,
"collective-stream-id-checker": 0.0006099999882280827,
"comparison-expander": 0.00041000000783242285,
"computation-deduplicator": 0.0009740000241436064,
"config-lowering": 0.00040600000647827983,
"constant_folding": 0.00021100000594742596,
"cse": 0.0007169999880716205,
"dce": 6.500000017695129e-05,
"dynamic-slice-transpose": 0.0003060000017285347,
"eliminate-redundant-compare": 0.00017600000137463212,
"emit-offloaded-dropout": 0.00033400001120753586,
"flatten-call-graph": 0.0003760000108741224,
"fuse-send-recv": 0.011651000007987022,
"hilo-conditional-to-select": 0.00012700000661425292,
"hilo::LegalizeAlias": 0.0029070000164210796,
"hilo::NeuronInstCombine": 0.0015899999998509884,
"hilo::NeuronOpFusion": 0.000295000005280599,
"hilo::ReplaceTokenTypeWithU8Pass": 0.00023499999952036887,
"hilo::ScheduleFusion": 4.600000102072954e-05,
"hilo::SixtyFourHack": 0.0004560000088531524,
"hilo::VerifyAliasing": 0.00018600000475998968,
"hlo-mac-count": 0.02524399943649769,
"io-con-pipe-begin": 0.00023999999393709004,
"io-con-pipe-end": 9.999999974752427e-07,
"io-layout-normalization": 0.08203800022602081,
"legalize-ccops-for-tensorizer": 0.00014600000577047467,
"legalize-compare": 0.0009549999958835542,
"lower-argminmax-custom-call": 0.0003000000142492354,
"map-inline": 0.0008740000193938613,
"metadata-naming": 0.06593800336122513,
"mlir::detail::OpToOpPassAdaptor": 0.00038400001358240843,
"mlir::hlo::MhloToPyPenguin": 0.5922709703445435,
"mlir::mhlo::LowerComplexExtraPass": 0.005900000222027302,
"mlir::mhlo::LowerComplexPass": 0.004220999777317047,
"native-to-custom-softmax": 0.0007440000190399587,
"native-to-custom-softmax-dx": 0.0006929999799467623,
"neuron-hlo-verifier": 0.3141849935054779,
"operand_upcaster": 0.0021929999347776175,
"post-par-pipe-begin": 9.999999974752427e-07,
"post-par-pipe-end": 0.0,
"post-partition-simplification": 0.3108839988708496,
"pre-hlo-begin": 1.9999999494757503e-05,
"pre-hlo-end": 9.999999974752427e-07,
"replace-minimum-constant": 0.0002690000110305846,
"reshape-mover": 0.00011000000085914508,
"simplify-concat": 0.06884700059890747,
"simplify-while-loops": 8.499999967170879e-05,
"transform-variadic-reduce": 0.0008679999737069011,
"tuple-simplifier": 0.00019299999985378236,
"unpack-nested-aws-ntwsr": 0.00036700000055134296,
"unroll-while-loop": 1.1000000085914508e-05
}
},
"cumsum": {
"compiletime": {
"CoalesceCCOp": 0.0003490447998046875,
"DMALocalityOpt": 0.00033211708068847656,
"DMAProfiler": 0.0012180805206298828,
"DataStreaming": 0.0005357265472412109,
"DoNothing": 0.00019812583923339844,
"ExpandISAMacro": 0.0012791156768798828,
"FactorizeBlkDims": 0.000789642333984375,
"InferPSumTensor": 0.001064300537109375,
"InferSharedMemLoc": 0.00042891502380371094,
"InsertCoreBarrier": 0.00044727325439453125,
"LateLegalizeInst": 0.0007555484771728516,
"LateNeuronInstComb": 0.0008594989776611328,
"LegalizeSundaAccess": 0.0021588802337646484,
"LegalizeType": 0.0003390312194824219,
"LowerBroadcast": 0.0003108978271484375,
"LowerIntrinsics": 0.0003314018249511719,
"LowerTranspose": 0.0003635883331298828,
"NeuronInstComb": 0.0009188652038574219,
"NeuronLICM": 0.0006546974182128906,
"NeuronSimplifyPredicates": 0.004039287567138672,
"NeuronValueNumbering": 0.0005695819854736328,
"SFKVectorizer": 0.006131172180175781,
"SimpleAllReduceTiling": 0.0004286766052246094,
"SimplifyNeuronTensor": 0.0011782646179199219,
"SpillPSum": 0.0007271766662597656,
"WeightCoalescing": 0.00041985511779785156
}
},
"sg00": {
"hilo": {
"ArithmeticIntensity": 3.148798704147339,
"HloMacCount": 1972166656.0,
"Traffic": 1252647040.0
}
},
"sg0000": {
"compiletime": {
"AGOrderingAnalysisPass": 5.95287561416626,
"AffinePredicateResolution": 0.49277400970458984,
"AliasDependencyElimination": 0.01732039451599121,
"AliasDependencyInduction": 18.183685302734375,
"AliasDependencyReset": 18.999177932739258,
"BFComputeCutting": 0.3126344680786133,
"BirCodeGenLoop": 1.8802330493927002,
"CCOpFusion": 0.8809163570404053,
"CanonicalizeDAGForPGTiling": 0.17032384872436523,
"CanonicalizeIR": 0.7969982624053955,
"CoalesceCCOp": 0.6308276653289795,
"CommuteConcat": 0.02859354019165039,
"DMALocalityOpt": 0.036058664321899414,
"DMAProfiler": 0.1385490894317627,
"DMATilingProfiler": 0.08906126022338867,
"DataLocalityOpt": 4.2657470703125,
"DataStreaming": 0.1718294620513916,
"DeConcat": 0.06563258171081543,
"DeadCodeElimination": 0.0297544002532959,
"DeadStoreElimination": 1.2178454399108887,
"DelinearIndices": 0.35585713386535645,
"Delinearization": 0.6069140434265137,
"DelinearizeSPMD": 0.6448159217834473,
"DoNothing": 0.00018167495727539063,
"DramToDramTranspose": 0.3358180522918701,
"DumpGraphAndMetadata": 0.17146539688110352,
"EliminateDivs": 1.3107364177703857,
"ExpandBatchNorm": 0.48049426078796387,
"ExpandISAMacro": 0.07998967170715332,
"FactorizeBlkDims": 0.5875980854034424,
"FactorizeThreadAxesInFreeDims": 0.09288716316223145,
"FlattenMacroLoop": 0.0829305648803711,
"GenericAccessSimplifier": 0.026407241821289063,
"InferInitValue": 1.2918815612792969,
"InferIntrinsicOnCC": 0.31558942794799805,
"InferNeuronTensor": 1.773860216140747,
"InferNonlocalTensors": 6.451204299926758,
"InferPSumTensor": 1.4296114444732666,
"InferShardAxis": 10.398513793945313,
"InferSharedMemLoc": 0.12653231620788574,
"InlineNativeKernels": 0.051079511642456055,
"InsertCoreBarrier": 0.36295533180236816,
"InsertIOTransposes": 1.088529348373413,
"InsertImplicitShardAxisBeforeISel": 0.4208717346191406,
"InsertLocalTransposes": 0.6978082656860352,
"InsertOffloadedTransposes": 0.13763642311096191,
"LICM": 0.11658525466918945,
"LateLegalizeInst": 0.5564777851104736,
"LateLegalizePostSplit": 0.10899925231933594,
"LateLowerReshapeOp": 0.03313422203063965,
"LateLowerTensorOp": 6.405031204223633,
"LateNeuronInstComb": 1.0862395763397217,
"LayoutPreprocessing": 1.1262352466583252,
"LayoutPreprocessingAndAnalysis": 1.4945013523101807,
"LayoutRequirementAnalysis": 0.36240172386169434,
"LegalizeCCOpLayout": 1.2990221977233887,
"LegalizeOpLevelAlias": 1.2160041332244873,
"LegalizePartitionReduce": 0.0940859317779541,
"LegalizeSundaAccess": 3.4090025424957275,
"LegalizeSundaMacro": 0.6391239166259766,
"LegalizeType": 0.12801194190979004,
"LocalLayoutOpt": 0.6706066131591797,
"LoopFusion": 0.28072071075439453,
"LoopSplitting": 0.07517576217651367,
"LowerBroadcast": 0.06457281112670898,
"LowerCCOpBlockAxis": 0.20669817924499512,
"LowerComplexBroadcast": 0.1663343906402588,
"LowerIntrinsics": 0.9722909927368164,
"LowerShardAxis": 0.24805045127868652,
"LowerTensorOp": 3.8965094089508057,
"LowerToSendRecv": 0.18735170364379883,
"LowerTranspose": 0.47939133644104004,
"MacroGeneration": 2.835282802581787,
"MaskPropagation": 0.1340806484222412,
"MemcpyElimination": 28.533254623413086,
"MutateDataType": 0.05120110511779785,
"NeuronAliasDependencyInduction": 0.05350494384765625,
"NeuronAliasDependencyReset": 0.07738471031188965,
"NeuronInstComb": 0.4073524475097656,
"NeuronLICM": 0.2638108730316162,
"NeuronLoopFusion": 1.1017746925354004,
"NeuronLoopInterchange": 0.05889534950256348,
"NeuronSimplifier": 0.5097112655639648,
"NeuronSimplifyPredicates": 0.21783137321472168,
"NeuronValueNumbering": 0.12063908576965332,
"OptimizeAliasedCopyChain": 0.31018686294555664,
"OptimizeNKIKernels": 1.1799430847167969,
"PAGLayoutOpt": 17.110271453857422,
"PComputeCutting": 1.0609712600708008,
"PGLayoutTilingPipeline": 50.19084930419922,
"PGTiling": 11.0619478225708,
"PadElimination": 0.012235879898071289,
"ParAxesAnnotation": 16.403249740600586,
"PartialLoopFusion": 1.436572551727295,
"PartialSimdFusion": 0.8934595584869385,
"PerfectLoopNest": 0.07143855094909668,
"RecognizeOpIdiom": 0.13362383842468262,
"Recompute": 0.0345001220703125,
"RelaxPredicates": 0.11409568786621094,
"Rematerialization": 0.22487568855285645,
"RemoveShardedPartitionAxes": 0.8885042667388916,
"ReshapeWeights": 0.02379584312438965,
"ResolveAccessConflict": 0.46589231491088867,
"ResolveComplicatePredicates": 0.5794906616210938,
"RewriteReplicationMatmul": 0.040407657623291016,
"RewriteWeights": 0.08104610443115234,
"SFKVectorizer": 11.707208633422852,
"ShardingPropagationAnalysis": 0.7053203582763672,
"SimpleAllReduceTiling": 0.2163074016571045,
"Simplifier": 0.12051248550415039,
"SimplifyMacroPredicates": 0.27616286277770996,
"SimplifyNeuronTensor": 0.4094421863555908,
"SimplifySlice": 0.026907682418823242,
"SimplifyTensor": 0.25261998176574707,
"SpillPSum": 0.6899855136871338,
"SplitAPUnionSets": 0.49262213706970215,
"SplitAccGrp": 0.04794502258300781,
"StaticProfiler": 0.13262128829956055,
"StaticTransposeLocalTensor": 0.6367781162261963,
"SundaISel": 1.6440446376800537,
"TCTransform": 0.030546188354492188,
"TensorInitialization": 0.24484515190124512,
"TensorOpSimplifier": 3.413067102432251,
"TensorOpTransform": 18.72884178161621,
"TileCCOps": 0.18365025520324707,
"TilingProfiler": 0.8507044315338135,
"TransformConvOp": 1.1984179019927979,
"TritiumFusion": 1.2713313102722168,
"ValueNumbering": 0.08294677734375,
"VectorizeDMA": 2.601423501968384,
"VectorizeMatMult": 0.10021328926086426,
"WeightCoalescing": 0.06349992752075195,
"ZeroSizeTensorElimination": 0.00041866302490234375
},
"tensorizer": {
"DMATilingProfiler::TotalInstructionsAfterTiling": 63618,
"StaticProfiler::AifUb": 27.032575607299805,
"StaticProfiler::ArithmeticIntensityTensorizer": 35.5345458984375,
"StaticProfiler::AverageDmaLength": 2744.118896484375,
"StaticProfiler::AverageFractalPeUtilization": 98.06681823730469,
"StaticProfiler::AveragePartitionUtilization": 88.63555145263672,
"StaticProfiler::AveragePeUtilization": 75.70149230957031,
"StaticProfiler::DDRTransferBytes": 1130475120,
"StaticProfiler::InternalTransferBytes": 288112224,
"StaticProfiler::LoadExpanded": 338844,
"StaticProfiler::LocalizationEfficiency": 131.45083618164063,
"StaticProfiler::LocalizationEfficiencyIgnoreNonlocal": 134.9540252685547,
"StaticProfiler::StoreExpanded": 4273,
"StaticProfiler::TotalDMAExpanded": 343117,
"StaticProfiler::TotalDynamicInstancesCount": 81003,
"StaticProfiler::TotalDynamicInstancesWithMmPackedCount": 66628,
"StaticProfiler::TotalLNCComm": 0,
"StaticProfiler::TotalLNCCommTransfer": 0,
"TilingProfiler::AveragePartitionUtilizationAfterTiling": 0,
"TilingProfiler::AveragePeUtilizationAfterTiling": 0,
"TilingProfiler::BatchnormInstructionsAfterTiling": 0,
"TilingProfiler::DmaInstructionsAfterTiling": 0,
"TilingProfiler::GenericInstructionsAfterTiling": 123,
"TilingProfiler::MatMultInstructionsAfterTiling": 40384,
"TilingProfiler::NumPfTransposes": 292,
"TilingProfiler::NumPfTransposesForIo": 30,
"TilingProfiler::NumPfTransposesForLocal": 142,
"TilingProfiler::NumPfTransposesForNonlocal": 120,
"TilingProfiler::PfTransposeInstructions": 14486,
"TilingProfiler::PfTransposeInstructionsForIo": 11952,
"TilingProfiler::PfTransposeInstructionsForLocal": 872,
"TilingProfiler::PfTransposeInstructionsForNonlocal": 1662,
"TilingProfiler::ReduceInstructionsAfterTiling": 563,
"TilingProfiler::SimdInstructionsAfterTiling": 3485,
"TilingProfiler::TotalInstructionsAfterTiling": 0,
"TransformConvOp::Conv1d_depthwise_bf01_oi01_bf01": 0,
"TransformConvOp::Conv2d_dw_fb01_io01_01bf_rep_nhwc_Pcinh": 0,
"TransformConvOp::Conv2d_pbp_0f1b_0i1o_01fb_experimental_1": 0,
"TransformConvOp::Conv2d_pbp_fb01_io01_01bf_experimental_1": 0,
"TransformConvOp::conv2d_column_packing": 0,
"TransformConvOp::conv2d_column_packing_1": 0,
"TransformConvOp::conv2d_column_packing_io10": 0,
"TransformConvOp::conv2d_depthwise_f01b_o01i_bf01": 0
}
},
"topk": {
"compiletime": {
"CoalesceCCOp": 0.0037636756896972656,
"DMALocalityOpt": 0.002650737762451172,
"DMAProfiler": 0.003582477569580078,
"DataStreaming": 0.006698131561279297,
"DoNothing": 0.00016760826110839844,
"ExpandISAMacro": 0.0037276744842529297,
"FactorizeBlkDims": 0.012275934219360352,
"InferPSumTensor": 0.01163172721862793,
"InferSharedMemLoc": 0.002855539321899414,
"InsertCoreBarrier": 0.0032694339752197266,
"LateLegalizeInst": 0.007744550704956055,
"LateNeuronInstComb": 0.008387565612792969,
"LegalizeSundaAccess": 0.014924049377441406,
"LegalizeType": 0.009481191635131836,
"LowerBroadcast": 0.003381013870239258,
"LowerIntrinsics": 0.0036077499389648438,
"LowerTranspose": 0.003323078155517578,
"NeuronInstComb": 0.008477210998535156,
"NeuronLICM": 0.009244441986083984,
"NeuronSimplifyPredicates": 0.00380706787109375,
"NeuronValueNumbering": 0.004020214080810547,
"SFKVectorizer": 0.03467822074890137,
"SimpleAllReduceTiling": 0.003582477569580078,
"SimplifyNeuronTensor": 0.050054073333740234,
"SpillPSum": 0.02035689353942871,
"WeightCoalescing": 0.0034646987915039063
}
}
}

3
token_generation_model/_tp0_bk4/graph.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2840765467a4771e8bfbd46ac14d381ab5b0ca2a5728c2c4e45bcaecd45c75d2
size 4097024

4484
token_generation_model/_tp0_bk4/log-neuron-cc.txt Normal file
View File

File diff suppressed because it is too large Load Diff

3
token_generation_model/_tp0_bk4/metaneff.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:d29e9e51be8d730fedcaafdffddee32ee55e95335b0070a546ddb2f8bc39ed32
size 2464555

3
token_generation_model/_tp0_bk4/model.MODULE_b4b44a47076a167bd9a5+795ee7cc.hlo_module.pb Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:0e834b15d13ccd9cce8994a607f90a7511bc74cf2044170c619756122300d806
size 2550451

3
token_generation_model/_tp0_bk4/model.MODULE_b4b44a47076a167bd9a5+795ee7cc.neff Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:2840765467a4771e8bfbd46ac14d381ab5b0ca2a5728c2c4e45bcaecd45c75d2
size 4097024

224
token_generation_model/_tp0_bk4/neuron_config.json Normal file
View File

@@ -0,0 +1,224 @@
{
"_attn_implementation_autoset": false,
"_name_or_path": "/models/Qwen3-1.7B/",
"add_cross_attention": false,
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"attribute_map": {},
"bad_words_ids": null,
"begin_suppress_tokens": null,
"bos_token_id": 151643,
"chunk_size_feed_forward": 0,
"cross_attention_hidden_size": null,
"decoder_start_token_id": null,
"diversity_penalty": 0.0,
"do_sample": false,
"early_stopping": false,
"encoder_no_repeat_ngram_size": 0,
"eos_token_id": 151645,
"exponential_decay_length_penalty": null,
"finetuning_task": null,
"forced_bos_token_id": null,
"forced_eos_token_id": null,
"fused_spec_config": null,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2048,
"id2label": {
"0": "LABEL_0",
"1": "LABEL_1"
},
"initializer_range": 0.02,
"intermediate_size": 6144,
"is_decoder": false,
"is_encoder_decoder": false,
"label2id": {
"LABEL_0": 0,
"LABEL_1": 1
},
"length_penalty": 1.0,
"max_length": 20,
"max_position_embeddings": 40960,
"max_window_layers": 28,
"metadata": null,
"min_length": 0,
"model_type": "qwen3",
"neuron_config": {
"activation_quantization_type": null,
"allow_input_truncation": false,
"apply_seq_ids_mask": false,
"async_mode": false,
"attention_dp_degree": 1,
"attention_dtype": null,
"attn_block_cte_nki_kernel_enabled": false,
"attn_block_tkg_nki_kernel_cache_update": false,
"attn_block_tkg_nki_kernel_cascaded_attention": false,
"attn_block_tkg_nki_kernel_enabled": false,
"attn_cls": {
"__module__": "neuronx_distributed_inference.models.qwen3.modeling_qwen3",
"__name__": "NeuronQwen3Attention"
},
"attn_kernel_enabled": null,
"attn_tkg_builtin_kernel_enabled": false,
"attn_tkg_nki_kernel_enabled": false,
"batch_size": 4,
"bucket_n_active_tokens": false,
"buckets": [
2048
],
"cast_type": "config",
"cc_pipeline_tiling_factor": 1,
"chunked_prefill_config": null,
"context_encoding_buckets": null,
"cp_degree": 1,
"ctx_batch_size": 1,
"disable_kv_cache_tiling": false,
"draft_model_modules_to_not_convert": null,
"enable_bucketing": true,
"enable_cte_modular_flow": false,
"enable_eagle_draft_input_norm": false,
"enable_eagle_speculation": false,
"enable_fused_speculation": false,
"enable_long_context_mode": false,
"enable_output_completion_notifications": false,
"enable_spill_reload_dge": false,
"enable_token_tree": false,
"ep_degree": 1,
"expert_mlp_nki_kernel_enabled": null,
"flash_decoding_enabled": false,
"fused_qkv": false,
"fused_rmsnorm_skip_gamma": false,
"is_block_kv_layout": null,
"is_chunked_prefill": false,
"is_continuous_batching": true,
"is_eagle_draft": false,
"is_medusa": false,
"is_prefill_stage": false,
"is_prefix_caching": false,
"k_cache_transposed": false,
"kv_cache_batch_size": 4,
"kv_cache_padding_size": 0,
"kv_cache_quant": false,
"kv_cache_tiling": false,
"layer_boundary_markers": false,
"lm_head_pad": true,
"lm_head_pad_alignment_size": 1,
"local_ranks_size": 4,
"logical_nc_config": 2,
"lora_config": null,
"max_batch_size": 4,
"max_context_length": 2048,
"max_length": 2048,
"max_new_tokens": null,
"medusa_speculation_length": 0,
"medusa_tree": null,
"mlp_kernel_enabled": false,
"mlp_kernel_fuse_residual_add": false,
"modules_to_not_convert": null,
"moe_fused_nki_kernel_enabled": null,
"n_active_tokens": 1,
"n_positions": 2048,
"num_medusa_heads": 0,
"on_cpu": false,
"on_device_sampling_config": {
"deterministic": false,
"do_sample": false,
"dynamic": true,
"global_topk": 256,
"on_device_sampling_config": true,
"temperature": 1.0,
"top_k": 1,
"top_k_kernel_enabled": false,
"top_p": 1.0
},
"output_logits": false,
"overrides_torch_dtype": true,
"pa_block_size": 2048,
"pa_num_blocks": 4,
"padding_side": "right",
"pp_degree": 1,
"prefix_buckets": null,
"qk_layernorm": false,
"qkv_kernel_enabled": false,
"qkv_kernel_fuse_residual_add": false,
"qkv_kernel_nbsd_layout": false,
"quantization_dtype": "int8",
"quantization_type": "per_tensor_symmetric",
"quantize_clamp_bound": Infinity,
"quantized": false,
"quantized_checkpoints_path": null,
"quantized_mlp_kernel_enabled": false,
"rmsnorm_quantize_kernel_enabled": false,
"router_topk_nki_kernel_enabled": null,
"rpl_reduce_dtype": null,
"save_sharded_checkpoint": true,
"scratchpad_page_size": null,
"seq_len": 2048,
"seq_len_threshold_for_cc_tiling": 16384,
"sequence_parallel_enabled": false,
"shared_mlp_nki_kernel_enabled": null,
"skip_sharding": false,
"skip_warmup": false,
"spec_batch_size": 4,
"speculation_length": 0,
"start_rank_id": 0,
"strided_context_parallel_kernel_enabled": false,
"target": null,
"tensor_capture_config": null,
"tile_cc": false,
"tkg_batch_size": 4,
"token_generation_buckets": [
2048
],
"token_tree_config": null,
"torch_dtype": "bfloat16",
"tp_degree": 4,
"vocab_parallel": false,
"weight_gather_seq_len_threshold": 32768,
"weights_to_skip_layout_optimization": [],
"world_size": 4
},
"no_repeat_ngram_size": 0,
"num_attention_heads": 16,
"num_beam_groups": 1,
"num_beams": 1,
"num_cores_per_group": 1,
"num_hidden_layers": 28,
"num_key_value_heads": 8,
"num_return_sequences": 1,
"output_attentions": false,
"output_hidden_states": false,
"output_scores": false,
"pad_token_id": 0,
"prefix": null,
"problem_type": null,
"pruned_heads": {},
"remove_invalid_values": false,
"repetition_penalty": 1.0,
"return_dict": true,
"return_dict_in_generate": false,
"rms_norm_eps": 1e-06,
"rope_scaling": null,
"rope_theta": 1000000,
"sep_token_id": null,
"sliding_window": null,
"suppress_tokens": null,
"task_specific_params": null,
"temperature": 1.0,
"tf_legacy_loss": false,
"tie_encoder_decoder": false,
"tie_word_embeddings": true,
"tokenizer_class": null,
"top_k": 50,
"top_p": 1.0,
"torchscript": false,
"transformers_version": "4.51.0",
"typical_p": 1.0,
"use_bfloat16": false,
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 151936
}