Yolov5s run on jetson nx probloms

957631806 · July 6, 2022, 4:09am

hello all，now i want to port yolov5s to jetson nx。
1，convert yolov5s.pt to yolov5s.onnx
(yolov5s.pt download from here https://github.com/ultralytics/yolov5/releases/download/v6.1/yolov5s.pt)
(convert to onnx command:python3 export.py --weights yolov5s.pt --include onnx --simplify)
2，convert to tersorrt engine use the command: ./trtexec --onnx=yolov5s.onnx --saveEngine=yolov5s.trt
the output is，note the int64 to int32 warning
./trtexec --onnx=yolov5s.onnx --saveEngine=yolov5s.trt
&&&& RUNNING TensorRT.trtexec # ./trtexec --onnx=yolov5s.onnx --saveEngine=yolov5s.trt
[07/06/2022-11:51:37] [I] === Model Options ===
[07/06/2022-11:51:37] [I] Format: ONNX
[07/06/2022-11:51:37] [I] Model: yolov5s.onnx
[07/06/2022-11:51:37] [I] Output:
[07/06/2022-11:51:37] [I] === Build Options ===
[07/06/2022-11:51:37] [I] Max batch: 1
[07/06/2022-11:51:37] [I] Workspace: 16 MB
[07/06/2022-11:51:37] [I] minTiming: 1
[07/06/2022-11:51:37] [I] avgTiming: 8
[07/06/2022-11:51:37] [I] Precision: FP32
[07/06/2022-11:51:37] [I] Calibration:
[07/06/2022-11:51:37] [I] Safe mode: Disabled
[07/06/2022-11:51:37] [I] Save engine: yolov5s.trt
[07/06/2022-11:51:37] [I] Load engine:
[07/06/2022-11:51:37] [I] Builder Cache: Enabled
[07/06/2022-11:51:37] [I] NVTX verbosity: 0
[07/06/2022-11:51:37] [I] Inputs format: fp32:CHW
[07/06/2022-11:51:37] [I] Outputs format: fp32:CHW
[07/06/2022-11:51:37] [I] Input build shapes: model
[07/06/2022-11:51:37] [I] Input calibration shapes: model
[07/06/2022-11:51:37] [I] === System Options ===
[07/06/2022-11:51:37] [I] Device: 0
[07/06/2022-11:51:37] [I] DLACore:
[07/06/2022-11:51:37] [I] Plugins:
[07/06/2022-11:51:37] [I] === Inference Options ===
[07/06/2022-11:51:37] [I] Batch: 1
[07/06/2022-11:51:37] [I] Input inference shapes: model
[07/06/2022-11:51:37] [I] Iterations: 10
[07/06/2022-11:51:37] [I] Duration: 3s (+ 200ms warm up)
[07/06/2022-11:51:37] [I] Sleep time: 0ms
[07/06/2022-11:51:37] [I] Streams: 1
[07/06/2022-11:51:37] [I] ExposeDMA: Disabled
[07/06/2022-11:51:37] [I] Spin-wait: Disabled
[07/06/2022-11:51:37] [I] Multithreading: Disabled
[07/06/2022-11:51:37] [I] CUDA Graph: Disabled
[07/06/2022-11:51:37] [I] Skip inference: Disabled
[07/06/2022-11:51:37] [I] Inputs:
[07/06/2022-11:51:37] [I] === Reporting Options ===
[07/06/2022-11:51:37] [I] Verbose: Disabled
[07/06/2022-11:51:37] [I] Averages: 10 inferences
[07/06/2022-11:51:37] [I] Percentile: 99
[07/06/2022-11:51:37] [I] Dump output: Disabled
[07/06/2022-11:51:37] [I] Profile: Disabled
[07/06/2022-11:51:37] [I] Export timing to JSON file:
[07/06/2022-11:51:37] [I] Export output to JSON file:
[07/06/2022-11:51:37] [I] Export profile to JSON file:
[07/06/2022-11:51:37] [I]

Input filename: yolov5s.onnx
ONNX IR version: 0.0.6
Opset version: 12
Producer name: pytorch
Producer version: 1.8
Domain:
Model version: 0
Doc string:

[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:220: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:40] [W] [TRT] onnx2trt_utils.cpp:246: One or more weights outside the range of INT32 was clamped
[07/06/2022-11:51:41] [I] [TRT]
[07/06/2022-11:51:41] [I] [TRT] --------------- Layers running on DLA:
[07/06/2022-11:51:41] [I] [TRT]
[07/06/2022-11:51:41] [I] [TRT] --------------- Layers running on GPU:
[07/06/2022-11:51:41] [I] [TRT] Slice_4, Slice_9, Slice_14, Slice_19, Slice_24, Slice_29, Slice_34, Slice_39, Conv_41, PWN(PWN((Unnamed Layer* 14) [Constant] + (Unnamed Layer* 15) [Shuffle], PWN(PWN((Unnamed Layer* 10) [Constant] + (Unnamed Layer* 11) [Shuffle] + Add_43, Clip_46), Div_48)), Mul_49), Conv_50, PWN(PWN((Unnamed Layer* 23) [Constant] + (Unnamed Layer* 24) [Shuffle], PWN(PWN((Unnamed Layer* 19) [Constant] + (Unnamed Layer* 20) [Shuffle] + Add_52, Clip_55), Div_57)), Mul_58), Conv_88 || Conv_59, PWN(PWN((Unnamed Layer* 32) [Constant] + (Unnamed Layer* 33) [Shuffle], PWN(PWN((Unnamed Layer* 28) [Constant] + (Unnamed Layer* 29) [Shuffle] + Add_61, Clip_64), Div_66)), Mul_67), Conv_68, PWN(PWN((Unnamed Layer* 41) [Constant] + (Unnamed Layer* 42) [Shuffle], PWN(PWN((Unnamed Layer* 37) [Constant] + (Unnamed Layer* 38) [Shuffle] + Add_70, Clip_73), Div_75)), Mul_76), Conv_77, PWN(PWN(PWN((Unnamed Layer* 50) [Constant] + (Unnamed Layer* 51) [Shuffle], PWN(PWN((Unnamed Layer* 46) [Constant] + (Unnamed Layer* 47) [Shuffle] + Add_79, Clip_82), Div_84)), Mul_85), Add_86), Conv_87, 254 copy, BatchNormalization_90, LeakyRelu_91, Conv_92, PWN(PWN((Unnamed Layer* 65) [Constant] + (Unnamed Layer* 66) [Shuffle], PWN(PWN((Unnamed Layer* 61) [Constant] + (Unnamed Layer* 62) [Shuffle] + Add_94, Clip_97), Div_99)), Mul_100), Conv_101, PWN(PWN((Unnamed Layer* 74) [Constant] + (Unnamed Layer* 75) [Shuffle], PWN(PWN((Unnamed Layer* 70) [Constant] + (Unnamed Layer* 71) [Shuffle] + Add_103, Clip_106), Div_108)), Mul_109), Conv_177 || Conv_110, PWN(PWN((Unnamed Layer* 83) [Constant] + (Unnamed Layer* 84) [Shuffle], PWN(PWN((Unnamed Layer* 79) [Constant] + (Unnamed Layer* 80) [Shuffle] + Add_112, Clip_115), Div_117)), Mul_118), Conv_119, PWN(PWN((Unnamed Layer* 92) [Constant] + (Unnamed Layer* 93) [Shuffle], PWN(PWN((Unnamed Layer* 88) [Constant] + (Unnamed Layer* 89) [Shuffle] + Add_121, Clip_124), Div_126)), Mul_127), Conv_128, PWN(PWN(PWN((Unnamed Layer* 101) [Constant] + (Unnamed Layer* 102) [Shuffle], PWN(PWN((Unnamed Layer* 97) [Constant] + (Unnamed Layer* 98) [Shuffle] + Add_130, Clip_133), Div_135)), Mul_136), Add_137), Conv_138, PWN(PWN((Unnamed Layer* 111) [Constant] + (Unnamed Layer* 112) [Shuffle], PWN(PWN((Unnamed Layer* 107) [Constant] + (Unnamed Layer* 108) [Shuffle] + Add_140, Clip_143), Div_145)), Mul_146), Conv_147, PWN(PWN(PWN((Unnamed Layer* 120) [Constant] + (Unnamed Layer* 121) [Shuffle], PWN(PWN((Unnamed Layer* 116) [Constant] + (Unnamed Layer* 117) [Shuffle] + Add_149, Clip_152), Div_154)), Mul_155), Add_156), Conv_157, PWN(PWN((Unnamed Layer* 130) [Constant] + (Unnamed Layer* 131) [Shuffle], PWN(PWN((Unnamed Layer* 126) [Constant] + (Unnamed Layer* 127) [Shuffle] + Add_159, Clip_162), Div_164)), Mul_165), Conv_166, PWN(PWN(PWN((Unnamed Layer* 139) [Constant] + (Unnamed Layer* 140) [Shuffle], PWN(PWN((Unnamed Layer* 135) [Constant] + (Unnamed Layer* 136) [Shuffle] + Add_168, Clip_171), Div_173)), Mul_174), Add_175), Conv_176, 343 copy, BatchNormalization_179, LeakyRelu_180, Conv_181, PWN(PWN((Unnamed Layer* 154) [Constant] + (Unnamed Layer* 155) [Shuffle], PWN(PWN((Unnamed Layer* 150) [Constant] + (Unnamed Layer* 151) [Shuffle] + Add_183, Clip_186), Div_188)), Mul_189), Conv_190, PWN(PWN((Unnamed Layer* 163) [Constant] + (Unnamed Layer* 164) [Shuffle], PWN(PWN((Unnamed Layer* 159) [Constant] + (Unnamed Layer* 160) [Shuffle] + Add_192, Clip_195), Div_197)), Mul_198), Conv_266 || Conv_199, PWN(PWN((Unnamed Layer* 172) [Constant] + (Unnamed Layer* 173) [Shuffle], PWN(PWN((Unnamed Layer* 168) [Constant] + (Unnamed Layer* 169) [Shuffle] + Add_201, Clip_204), Div_206)), Mul_207), Conv_208, PWN(PWN((Unnamed Layer* 181) [Constant] + (Unnamed Layer* 182) [Shuffle], PWN(PWN((Unnamed Layer* 177) [Constant] + (Unnamed Layer* 178) [Shuffle] + Add_210, Clip_213), Div_215)), Mul_216), Conv_217, PWN(PWN(PWN((Unnamed Layer* 190) [Constant] + (Unnamed Layer* 191) [Shuffle], PWN(PWN((Unnamed Layer* 186) [Constant] + (Unnamed Layer* 187) [Shuffle] + Add_219, Clip_222), Div_224)), Mul_225), Add_226), Conv_227, PWN(PWN((Unnamed Layer* 200) [Constant] + (Unnamed Layer* 201) [Shuffle], PWN(PWN((Unnamed Layer* 196) [Constant] + (Unnamed Layer* 197) [Shuffle] + Add_229, Clip_232), Div_234)), Mul_235), Conv_236, PWN(PWN(PWN((Unnamed Layer* 209) [Constant] + (Unnamed Layer* 210) [Shuffle], PWN(PWN((Unnamed Layer* 205) [Constant] + (Unnamed Layer* 206) [Shuffle] + Add_238, Clip_241), Div_243)), Mul_244), Add_245), Conv_246, PWN(PWN((Unnamed Layer* 219) [Constant] + (Unnamed Layer* 220) [Shuffle], PWN(PWN((Unnamed Layer* 215) [Constant] + (Unnamed Layer* 216) [Shuffle] + Add_248, Clip_251), Div_253)), Mul_254), Conv_255, PWN(PWN(PWN((Unnamed Layer* 228) [Constant] + (Unnamed Layer* 229) [Shuffle], PWN(PWN((Unnamed Layer* 224) [Constant] + (Unnamed Layer* 225) [Shuffle] + Add_257, Clip_260), Div_262)), Mul_263), Add_264), Conv_265, 432 copy, BatchNormalization_268, LeakyRelu_269, Conv_270, PWN(PWN((Unnamed Layer* 243) [Constant] + (Unnamed Layer* 244) [Shuffle], PWN(PWN((Unnamed Layer* 239) [Constant] + (Unnamed Layer* 240) [Shuffle] + Add_272, Clip_275), Div_277)), Mul_278), Conv_279, PWN(PWN((Unnamed Layer* 252) [Constant] + (Unnamed Layer* 253) [Shuffle], PWN(PWN((Unnamed Layer* 248) [Constant] + (Unnamed Layer* 249) [Shuffle] + Add_281, Clip_284), Div_286)), Mul_287), Conv_288, PWN(PWN((Unnamed Layer* 261) [Constant] + (Unnamed Layer* 262) [Shuffle], PWN(PWN((Unnamed Layer* 257) [Constant] + (Unnamed Layer* 258) [Shuffle] + Add_290, Clip_293), Div_295)), Mul_296), MaxPool_299, MaxPool_298, MaxPool_297, 462 copy, Conv_301, PWN(PWN((Unnamed Layer* 274) [Constant] + (Unnamed Layer* 275) [Shuffle], PWN(PWN((Unnamed Layer* 270) [Constant] + (Unnamed Layer* 271) [Shuffle] + Add_303, Clip_306), Div_308)), Mul_309), Conv_338 || Conv_310, PWN(PWN((Unnamed Layer* 283) [Constant] + (Unnamed Layer* 284) [Shuffle], PWN(PWN((Unnamed Layer* 279) [Constant] + (Unnamed Layer* 280) [Shuffle] + Add_312, Clip_315), Div_317)), Mul_318), Conv_319, PWN(PWN((Unnamed Layer* 292) [Constant] + (Unnamed Layer* 293) [Shuffle], PWN(PWN((Unnamed Layer* 288) [Constant] + (Unnamed Layer* 289) [Shuffle] + Add_321, Clip_324), Div_326)), Mul_327), Conv_328, PWN(PWN((Unnamed Layer* 301) [Constant] + (Unnamed Layer* 302) [Shuffle], PWN(PWN((Unnamed Layer* 297) [Constant] + (Unnamed Layer* 298) [Shuffle] + Add_330, Clip_333), Div_335)), Mul_336), Conv_337, 504 copy, BatchNormalization_340, LeakyRelu_341, Conv_342, PWN(PWN((Unnamed Layer* 315) [Constant] + (Unnamed Layer* 316) [Shuffle], PWN(PWN((Unnamed Layer* 311) [Constant] + (Unnamed Layer* 312) [Shuffle] + Add_344, Clip_347), Div_349)), Mul_350), Conv_351, PWN(PWN((Unnamed Layer* 324) [Constant] + (Unnamed Layer* 325) [Shuffle], PWN(PWN((Unnamed Layer* 320) [Constant] + (Unnamed Layer* 321) [Shuffle] + Add_353, Clip_356), Div_358)), Mul_359), Resize_361, 530 copy, Conv_391 || Conv_363, PWN(PWN((Unnamed Layer* 335) [Constant] + (Unnamed Layer* 336) [Shuffle], PWN(PWN((Unnamed Layer* 331) [Constant] + (Unnamed Layer* 332) [Shuffle] + Add_365, Clip_368), Div_370)), Mul_371), Conv_372, PWN(PWN((Unnamed Layer* 344) [Constant] + (Unnamed Layer* 345) [Shuffle], PWN(PWN((Unnamed Layer* 340) [Constant] + (Unnamed Layer* 341) [Shuffle] + Add_374, Clip_377), Div_379)), Mul_380), Conv_381, PWN(PWN((Unnamed Layer* 353) [Constant] + (Unnamed Layer* 354) [Shuffle], PWN(PWN((Unnamed Layer* 349) [Constant] + (Unnamed Layer* 350) [Shuffle] + Add_383, Clip_386), Div_388)), Mul_389), Conv_390, 560 copy, BatchNormalization_393, LeakyRelu_394, Conv_395, PWN(PWN((Unnamed Layer* 367) [Constant] + (Unnamed Layer* 368) [Shuffle], PWN(PWN((Unnamed Layer* 363) [Constant] + (Unnamed Layer* 364) [Shuffle] + Add_397, Clip_400), Div_402)), Mul_403), Conv_404, PWN(PWN((Unnamed Layer* 376) [Constant] + (Unnamed Layer* 377) [Shuffle], PWN(PWN((Unnamed Layer* 372) [Constant] + (Unnamed Layer* 373) [Shuffle] + Add_406, Clip_409), Div_411)), Mul_412), Resize_414, 586 copy, Conv_444 || Conv_416, PWN(PWN((Unnamed Layer* 387) [Constant] + (Unnamed Layer* 388) [Shuffle], PWN(PWN((Unnamed Layer* 383) [Constant] + (Unnamed Layer* 384) [Shuffle] + Add_418, Clip_421), Div_423)), Mul_424), Conv_425, PWN(PWN((Unnamed Layer* 396) [Constant] + (Unnamed Layer* 397) [Shuffle], PWN(PWN((Unnamed Layer* 392) [Constant] + (Unnamed Layer* 393) [Shuffle] + Add_427, Clip_430), Div_432)), Mul_433), Conv_434, PWN(PWN((Unnamed Layer* 405) [Constant] + (Unnamed Layer* 406) [Shuffle], PWN(PWN((Unnamed Layer* 401) [Constant] + (Unnamed Layer* 402) [Shuffle] + Add_436, Clip_439), Div_441)), Mul_442), Conv_443, 616 copy, BatchNormalization_446, LeakyRelu_447, Conv_448, PWN(PWN((Unnamed Layer* 419) [Constant] + (Unnamed Layer* 420) [Shuffle], PWN(PWN((Unnamed Layer* 415) [Constant] + (Unnamed Layer* 416) [Shuffle] + Add_450, Clip_453), Div_455)), Mul_456), Conv_559, Reshape_573 + Transpose_574, Sigmoid_575, Split_576_2, Split_576_1, Split_576, Conv_457, PWN(PWN((Unnamed Layer* 428) [Constant] + (Unnamed Layer* 429) [Shuffle], PWN(PWN((Unnamed Layer* 424) [Constant] + (Unnamed Layer* 425) [Shuffle] + Add_459, Clip_462), Div_464)), Mul_465), 581 copy, (Unnamed Layer* 532) [Constant] + (Unnamed Layer* 533) [Shuffle] + Mul_578, Conv_495 || Conv_467, PWN(PWN((Unnamed Layer* 438) [Constant] + (Unnamed Layer* 439) [Shuffle], PWN(PWN((Unnamed Layer* 434) [Constant] + (Unnamed Layer* 435) [Shuffle] + Add_469, Clip_472), Div_474)), Mul_475), PWN((Unnamed Layer* 543) [Constant] + (Unnamed Layer* 544) [Shuffle], PWN((Unnamed Layer* 540) [Constant] + (Unnamed Layer* 541) [Shuffle] + Mul_584, Pow_586)), (Unnamed Layer* 535) [Constant] + Add_580, Conv_476, PWN(PWN((Unnamed Layer* 447) [Constant] + (Unnamed Layer* 448) [Shuffle], PWN(PWN((Unnamed Layer* 443) [Constant] + (Unnamed Layer* 444) [Shuffle] + Add_478, Clip_481), Div_483)), Mul_484), (Unnamed Layer* 546) [Constant] + Mul_588, (Unnamed Layer* 537) [Constant] + (Unnamed Layer* 538) [Shuffle] + Mul_582, 760 copy, 766 copy, 754 copy, Reshape_592, Conv_485, PWN(PWN((Unnamed Layer* 456) [Constant] + (Unnamed Layer* 457) [Shuffle], PWN(PWN((Unnamed Layer* 452) [Constant] + (Unnamed Layer* 453) [Shuffle] + Add_487, Clip_490), Div_492)), Mul_493), Conv_494, 666 copy, 667 copy, BatchNormalization_497, LeakyRelu_498, Conv_499, PWN(PWN((Unnamed Layer* 470) [Constant] + (Unnamed Layer* 471) [Shuffle], PWN(PWN((Unnamed Layer* 466) [Constant] + (Unnamed Layer* 467) [Shuffle] + Add_501, Clip_504), Div_506)), Mul_507), Conv_593, Reshape_607 + Transpose_608, Sigmoid_609, Split_610_2, Split_610_1, Split_610, Conv_508, PWN(PWN((Unnamed Layer* 479) [Constant] + (Unnamed Layer* 480) [Shuffle], PWN(PWN((Unnamed Layer* 475) [Constant] + (Unnamed Layer* 476) [Shuffle] + Add_510, Clip_513), Div_515)), Mul_516), 688 copy, 525 copy, (Unnamed Layer* 557) [Constant] + (Unnamed Layer* 558) [Shuffle] + Mul_612, Conv_546 || Conv_518, PWN(PWN((Unnamed Layer* 489) [Constant] + (Unnamed Layer* 490) [Shuffle], PWN(PWN((Unnamed Layer* 485) [Constant] + (Unnamed Layer* 486) [Shuffle] + Add_520, Clip_523), Div_525)), Mul_526), PWN((Unnamed Layer* 568) [Constant] + (Unnamed Layer* 569) [Shuffle], PWN((Unnamed Layer* 565) [Constant] + (Unnamed Layer* 566) [Shuffle] + Mul_618, Pow_620)), (Unnamed Layer* 560) [Constant] + Add_614, Conv_527, PWN(PWN((Unnamed Layer* 498) [Constant] + (Unnamed Layer* 499) [Shuffle], PWN(PWN((Unnamed Layer* 494) [Constant] + (Unnamed Layer* 495) [Shuffle] + Add_529, Clip_532), Div_534)), Mul_535), (Unnamed Layer* 571) [Constant] + Mul_622, (Unnamed Layer* 562) [Constant] + (Unnamed Layer* 563) [Shuffle] + Mul_616, 804 copy, 810 copy, 798 copy, Reshape_626, Conv_536, PWN(PWN((Unnamed Layer* 507) [Constant] + (Unnamed Layer* 508) [Shuffle], PWN(PWN((Unnamed Layer* 503) [Constant] + (Unnamed Layer* 504) [Shuffle] + Add_538, Clip_541), Div_543)), Mul_544), Conv_545, 717 copy, 718 copy, BatchNormalization_548, LeakyRelu_549, Conv_550, PWN(PWN((Unnamed Layer* 521) [Constant] + (Unnamed Layer* 522) [Shuffle], PWN(PWN((Unnamed Layer* 517) [Constant] + (Unnamed Layer* 518) [Shuffle] + Add_552, Clip_555), Div_557)), Mul_558), Conv_627, Reshape_641 + Transpose_642, Sigmoid_643, Split_644_2, Split_644_1, Split_644, (Unnamed Layer* 582) [Constant] + (Unnamed Layer* 583) [Shuffle] + Mul_646, PWN((Unnamed Layer* 593) [Constant] + (Unnamed Layer* 594) [Shuffle], PWN((Unnamed Layer* 590) [Constant] + (Unnamed Layer* 591) [Shuffle] + Mul_652, Pow_654)), (Unnamed Layer* 585) [Constant] + Add_648, (Unnamed Layer* 596) [Constant] + Mul_656, (Unnamed Layer* 587) [Constant] + (Unnamed Layer* 588) [Shuffle] + Mul_650, 848 copy, 854 copy, 842 copy, Reshape_660, 774 copy, 818 copy, 862 copy,
[07/06/2022-11:51:50] [I] [TRT] Some tactics do not have sufficient workspace memory to run. Increasing workspace size may increase performance, please check verbose output.
[07/06/2022-11:55:43] [I] [TRT] Detected 1 inputs and 4 output network tensors.
[07/06/2022-11:55:44] [I] Starting inference threads
[07/06/2022-11:55:47] [I] Warmup completed 5 queries over 200 ms
[07/06/2022-11:55:47] [I] Timing trace has 43 queries over 3.15354 s
[07/06/2022-11:55:47] [I] Trace averages of 10 runs:
[07/06/2022-11:55:47] [I] Average on 10 runs - GPU latency: 58.0234 ms - Host latency: 58.8179 ms (end to end 58.889 ms, enqueue 7.25994 ms)
[07/06/2022-11:55:47] [I] Average on 10 runs - GPU latency: 77.7581 ms - Host latency: 78.5935 ms (end to end 78.6058 ms, enqueue 5.8263 ms)
[07/06/2022-11:55:47] [I] Average on 10 runs - GPU latency: 82.3244 ms - Host latency: 83.1825 ms (end to end 83.3682 ms, enqueue 5.49225 ms)
[07/06/2022-11:55:47] [I] Average on 10 runs - GPU latency: 70.3519 ms - Host latency: 71.1755 ms (end to end 71.2973 ms, enqueue 4.90483 ms)
[07/06/2022-11:55:47] [I] Host Latency
[07/06/2022-11:55:47] [I] min: 46.4056 ms (end to end 46.4195 ms)
[07/06/2022-11:55:47] [I] max: 113.098 ms (end to end 113.105 ms)
[07/06/2022-11:55:47] [I] mean: 73.2449 ms (end to end 73.3366 ms)
[07/06/2022-11:55:47] [I] median: 60.6202 ms (end to end 60.627 ms)
[07/06/2022-11:55:47] [I] percentile: 113.098 ms at 99% (end to end 113.105 ms at 99%)
[07/06/2022-11:55:47] [I] throughput: 13.6355 qps
[07/06/2022-11:55:47] [I] walltime: 3.15354 s
[07/06/2022-11:55:47] [I] Enqueue Time
[07/06/2022-11:55:47] [I] min: 4.31714 ms
[07/06/2022-11:55:47] [I] max: 10.6324 ms
[07/06/2022-11:55:47] [I] median: 5.27222 ms
[07/06/2022-11:55:47] [I] GPU Compute
[07/06/2022-11:55:47] [I] min: 45.7134 ms
[07/06/2022-11:55:47] [I] max: 112.401 ms
[07/06/2022-11:55:47] [I] mean: 72.4188 ms
[07/06/2022-11:55:47] [I] median: 59.8231 ms
[07/06/2022-11:55:47] [I] percentile: 112.401 ms at 99%
[07/06/2022-11:55:47] [I] total compute time: 3.11401 s
&&&& PASSED TensorRT.trtexec # ./trtexec --onnx=yolov5s.onnx --saveEngine=yolov5s.trt
3，convert to tensorrt engine seems ok excepts the warning，but when I user tensorrt to infer，the output is Partial correct。I compare the tensorrt engine with the onnx output which run on cpu，214200(25200*85) floats，33607 is not equal，the other is equal.
4，my jetson info is

NVIDIA Jetson Xavier NX (Developer Kit Version)
- Jetpack 4.4.1 [L4T 32.4.4]
- NV Power Mode: MODE_15W_6CORE - Type: 2
- jetson_stats.service: active
Libraries:
- CUDA: 10.2.89
- cuDNN: 8.0.0.180
- TensorRT: 7.1.3.0
- Visionworks: NOT_INSTALLED
- OpenCV: 4.1.1 compiled CUDA: NO
- VPI: NOT_INSTALLED
- Vulkan: 1.2.70
  Please correct me if I am wrong. thanyou very much

TomNVIDIA · July 6, 2022, 4:27pm

Hi,

I am moving your topic to the Jetson Xavier NX category so our support team has visibility.

AastaLLL · July 7, 2022, 3:53am

Hi,

We have several newer releases already.
Could you upgrade your environment to JetPack 4.6.2 or 5.0.1 DP and try it again?

Thanks.

957631806 · July 7, 2022, 5:48am

how can i upgrade the jetpack version？

AastaLLL · July 8, 2022, 2:23am

Hi,

For JetPack 4.6.x, you can find the SD card image:

For JetPack 5.0, please reflash your device with SDK manager.

Thanks.

system · July 27, 2022, 4:02am

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.

Yolov5s run on jetson nx probloms

Input filename: yolov5s.onnx ONNX IR version: 0.0.6 Opset version: 12 Producer name: pytorch Producer version: 1.8 Domain: Model version: 0 Doc string:

Input filename: yolov5s.onnx
ONNX IR version: 0.0.6
Opset version: 12
Producer name: pytorch
Producer version: 1.8
Domain:
Model version: 0
Doc string: