Open main menu

DAVE Developer's Wiki β

Changes

ML-TN-001 - AI at the edge: comparison of different embedded platforms - Part 3

445 bytes added, 13:48, 12 October 2020
no edit summary
red_apple_1.jpg
[Time] LoadImage 0.0128658 0161657 ms[Time] PreprocessImage 0.00510821 00521947 ms[Time] SetInputImageInHWCFP32 0.00115788 00115946 ms[DPU Time] dpuSetInputTensorInHWCFP32 1.70365 67499 ms[DPU Time] dpuRunTask 1.96536 99908 ms[DPU Time] dpuGetOutputTensorInHWCFP32 0.00884 00864 ms[DPU Time] dpuRunSoftmax 0.00472 00476 ms[DPU tot time] 3.68257 68747 ms[DPU throughput] 271.549 189 FPS
[Time] CpuArgmax<float> 1.2e3e-07 ms[Time] CpuSoftmax 91.6e01e-07 06 ms[Time] RunCustomCNN 0.0108261 0109225 ms[Time] TopK 8.17e04e-06 ms1) red_apple : 0.999664999665
2) orange : 0.00033535
3) avocado hand : 18.85477e76131e-07084) hand avocado : 19.52249e23435e-08095) banana : 1.13961e38833e-12116) green_apple : 3.13808e44132e-1714
_______________________________________________________________
red_apple_1.jpg
[Time] LoadImage 0.012864 0163798 ms[Time] PreprocessImage 0.0051113 00518644 ms[Time] SetInputImageInHWCFP32 0.00120038 00132577 ms[DNNDK] Performance profile - DPU Kernel "pruned_dbg_custom_cnn_0dbg_custom_cnn_0" DPU Task "pruned_dbg_custom_cnn_0dbg_custom_cnn_0-1"
=====================================================================================================
ID NodeName Workload(MOP) Mem(MB) RunTime(ms) Perf(GOPS) Utilization MB/S
1 conv2d_1_Conv2D 85.163 0.53 0.362 363 235234.3 6 19.1% 14571453.7 2 conv2d_2_Conv2D 218.991 0.48 0.206 205 10631068.1 2 86.59% 23182330.81 3 conv2d_3_Conv2D 99.680 0.15 0.109 108 914923.5 0 7475.41% 13831396.75 4 conv2d_4_Conv2D 84.935 0.13 0.090 089 943954.7 3 7677.87% 14701487.72 5 dense_1_MatMul 9.437 4.52 1.101 106 8.6 5 0.7% 41084089.15 6 predictions_MatMul 0.003 0.00 0.012 013 0.3 2 0.0% 158145.08
Total Nodes In Avg:
All 498.209 6.10 1.880 884 265264.0 4 21.65% 32423235.46
=====================================================================================================
[Time] CpuArgmax<float> 1.3e-07 ms
[Time] CpuSoftmax 1.08e9e-06 07 ms[Time] RunCustomCNN 0.0108636 0110171 ms[Time] TopK 78.83e34e-06 ms1) red_apple : 0.999664999665
2) orange : 0.00033535
3) avocado hand : 18.85477e76131e-07084) hand avocado : 19.52249e23435e-08095) banana : 1.13961e38833e-12116) green_apple : 3.13808e44132e-1714
_______________________________________________________________
</pre>
VAITRACE JSON
<pre>
VAITRACE JSON CONFIG{ "options": { "runmode": "normal", "cmd": "./bin/customCNNclassification -w split -i ./images -c ./custom_images -r 10 -t 1 -M ./bin/dpu_custom_cnn_0.elf", "output": "./trace_customCNN_vart.xat", "timeout": 10 }, "trace": { "enable_trace_list": ["vitis-ai-library", "vart", "opencv", "custom"] }, "trace_custom": ["ListDirectory", "GetImageFileNames", "TopK", "CpuArgmax", "CpuSoftmax", "GetPredictedLabels", "GetGroundTruthLabels", "SliceVector"]}
</pre>
U0019
dave_user
207
edits
Retrieved from "https://wiki.dave.eu/index.php/Special:MobileDiff/10677"