diff --git "a/yolo_nas_pose_m_int8.onnx.best.engine.log" "b/yolo_nas_pose_m_int8.onnx.best.engine.log" --- "a/yolo_nas_pose_m_int8.onnx.best.engine.log" +++ "b/yolo_nas_pose_m_int8.onnx.best.engine.log" @@ -1,333 +1,338 @@ &&&& RUNNING TensorRT.trtexec [TensorRT v8502] # /usr/src/tensorrt/bin/trtexec --onnx=yolo_nas_pose_m_int8.onnx --best --avgRuns=100 --duration=15 --saveEngine=yolo_nas_pose_m_int8.onnx.best.engine -[12/28/2023-10:38:43] [I] === Model Options === -[12/28/2023-10:38:43] [I] Format: ONNX -[12/28/2023-10:38:43] [I] Model: yolo_nas_pose_m_int8.onnx -[12/28/2023-10:38:43] [I] Output: -[12/28/2023-10:38:43] [I] === Build Options === -[12/28/2023-10:38:43] [I] Max batch: explicit batch -[12/28/2023-10:38:43] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default -[12/28/2023-10:38:43] [I] minTiming: 1 -[12/28/2023-10:38:43] [I] avgTiming: 8 -[12/28/2023-10:38:43] [I] Precision: FP32+FP16+INT8 -[12/28/2023-10:38:43] [I] LayerPrecisions: -[12/28/2023-10:38:43] [I] Calibration: Dynamic -[12/28/2023-10:38:43] [I] Refit: Disabled -[12/28/2023-10:38:43] [I] Sparsity: Disabled -[12/28/2023-10:38:43] [I] Safe mode: Disabled -[12/28/2023-10:38:43] [I] DirectIO mode: Disabled -[12/28/2023-10:38:43] [I] Restricted mode: Disabled -[12/28/2023-10:38:43] [I] Build only: Disabled -[12/28/2023-10:38:43] [I] Save engine: yolo_nas_pose_m_int8.onnx.best.engine -[12/28/2023-10:38:43] [I] Load engine: -[12/28/2023-10:38:43] [I] Profiling verbosity: 0 -[12/28/2023-10:38:43] [I] Tactic sources: Using default tactic sources -[12/28/2023-10:38:43] [I] timingCacheMode: local -[12/28/2023-10:38:43] [I] timingCacheFile: -[12/28/2023-10:38:43] [I] Heuristic: Disabled -[12/28/2023-10:38:43] [I] Preview Features: Use default preview flags. -[12/28/2023-10:38:43] [I] Input(s)s format: fp32:CHW -[12/28/2023-10:38:43] [I] Output(s)s format: fp32:CHW -[12/28/2023-10:38:43] [I] Input build shapes: model -[12/28/2023-10:38:43] [I] Input calibration shapes: model -[12/28/2023-10:38:43] [I] === System Options === -[12/28/2023-10:38:43] [I] Device: 0 -[12/28/2023-10:38:43] [I] DLACore: -[12/28/2023-10:38:43] [I] Plugins: -[12/28/2023-10:38:43] [I] === Inference Options === -[12/28/2023-10:38:43] [I] Batch: Explicit -[12/28/2023-10:38:43] [I] Input inference shapes: model -[12/28/2023-10:38:43] [I] Iterations: 10 -[12/28/2023-10:38:43] [I] Duration: 15s (+ 200ms warm up) -[12/28/2023-10:38:43] [I] Sleep time: 0ms -[12/28/2023-10:38:43] [I] Idle time: 0ms -[12/28/2023-10:38:43] [I] Streams: 1 -[12/28/2023-10:38:43] [I] ExposeDMA: Disabled -[12/28/2023-10:38:43] [I] Data transfers: Enabled -[12/28/2023-10:38:43] [I] Spin-wait: Disabled -[12/28/2023-10:38:43] [I] Multithreading: Disabled -[12/28/2023-10:38:43] [I] CUDA Graph: Disabled -[12/28/2023-10:38:43] [I] Separate profiling: Disabled -[12/28/2023-10:38:43] [I] Time Deserialize: Disabled -[12/28/2023-10:38:43] [I] Time Refit: Disabled -[12/28/2023-10:38:43] [I] NVTX verbosity: 0 -[12/28/2023-10:38:43] [I] Persistent Cache Ratio: 0 -[12/28/2023-10:38:43] [I] Inputs: -[12/28/2023-10:38:43] [I] === Reporting Options === -[12/28/2023-10:38:43] [I] Verbose: Disabled -[12/28/2023-10:38:43] [I] Averages: 100 inferences -[12/28/2023-10:38:43] [I] Percentiles: 90,95,99 -[12/28/2023-10:38:43] [I] Dump refittable layers:Disabled -[12/28/2023-10:38:43] [I] Dump output: Disabled -[12/28/2023-10:38:43] [I] Profile: Disabled -[12/28/2023-10:38:43] [I] Export timing to JSON file: -[12/28/2023-10:38:43] [I] Export output to JSON file: -[12/28/2023-10:38:43] [I] Export profile to JSON file: -[12/28/2023-10:38:43] [I] -[12/28/2023-10:38:43] [I] === Device Information === -[12/28/2023-10:38:43] [I] Selected Device: Orin -[12/28/2023-10:38:43] [I] Compute Capability: 8.7 -[12/28/2023-10:38:43] [I] SMs: 8 -[12/28/2023-10:38:43] [I] Compute Clock Rate: 0.624 GHz -[12/28/2023-10:38:43] [I] Device Global Memory: 7471 MiB -[12/28/2023-10:38:43] [I] Shared Memory per SM: 164 KiB -[12/28/2023-10:38:43] [I] Memory Bus Width: 128 bits (ECC disabled) -[12/28/2023-10:38:43] [I] Memory Clock Rate: 0.624 GHz -[12/28/2023-10:38:43] [I] -[12/28/2023-10:38:43] [I] TensorRT version: 8.5.2 -[12/28/2023-10:38:43] [I] [TRT] [MemUsageChange] Init CUDA: CPU +220, GPU +0, now: CPU 249, GPU 2979 (MiB) -[12/28/2023-10:38:46] [I] [TRT] [MemUsageChange] Init builder kernel library: CPU +302, GPU +284, now: CPU 574, GPU 3284 (MiB) -[12/28/2023-10:38:46] [I] Start parsing network model -[12/28/2023-10:38:46] [I] [TRT] ---------------------------------------------------------------- -[12/28/2023-10:38:46] [I] [TRT] Input filename: yolo_nas_pose_m_int8.onnx -[12/28/2023-10:38:46] [I] [TRT] ONNX IR version: 0.0.8 -[12/28/2023-10:38:46] [I] [TRT] Opset version: 17 -[12/28/2023-10:38:46] [I] [TRT] Producer name: pytorch -[12/28/2023-10:38:46] [I] [TRT] Producer version: 2.1.2 -[12/28/2023-10:38:46] [I] [TRT] Domain: -[12/28/2023-10:38:46] [I] [TRT] Model version: 0 -[12/28/2023-10:38:46] [I] [TRT] Doc string: -[12/28/2023-10:38:46] [I] [TRT] ---------------------------------------------------------------- -[12/28/2023-10:38:50] [I] Finish parsing network model -[12/28/2023-10:38:54] [I] [TRT] ---------- Layers Running on DLA ---------- -[12/28/2023-10:38:54] [I] [TRT] ---------- Layers Running on GPU ---------- -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation1] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/pre_process/pre_process.0/Cast.../pre_process/pre_process.2/Mul]} -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1274) [Constant] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1275) [Constant] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1276) [Constant] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stem/conv/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stem.conv.rbr_reparam.weight + /model/backbone/stem/conv/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stem/conv/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.downsample.rbr_reparam.weight + /model/backbone/stage1/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/downsample/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv2.conv.weight + /model/backbone/stage1/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv1.conv.weight + /model/backbone/stage1/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage1.blocks.bottlenecks.0.alpha + (Unnamed Layer* 509) [Shuffle] + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage1.blocks.bottlenecks.1.alpha + (Unnamed Layer* 525) [Shuffle] + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv3.conv.weight + /model/backbone/stage1/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.reduce_skip2.conv.weight + /model/neck/neck2/reduce_skip2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_skip2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.downsample.rbr_reparam.weight + /model/backbone/stage2/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/downsample/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.downsample.conv.weight + /model/neck/neck2/downsample/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/downsample/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv2.conv.weight + /model/backbone/stage2/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv1.conv.weight + /model/backbone/stage2/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.0.alpha + (Unnamed Layer* 572) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.1.alpha + (Unnamed Layer* 588) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.2.alpha + (Unnamed Layer* 604) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_2 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv3.conv.weight + /model/backbone/stage2/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_skip2.conv.weight + /model/neck/neck1/reduce_skip2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_skip2/conv/Conv || model.neck.neck2.reduce_skip1.conv.weight + /model/neck/neck2/reduce_skip1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_skip1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.downsample.rbr_reparam.weight + /model/backbone/stage3/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/downsample/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.downsample.conv.weight + /model/neck/neck1/downsample/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/downsample/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv2.conv.weight + /model/backbone/stage3/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv1.conv.weight + /model/backbone/stage3/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.0.alpha + (Unnamed Layer* 654) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.1.alpha + (Unnamed Layer* 670) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.2.alpha + (Unnamed Layer* 686) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.3.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.3.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.3.alpha + (Unnamed Layer* 702) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.4.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.4.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.4.alpha + (Unnamed Layer* 718) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_4 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_3 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_2 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv3.conv.weight + /model/backbone/stage3/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_skip1.conv.weight + /model/neck/neck1/reduce_skip1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_skip1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.downsample.rbr_reparam.weight + /model/backbone/stage4/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/downsample/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv2.conv.weight + /model/backbone/stage4/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv1.conv.weight + /model/backbone/stage4/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage4.blocks.bottlenecks.0.alpha + (Unnamed Layer* 759) [Shuffle] + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage4.blocks.bottlenecks.1.alpha + (Unnamed Layer* 775) [Shuffle] + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv3.conv.weight + /model/backbone/stage4/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.context_module.cv1.conv.weight + /model/backbone/context_module/cv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/context_module/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.2/MaxPool -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.1/MaxPool -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.0/MaxPool -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/backbone/context_module/m.2/MaxPool_output_0 copy -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.context_module.cv2.conv.weight + /model/backbone/context_module/cv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/context_module/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.conv.conv.weight + /model/neck/neck1/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/conv/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] DECONVOLUTION: model.neck.neck1.upsample.weight + /model/neck/neck1/upsample/_weight_quantizer/QuantizeLinear + /model/neck/neck1/upsample/ConvTranspose -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_after_concat.conv.weight + /model/neck/neck1/reduce_after_concat/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_after_concat/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv2.conv.weight + /model/neck/neck1/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv1.conv.weight + /model/neck/neck1/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck1.blocks.bottlenecks.0.alpha + (Unnamed Layer* 840) [Shuffle] + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck1.blocks.bottlenecks.1.alpha + (Unnamed Layer* 856) [Shuffle] + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv3.conv.weight + /model/neck/neck1/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.conv.conv.weight + /model/neck/neck2/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/conv/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] DECONVOLUTION: model.neck.neck2.upsample.weight + /model/neck/neck2/upsample/_weight_quantizer/QuantizeLinear + /model/neck/neck2/upsample/ConvTranspose -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/Concat_/model/neck/neck2/reduce_skip1/act/Relu_output_0_clone_1 copy -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.reduce_after_concat.conv.weight + /model/neck/neck2/reduce_after_concat/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_after_concat/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv2.conv.weight + /model/neck/neck2/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv1.conv.weight + /model/neck/neck2/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.0.alpha + (Unnamed Layer* 905) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.1.alpha + (Unnamed Layer* 921) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.2.alpha + (Unnamed Layer* 937) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv3.conv.weight + /model/neck/neck2/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.bbox_stem.seq.conv.weight + /model/heads/head1/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/bbox_stem/seq/conv/Conv || model.heads.head1.pose_stem.seq.conv.weight + /model/heads/head1/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_stem/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.conv.conv.weight + /model/neck/neck3/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/conv/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.reg_convs.0.seq.conv.weight + /model/heads/head1/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head1.cls_convs.0.seq.conv.weight + /model/heads/head1/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/cls_convs/cls_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_convs.0.seq.conv.weight + /model/heads/head1/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_convs/pose_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv2.conv.weight + /model/neck/neck3/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv1.conv.weight + /model/neck/neck3/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.cls_pred.weight + /model/heads/head1/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/cls_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.reg_pred.weight + /model/heads/head1/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/reg_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_convs.1.seq.conv.weight + /model/heads/head1/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_convs/pose_convs.1/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape + /model/heads/Transpose -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_pred.weight + /model/heads/head1/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.0.cv1.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.0.cv2.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck3.blocks.bottlenecks.0.alpha + (Unnamed Layer* 1002) [Shuffle] + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.1.cv1.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.1.cv2.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck3.blocks.bottlenecks.1.alpha + (Unnamed Layer* 1043) [Shuffle] + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv3.conv.weight + /model/neck/neck3/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_stem.seq.conv.weight + /model/heads/head2/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_stem/seq/conv/Conv || model.heads.head2.bbox_stem.seq.conv.weight + /model/heads/head2/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/bbox_stem/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.conv.conv.weight + /model/neck/neck4/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/conv/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.reg_convs.0.seq.conv.weight + /model/heads/head2/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head2.cls_convs.0.seq.conv.weight + /model/heads/head2/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/cls_convs/cls_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_convs.0.seq.conv.weight + /model/heads/head2/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_convs/pose_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv2.conv.weight + /model/neck/neck4/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv1.conv.weight + /model/neck/neck4/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.cls_pred.weight + /model/heads/head2/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/cls_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.reg_pred.weight + /model/heads/head2/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/reg_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_convs.1.seq.conv.weight + /model/heads/head2/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_convs/pose_convs.1/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape_4 + /model/heads/Transpose_3 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_pred.weight + /model/heads/head2/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.0.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax_1 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.0.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv_1 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.0.alpha + (Unnamed Layer* 1108) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.1.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.1.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.1.alpha + (Unnamed Layer* 1149) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/_input_quantizer/QuantizeLinear -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.2.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.2.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv2/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.2.alpha + (Unnamed Layer* 1165) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/Add) -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv3.conv.weight + /model/neck/neck4/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv3/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.bbox_stem.seq.conv.weight + /model/heads/head3/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/bbox_stem/seq/conv/Conv || model.heads.head3.pose_stem.seq.conv.weight + /model/heads/head3/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_stem/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.reg_convs.0.seq.conv.weight + /model/heads/head3/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head3.cls_convs.0.seq.conv.weight + /model/heads/head3/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/cls_convs/cls_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.0.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.0/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.cls_pred.weight + /model/heads/head3/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/cls_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.reg_pred.weight + /model/heads/head3/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/reg_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.1.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.1/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape_8 + /model/heads/Transpose_6 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.2.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.2/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.2/seq/conv/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax_2 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_pred.weight + /model/heads/head3/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_pred/Conv -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv_2 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/model/heads/head1/Slice_1.../post_process/Reshape_2]} -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] NMS: batched_nms_250 -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] DEVICE_TO_SHAPE_HOST: (Unnamed Layer* 1278) [NMS]_1_output[DevicetoShapeHostCopy] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation2] -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/model/heads/head1/Slice...graph2_/Concat_5]} -[12/28/2023-10:38:54] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation3] -[12/28/2023-10:39:08] [I] [TRT] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +534, GPU +490, now: CPU 1413, GPU 4003 (MiB) -[12/28/2023-10:39:10] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +83, GPU +74, now: CPU 1496, GPU 4077 (MiB) -[12/28/2023-10:39:10] [I] [TRT] Local timing cache in use. Profiling results in this builder pass will not be stored. -[12/28/2023-11:16:43] [I] [TRT] Total Activation Memory: 7962418688 -[12/28/2023-11:16:43] [I] [TRT] Detected 1 inputs and 1 output network tensors. -[12/28/2023-11:16:52] [I] [TRT] Total Host Persistent Memory: 320224 -[12/28/2023-11:16:52] [I] [TRT] Total Device Persistent Memory: 38912 -[12/28/2023-11:16:52] [I] [TRT] Total Scratch Memory: 134217728 -[12/28/2023-11:16:52] [I] [TRT] [MemUsageStats] Peak memory usage of TRT CPU/GPU memory allocators: CPU 188 MiB, GPU 304 MiB -[12/28/2023-11:16:52] [I] [TRT] [BlockAssignment] Started assigning block shifts. This will take 165 steps to complete. -[12/28/2023-11:16:52] [I] [TRT] [BlockAssignment] Algorithm ShiftNTopDown took 49.1135ms to assign 14 blocks to 165 nodes requiring 151026176 bytes. -[12/28/2023-11:16:52] [I] [TRT] Total Activation Memory: 151026176 -[12/28/2023-11:16:56] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +1, GPU +0, now: CPU 1870, GPU 5260 (MiB) -[12/28/2023-11:16:56] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in building engine: CPU +37, GPU +64, now: CPU 37, GPU 64 (MiB) -[12/28/2023-11:16:57] [I] Engine built in 2293.98 sec. -[12/28/2023-11:16:57] [I] [TRT] Loaded engine size: 39 MiB -[12/28/2023-11:16:57] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +0, GPU -1, now: CPU 1274, GPU 5179 (MiB) -[12/28/2023-11:16:57] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +37, now: CPU 0, GPU 37 (MiB) -[12/28/2023-11:16:58] [I] Engine deserialized in 0.373087 sec. -[12/28/2023-11:16:58] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +0, GPU +0, now: CPU 1274, GPU 5179 (MiB) -[12/28/2023-11:16:58] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +144, now: CPU 0, GPU 181 (MiB) -[12/28/2023-11:16:58] [I] Setting persistentCacheLimit to 0 bytes. -[12/28/2023-11:16:58] [I] Using random values for input onnx::Cast_0 -[12/28/2023-11:16:58] [I] Created input binding for onnx::Cast_0 with dimensions 1x3x640x640 -[12/28/2023-11:16:58] [I] Using random values for output graph2_flat_predictions -[12/28/2023-11:16:58] [I] Created output binding for graph2_flat_predictions with dimensions -1x57 -[12/28/2023-11:16:58] [I] Starting inference -[12/28/2023-11:17:13] [I] Warmup completed 7 queries over 200 ms -[12/28/2023-11:17:13] [I] Timing trace has 674 queries over 15.0431 s -[12/28/2023-11:17:13] [I] -[12/28/2023-11:17:13] [I] === Trace details === -[12/28/2023-11:17:13] [I] Trace averages of 100 runs: -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 22.5286 ms - Host latency: 22.6467 ms (enqueue 22.5798 ms) -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 21.9606 ms - Host latency: 22.0766 ms (enqueue 22.0285 ms) -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 21.7562 ms - Host latency: 21.8686 ms (enqueue 21.8288 ms) -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 22.3657 ms - Host latency: 22.4842 ms (enqueue 22.4252 ms) -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 22.2386 ms - Host latency: 22.3543 ms (enqueue 22.2991 ms) -[12/28/2023-11:17:13] [I] Average on 100 runs - GPU latency: 22.0637 ms - Host latency: 22.1763 ms (enqueue 22.1317 ms) -[12/28/2023-11:17:13] [I] -[12/28/2023-11:17:13] [I] === Performance summary === -[12/28/2023-11:17:13] [I] Throughput: 44.8046 qps -[12/28/2023-11:17:13] [I] Latency: min = 20.2925 ms, max = 33.0699 ms, mean = 22.2836 ms, median = 22.1313 ms, percentile(90%) = 22.8691 ms, percentile(95%) = 23.7305 ms, percentile(99%) = 30.8477 ms -[12/28/2023-11:17:13] [I] Enqueue Time: min = 20.2573 ms, max = 32.9944 ms, mean = 22.23 ms, median = 22.0717 ms, percentile(90%) = 22.8018 ms, percentile(95%) = 23.6631 ms, percentile(99%) = 29.8916 ms -[12/28/2023-11:17:13] [I] H2D Latency: min = 0.0800781 ms, max = 0.120361 ms, mean = 0.0955699 ms, median = 0.0957031 ms, percentile(90%) = 0.0986328 ms, percentile(95%) = 0.0991211 ms, percentile(99%) = 0.100586 ms -[12/28/2023-11:17:13] [I] GPU Compute Time: min = 20.1782 ms, max = 32.9492 ms, mean = 22.1676 ms, median = 22.0125 ms, percentile(90%) = 22.7468 ms, percentile(95%) = 23.6074 ms, percentile(99%) = 30.7334 ms -[12/28/2023-11:17:13] [I] D2H Latency: min = 0.00292969 ms, max = 0.0585938 ms, mean = 0.020397 ms, median = 0.0224609 ms, percentile(90%) = 0.0273438 ms, percentile(95%) = 0.0292969 ms, percentile(99%) = 0.0351562 ms -[12/28/2023-11:17:13] [I] Total Host Walltime: 15.0431 s -[12/28/2023-11:17:13] [I] Total GPU Compute Time: 14.941 s -[12/28/2023-11:17:13] [I] Explanations of the performance metrics are printed in the verbose logs. -[12/28/2023-11:17:13] [I] +[01/04/2024-15:38:47] [I] === Model Options === +[01/04/2024-15:38:47] [I] Format: ONNX +[01/04/2024-15:38:47] [I] Model: yolo_nas_pose_m_int8.onnx +[01/04/2024-15:38:47] [I] Output: +[01/04/2024-15:38:47] [I] === Build Options === +[01/04/2024-15:38:47] [I] Max batch: explicit batch +[01/04/2024-15:38:47] [I] Memory Pools: workspace: default, dlaSRAM: default, dlaLocalDRAM: default, dlaGlobalDRAM: default +[01/04/2024-15:38:47] [I] minTiming: 1 +[01/04/2024-15:38:47] [I] avgTiming: 8 +[01/04/2024-15:38:47] [I] Precision: FP32+FP16+INT8 +[01/04/2024-15:38:47] [I] LayerPrecisions: +[01/04/2024-15:38:47] [I] Calibration: Dynamic +[01/04/2024-15:38:47] [I] Refit: Disabled +[01/04/2024-15:38:47] [I] Sparsity: Disabled +[01/04/2024-15:38:47] [I] Safe mode: Disabled +[01/04/2024-15:38:47] [I] DirectIO mode: Disabled +[01/04/2024-15:38:47] [I] Restricted mode: Disabled +[01/04/2024-15:38:47] [I] Build only: Disabled +[01/04/2024-15:38:47] [I] Save engine: yolo_nas_pose_m_int8.onnx.best.engine +[01/04/2024-15:38:47] [I] Load engine: +[01/04/2024-15:38:47] [I] Profiling verbosity: 0 +[01/04/2024-15:38:47] [I] Tactic sources: Using default tactic sources +[01/04/2024-15:38:47] [I] timingCacheMode: local +[01/04/2024-15:38:47] [I] timingCacheFile: +[01/04/2024-15:38:47] [I] Heuristic: Disabled +[01/04/2024-15:38:47] [I] Preview Features: Use default preview flags. +[01/04/2024-15:38:47] [I] Input(s)s format: fp32:CHW +[01/04/2024-15:38:47] [I] Output(s)s format: fp32:CHW +[01/04/2024-15:38:47] [I] Input build shapes: model +[01/04/2024-15:38:47] [I] Input calibration shapes: model +[01/04/2024-15:38:47] [I] === System Options === +[01/04/2024-15:38:47] [I] Device: 0 +[01/04/2024-15:38:47] [I] DLACore: +[01/04/2024-15:38:47] [I] Plugins: +[01/04/2024-15:38:47] [I] === Inference Options === +[01/04/2024-15:38:47] [I] Batch: Explicit +[01/04/2024-15:38:47] [I] Input inference shapes: model +[01/04/2024-15:38:47] [I] Iterations: 10 +[01/04/2024-15:38:47] [I] Duration: 15s (+ 200ms warm up) +[01/04/2024-15:38:47] [I] Sleep time: 0ms +[01/04/2024-15:38:47] [I] Idle time: 0ms +[01/04/2024-15:38:47] [I] Streams: 1 +[01/04/2024-15:38:47] [I] ExposeDMA: Disabled +[01/04/2024-15:38:47] [I] Data transfers: Enabled +[01/04/2024-15:38:47] [I] Spin-wait: Disabled +[01/04/2024-15:38:47] [I] Multithreading: Disabled +[01/04/2024-15:38:47] [I] CUDA Graph: Disabled +[01/04/2024-15:38:47] [I] Separate profiling: Disabled +[01/04/2024-15:38:47] [I] Time Deserialize: Disabled +[01/04/2024-15:38:47] [I] Time Refit: Disabled +[01/04/2024-15:38:47] [I] NVTX verbosity: 0 +[01/04/2024-15:38:47] [I] Persistent Cache Ratio: 0 +[01/04/2024-15:38:47] [I] Inputs: +[01/04/2024-15:38:47] [I] === Reporting Options === +[01/04/2024-15:38:47] [I] Verbose: Disabled +[01/04/2024-15:38:47] [I] Averages: 100 inferences +[01/04/2024-15:38:47] [I] Percentiles: 90,95,99 +[01/04/2024-15:38:47] [I] Dump refittable layers:Disabled +[01/04/2024-15:38:47] [I] Dump output: Disabled +[01/04/2024-15:38:47] [I] Profile: Disabled +[01/04/2024-15:38:47] [I] Export timing to JSON file: +[01/04/2024-15:38:47] [I] Export output to JSON file: +[01/04/2024-15:38:47] [I] Export profile to JSON file: +[01/04/2024-15:38:47] [I] +[01/04/2024-15:38:47] [I] === Device Information === +[01/04/2024-15:38:47] [I] Selected Device: Orin +[01/04/2024-15:38:47] [I] Compute Capability: 8.7 +[01/04/2024-15:38:47] [I] SMs: 8 +[01/04/2024-15:38:47] [I] Compute Clock Rate: 0.624 GHz +[01/04/2024-15:38:47] [I] Device Global Memory: 7471 MiB +[01/04/2024-15:38:47] [I] Shared Memory per SM: 164 KiB +[01/04/2024-15:38:47] [I] Memory Bus Width: 128 bits (ECC disabled) +[01/04/2024-15:38:47] [I] Memory Clock Rate: 0.624 GHz +[01/04/2024-15:38:47] [I] +[01/04/2024-15:38:47] [I] TensorRT version: 8.5.2 +[01/04/2024-15:38:48] [I] [TRT] [MemUsageChange] Init CUDA: CPU +220, GPU +0, now: CPU 249, GPU 3844 (MiB) +[01/04/2024-15:38:50] [I] [TRT] [MemUsageChange] Init builder kernel library: CPU +302, GPU +284, now: CPU 574, GPU 4145 (MiB) +[01/04/2024-15:38:50] [I] Start parsing network model +[01/04/2024-15:38:52] [I] [TRT] ---------------------------------------------------------------- +[01/04/2024-15:38:52] [I] [TRT] Input filename: yolo_nas_pose_m_int8.onnx +[01/04/2024-15:38:52] [I] [TRT] ONNX IR version: 0.0.8 +[01/04/2024-15:38:52] [I] [TRT] Opset version: 17 +[01/04/2024-15:38:52] [I] [TRT] Producer name: pytorch +[01/04/2024-15:38:52] [I] [TRT] Producer version: 2.1.2 +[01/04/2024-15:38:52] [I] [TRT] Domain: +[01/04/2024-15:38:52] [I] [TRT] Model version: 0 +[01/04/2024-15:38:52] [I] [TRT] Doc string: +[01/04/2024-15:38:52] [I] [TRT] ---------------------------------------------------------------- +[01/04/2024-15:38:56] [I] Finish parsing network model +[01/04/2024-15:39:00] [I] [TRT] ---------- Layers Running on DLA ---------- +[01/04/2024-15:39:00] [I] [TRT] ---------- Layers Running on GPU ---------- +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation1] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/pre_process/pre_process.0/Cast.../pre_process/pre_process.2/Mul]} +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1276) [Constant] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1277) [Constant] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONSTANT: (Unnamed Layer* 1278) [Constant] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stem/conv/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stem.conv.rbr_reparam.weight + /model/backbone/stem/conv/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stem/conv/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.downsample.rbr_reparam.weight + /model/backbone/stage1/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/downsample/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv2.conv.weight + /model/backbone/stage1/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv1.conv.weight + /model/backbone/stage1/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage1.blocks.bottlenecks.0.alpha + (Unnamed Layer* 509) [Shuffle] + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage1/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage1.blocks.bottlenecks.1.alpha + (Unnamed Layer* 525) [Shuffle] + /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage1/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage1/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage1.blocks.conv3.conv.weight + /model/backbone/stage1/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage1/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.reduce_skip2.conv.weight + /model/neck/neck2/reduce_skip2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_skip2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.downsample.rbr_reparam.weight + /model/backbone/stage2/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/downsample/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.downsample.conv.weight + /model/neck/neck2/downsample/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/downsample/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv2.conv.weight + /model/backbone/stage2/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv1.conv.weight + /model/backbone/stage2/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.0.alpha + (Unnamed Layer* 572) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.1.alpha + (Unnamed Layer* 588) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage2.blocks.bottlenecks.2.alpha + (Unnamed Layer* 604) [Shuffle] + /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/Mul, /model/backbone/stage2/blocks/bottlenecks/bottlenecks.2/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_2 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage2/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage2.blocks.conv3.conv.weight + /model/backbone/stage2/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage2/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_skip2.conv.weight + /model/neck/neck1/reduce_skip2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_skip2/conv/Conv || model.neck.neck2.reduce_skip1.conv.weight + /model/neck/neck2/reduce_skip1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_skip1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.downsample.rbr_reparam.weight + /model/backbone/stage3/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/downsample/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.downsample.conv.weight + /model/neck/neck1/downsample/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/downsample/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv2.conv.weight + /model/backbone/stage3/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv1.conv.weight + /model/backbone/stage3/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.0.alpha + (Unnamed Layer* 654) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.1.alpha + (Unnamed Layer* 670) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.2.alpha + (Unnamed Layer* 686) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.2/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.3.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.3.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.3.alpha + (Unnamed Layer* 702) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.3/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.4.cv1.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.bottlenecks.4.cv2.rbr_reparam.weight + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage3.blocks.bottlenecks.4.alpha + (Unnamed Layer* 718) [Shuffle] + /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/Mul, /model/backbone/stage3/blocks/bottlenecks/bottlenecks.4/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_4 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_3 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_2 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage3/blocks/conv3/conv/_input_quantizer/QuantizeLinear_clone_0 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage3.blocks.conv3.conv.weight + /model/backbone/stage3/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage3/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_skip1.conv.weight + /model/neck/neck1/reduce_skip1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_skip1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.downsample.rbr_reparam.weight + /model/backbone/stage4/downsample/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/downsample/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv2.conv.weight + /model/backbone/stage4/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv1.conv.weight + /model/backbone/stage4/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage4.blocks.bottlenecks.0.alpha + (Unnamed Layer* 759) [Shuffle] + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/Mul, /model/backbone/stage4/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.backbone.stage4.blocks.bottlenecks.1.alpha + (Unnamed Layer* 775) [Shuffle] + /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/Mul, /model/backbone/stage4/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.stage4.blocks.conv3.conv.weight + /model/backbone/stage4/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/backbone/stage4/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.context_module.cv1.conv.weight + /model/backbone/context_module/cv1/conv/_weight_quantizer/QuantizeLinear + /model/backbone/context_module/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.2/MaxPool +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.1/MaxPool +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POOLING: /model/backbone/context_module/m.0/MaxPool +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/backbone/context_module/m.2/MaxPool_output_0 copy +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.backbone.context_module.cv2.conv.weight + /model/backbone/context_module/cv2/conv/_weight_quantizer/QuantizeLinear + /model/backbone/context_module/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.conv.conv.weight + /model/neck/neck1/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/conv/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck1/upsample/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] DECONVOLUTION: model.neck.neck1.upsample.weight + /model/neck/neck1/upsample/_weight_quantizer/QuantizeLinear + /model/neck/neck1/upsample/ConvTranspose +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.reduce_after_concat.conv.weight + /model/neck/neck1/reduce_after_concat/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/reduce_after_concat/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv2.conv.weight + /model/neck/neck1/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv1.conv.weight + /model/neck/neck1/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck1.blocks.bottlenecks.0.alpha + (Unnamed Layer* 840) [Shuffle] + /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck1/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck1.blocks.bottlenecks.1.alpha + (Unnamed Layer* 856) [Shuffle] + /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck1/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck1.blocks.conv3.conv.weight + /model/neck/neck1/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck1/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.conv.conv.weight + /model/neck/neck2/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/conv/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/upsample/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] DECONVOLUTION: model.neck.neck2.upsample.weight + /model/neck/neck2/upsample/_weight_quantizer/QuantizeLinear + /model/neck/neck2/upsample/ConvTranspose +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/Concat_/model/neck/neck2/reduce_skip1/act/Relu_output_0_clone_1 copy +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.reduce_after_concat.conv.weight + /model/neck/neck2/reduce_after_concat/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/reduce_after_concat/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv2.conv.weight + /model/neck/neck2/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv1.conv.weight + /model/neck/neck2/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.0.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.0.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.0.alpha + (Unnamed Layer* 905) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.1.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.1.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.1.alpha + (Unnamed Layer* 921) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.2.cv1.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv1/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.bottlenecks.2.cv2.rbr_reparam.weight + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/cv2/rbr_reparam/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck2.blocks.bottlenecks.2.alpha + (Unnamed Layer* 937) [Shuffle] + /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/Mul, /model/neck/neck2/blocks/bottlenecks/bottlenecks.2/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck2.blocks.conv3.conv.weight + /model/neck/neck2/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck2/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.bbox_stem.seq.conv.weight + /model/heads/head1/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/bbox_stem/seq/conv/Conv || model.heads.head1.pose_stem.seq.conv.weight + /model/heads/head1/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_stem/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.conv.conv.weight + /model/neck/neck3/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/conv/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.reg_convs.0.seq.conv.weight + /model/heads/head1/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head1.cls_convs.0.seq.conv.weight + /model/heads/head1/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/cls_convs/cls_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_convs.0.seq.conv.weight + /model/heads/head1/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_convs/pose_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck3/blocks/conv1/conv/_input_quantizer/QuantizeLinear_clone_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.cls_pred.weight + /model/heads/head1/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/cls_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.reg_pred.weight + /model/heads/head1/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/reg_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_convs.1.seq.conv.weight + /model/heads/head1/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_convs/pose_convs.1/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv2.conv.weight + /model/neck/neck3/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv1.conv.weight + /model/neck/neck3/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape + /model/heads/Transpose +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head1.pose_pred.weight + /model/heads/head1/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head1/pose_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.0.cv1.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.0.cv2.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck3.blocks.bottlenecks.0.alpha + (Unnamed Layer* 1003) [Shuffle] + /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck3/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.1.cv1.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.bottlenecks.1.cv2.conv.weight + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck3.blocks.bottlenecks.1.alpha + (Unnamed Layer* 1044) [Shuffle] + /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck3/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck3.blocks.conv3.conv.weight + /model/neck/neck3/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck3/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_stem.seq.conv.weight + /model/heads/head2/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_stem/seq/conv/Conv || model.heads.head2.bbox_stem.seq.conv.weight + /model/heads/head2/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/bbox_stem/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.conv.conv.weight + /model/neck/neck4/conv/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/conv/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.reg_convs.0.seq.conv.weight + /model/heads/head2/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head2.cls_convs.0.seq.conv.weight + /model/heads/head2/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/cls_convs/cls_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_convs.0.seq.conv.weight + /model/heads/head2/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_convs/pose_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/conv1/conv/_input_quantizer/QuantizeLinear_clone_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.cls_pred.weight + /model/heads/head2/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/cls_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.reg_pred.weight + /model/heads/head2/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/reg_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_convs.1.seq.conv.weight + /model/heads/head2/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_convs/pose_convs.1/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv2.conv.weight + /model/neck/neck4/blocks/conv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv1.conv.weight + /model/neck/neck4/blocks/conv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape_4 + /model/heads/Transpose_3 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head2.pose_pred.weight + /model/heads/head2/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head2/pose_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.0.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv_1 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.0.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.0.alpha + (Unnamed Layer* 1110) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.0/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.1.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.1.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.1.alpha + (Unnamed Layer* 1151) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.1/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] COPY: /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/_input_quantizer/QuantizeLinear +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.2.cv1.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv1/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.bottlenecks.2.cv2.conv.weight + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv2/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/cv2/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] POINTWISE: PWN(model.neck.neck4.blocks.bottlenecks.2.alpha + (Unnamed Layer* 1167) [Shuffle] + /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/Mul, /model/neck/neck4/blocks/bottlenecks/bottlenecks.2/Add) +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.neck.neck4.blocks.conv3.conv.weight + /model/neck/neck4/blocks/conv3/conv/_weight_quantizer/QuantizeLinear + /model/neck/neck4/blocks/conv3/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.bbox_stem.seq.conv.weight + /model/heads/head3/bbox_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/bbox_stem/seq/conv/Conv || model.heads.head3.pose_stem.seq.conv.weight + /model/heads/head3/pose_stem/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_stem/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.reg_convs.0.seq.conv.weight + /model/heads/head3/reg_convs/reg_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/reg_convs/reg_convs.0/seq/conv/Conv || model.heads.head3.cls_convs.0.seq.conv.weight + /model/heads/head3/cls_convs/cls_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/cls_convs/cls_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.0.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.0/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.0/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.cls_pred.weight + /model/heads/head3/cls_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/cls_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.reg_pred.weight + /model/heads/head3/reg_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/reg_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.1.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.1/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.1/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SHUFFLE: /model/heads/Reshape_8 + /model/heads/Transpose_6 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_convs.2.seq.conv.weight + /model/heads/head3/pose_convs/pose_convs.2/seq/conv/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_convs/pose_convs.2/seq/conv/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] SOFTMAX: /model/heads/Softmax_2 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: model.heads.head3.pose_pred.weight + /model/heads/head3/pose_pred/_weight_quantizer/QuantizeLinear + /model/heads/head3/pose_pred/Conv +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] CONVOLUTION: /model/heads/Conv_2 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/model/heads/head1/Slice_1.../post_process/Reshape_2]} +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] NMS: batched_nms_250 +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] DEVICE_TO_SHAPE_HOST: (Unnamed Layer* 1280) [NMS]_1_output[DevicetoShapeHostCopy] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation2] +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] MYELIN: {ForeignNode[/model/heads/head1/Slice...graph2_/Concat_5]} +[01/04/2024-15:39:00] [I] [TRT] [GpuLayer] TRAIN_STATION: [trainStation3] +[01/04/2024-15:39:01] [I] [TRT] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +534, GPU +595, now: CPU 1413, GPU 5045 (MiB) +[01/04/2024-15:39:02] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +83, GPU +132, now: CPU 1496, GPU 5177 (MiB) +[01/04/2024-15:39:02] [I] [TRT] Local timing cache in use. Profiling results in this builder pass will not be stored. +[01/04/2024-16:16:36] [I] [TRT] Total Activation Memory: 7962286592 +[01/04/2024-16:16:36] [I] [TRT] Detected 1 inputs and 1 output network tensors. +[01/04/2024-16:16:45] [I] [TRT] Total Host Persistent Memory: 318176 +[01/04/2024-16:16:45] [I] [TRT] Total Device Persistent Memory: 38912 +[01/04/2024-16:16:45] [I] [TRT] Total Scratch Memory: 134217728 +[01/04/2024-16:16:45] [I] [TRT] [MemUsageStats] Peak memory usage of TRT CPU/GPU memory allocators: CPU 188 MiB, GPU 304 MiB +[01/04/2024-16:16:45] [I] [TRT] [BlockAssignment] Started assigning block shifts. This will take 171 steps to complete. +[01/04/2024-16:16:45] [I] [TRT] [BlockAssignment] Algorithm ShiftNTopDown took 84.0411ms to assign 13 blocks to 171 nodes requiring 151022080 bytes. +[01/04/2024-16:16:45] [I] [TRT] Total Activation Memory: 151022080 +[01/04/2024-16:16:49] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +0, GPU +0, now: CPU 1869, GPU 5974 (MiB) +[01/04/2024-16:16:49] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in building engine: CPU +37, GPU +64, now: CPU 37, GPU 64 (MiB) +[01/04/2024-16:16:49] [I] Engine built in 2282.06 sec. +[01/04/2024-16:16:50] [I] [TRT] Loaded engine size: 39 MiB +[01/04/2024-16:16:50] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +0, GPU +0, now: CPU 1274, GPU 5805 (MiB) +[01/04/2024-16:16:50] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +37, now: CPU 0, GPU 37 (MiB) +[01/04/2024-16:16:50] [I] Engine deserialized in 0.215387 sec. +[01/04/2024-16:16:50] [I] [TRT] [MemUsageChange] Init cuDNN: CPU +0, GPU +0, now: CPU 1274, GPU 5805 (MiB) +[01/04/2024-16:16:50] [I] [TRT] [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +144, now: CPU 0, GPU 181 (MiB) +[01/04/2024-16:16:50] [I] Setting persistentCacheLimit to 0 bytes. +[01/04/2024-16:16:50] [I] Using random values for input onnx::Cast_0 +[01/04/2024-16:16:50] [I] Created input binding for onnx::Cast_0 with dimensions 1x3x640x640 +[01/04/2024-16:16:50] [I] Using random values for output graph2_flat_predictions +[01/04/2024-16:16:50] [I] Created output binding for graph2_flat_predictions with dimensions -1x57 +[01/04/2024-16:16:50] [I] Starting inference +[01/04/2024-16:17:05] [I] Warmup completed 7 queries over 200 ms +[01/04/2024-16:17:05] [I] Timing trace has 710 queries over 15.0472 s +[01/04/2024-16:17:05] [I] +[01/04/2024-16:17:05] [I] === Trace details === +[01/04/2024-16:17:05] [I] Trace averages of 100 runs: +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 21.0753 ms - Host latency: 21.189 ms (enqueue 21.1436 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 21.5315 ms - Host latency: 21.6483 ms (enqueue 21.5925 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 21.016 ms - Host latency: 21.1284 ms (enqueue 21.0889 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 20.9846 ms - Host latency: 21.0969 ms (enqueue 21.0589 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 21.0441 ms - Host latency: 21.1555 ms (enqueue 21.1173 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 20.8767 ms - Host latency: 20.989 ms (enqueue 20.9487 ms) +[01/04/2024-16:17:05] [I] Average on 100 runs - GPU latency: 20.8415 ms - Host latency: 20.9534 ms (enqueue 20.9146 ms) +[01/04/2024-16:17:05] [I] +[01/04/2024-16:17:05] [I] === Performance summary === +[01/04/2024-16:17:05] [I] Throughput: 47.1848 qps +[01/04/2024-16:17:05] [I] Latency: min = 19.9678 ms, max = 24.1187 ms, mean = 21.1609 ms, median = 21.0818 ms, percentile(90%) = 22.0253 ms, percentile(95%) = 22.418 ms, percentile(99%) = 22.8726 ms +[01/04/2024-16:17:05] [I] Enqueue Time: min = 19.918 ms, max = 23.605 ms, mean = 21.1187 ms, median = 21.0505 ms, percentile(90%) = 21.9624 ms, percentile(95%) = 22.3008 ms, percentile(99%) = 22.81 ms +[01/04/2024-16:17:05] [I] H2D Latency: min = 0.0810547 ms, max = 0.125977 ms, mean = 0.0965348 ms, median = 0.0966797 ms, percentile(90%) = 0.0986328 ms, percentile(95%) = 0.0991211 ms, percentile(99%) = 0.0996704 ms +[01/04/2024-16:17:05] [I] GPU Compute Time: min = 19.8594 ms, max = 24.0195 ms, mean = 21.0479 ms, median = 20.9727 ms, percentile(90%) = 21.9004 ms, percentile(95%) = 22.3013 ms, percentile(99%) = 22.7523 ms +[01/04/2024-16:17:05] [I] D2H Latency: min = 0.00292969 ms, max = 0.0576172 ms, mean = 0.0164532 ms, median = 0.0146484 ms, percentile(90%) = 0.0244141 ms, percentile(95%) = 0.0263672 ms, percentile(99%) = 0.0302734 ms +[01/04/2024-16:17:05] [I] Total Host Walltime: 15.0472 s +[01/04/2024-16:17:05] [I] Total GPU Compute Time: 14.944 s +[01/04/2024-16:17:05] [I] Explanations of the performance metrics are printed in the verbose logs. +[01/04/2024-16:17:05] [I] &&&& PASSED TensorRT.trtexec [TensorRT v8502] # /usr/src/tensorrt/bin/trtexec --onnx=yolo_nas_pose_m_int8.onnx --best --avgRuns=100 --duration=15 --saveEngine=yolo_nas_pose_m_int8.onnx.best.engine