Ticket Name: CCS/TDA2EVM5777: Question related to projects

Query Text:
Part Number: TDA2EVM5777 Other Parts Discussed in Thread: TDA2 Tool/software: Code Composer Studio [9:49 pm, 05/04/2020] Reshma: Sir I run the semantic segmentation use case in tda2x board and I got this result [9:50 pm, 05/04/2020] Reshma: Sir for optimization what is the procedure sir please help me sir what is the meaning of this sentence sir U have to understand where on tda2 the DL network is running what operation is happening in each network layer and then check for the possibility of optimization of those low-level operations please tell me, sir, how to optimize this can work on finding the scope for optimization and define how will u optimize it and also please reply sir and help me sir thanks reshma m

Responses:
>> what is the meaning of this sentence sir >> U have to understand where on tda2 the DL network is running what operation is happening in each network layer and then check for the possibility of optimization of those low-level operations We are also not sure, could you please check with him/her only. Thanks, Praveen

C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\import>tidl_model_import.out.exe tidl_import_jseg21.txt Caffe Network File : C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\caffe_jacinto_models_caffe\trained\image_segmentation\cityscapes5_jsegnet21v2\sparse\deploy.prototxt Caffe Model File : C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\caffe_jacinto_models_caffe\trained\image_segmentation\cityscapes5_jsegnet21v2\sparse\cityscapes5_jsegnet21v2_iter_32000.caffemodel TIDL Network File : C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\tidl_models\tidl_net_jsegnet21v2.bin TIDL Model File : C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\tidl_models\tidl_param_jsegnet21v2.bin Name of the Network : jsegnet21v2_deploy Num Inputs : 1 Num of Layer Detected : 27 0, TIDL_DataLayer , data 0, -1 , 1 , x , x , x , x , x , x , x , x , 0 , 0 , 0 , 0 , 0 , 1 , 3 , 512 , 1024 , 0 , 1, TIDL_BatchNormLayer , data/bias 1, 1 , 1 , 0 , x , x , x , x , x , x , x , 1 , 1 , 3 , 512 , 1024 , 1 , 3 , 512 , 1024 , 1572864 , 2, TIDL_ConvolutionLayer , conv1a 1, 1 , 1 , 1 , x , x , x , x , x , x , x , 2 , 1 , 3 , 512 , 1024 , 1 , 32 , 256 , 512 , 314572800 , 3, TIDL_ConvolutionLayer , conv1b 1, 1 , 1 , 2 , x , x , x , x , x , x , x , 3 , 1 , 32 , 256 , 512 , 1 , 32 , 128 , 256 , 301989888 , 4, TIDL_ConvolutionLayer , res2a_branch2a 1, 1 , 1 , 3 , x , x , x , x , x , x , x , 4 , 1 , 32 , 128 , 256 , 1 , 64 , 128 , 256 , 603979776 , 5, TIDL_ConvolutionLayer , res2a_branch2b 1, 1 , 1 , 4 , x , x , x , x , x , x , x , 5 , 1 , 64 , 128 , 256 , 1 , 64 , 64 , 128 , 301989888 , 6, TIDL_ConvolutionLayer , res3a_branch2a 1, 1 , 1 , 5 , x , x , x , x , x , x , x , 6 , 1 , 64 , 64 , 128 , 1 , 128 , 64 , 128 , 603979776 , 7, TIDL_ConvolutionLayer , res3a_branch2b 1, 1 , 1 , 6 , x , x , x , x , x , x , x , 7 , 1 , 128 , 64 , 128 , 1 , 128 , 64 , 128 , 301989888 , 8, TIDL_PoolingLayer , pool3 1, 1 , 1 , 7 , x , x , x , x , x , x , x , 8 , 1 , 128 , 64 , 128 , 1 , 128 , 32 , 64 , 1048576 , 9, TIDL_ConvolutionLayer , res4a_branch2a 1, 1 , 1 , 8 , x , x , x , x , x , x , x , 9 , 1 , 128 , 32 , 64 , 1 , 256 , 32 , 64 , 603979776 , 10, TIDL_ConvolutionLayer , res4a_branch2b 1, 1 , 1 , 9 , x , x , x , x , x , x , x , 10 , 1 , 256 , 32 , 64 , 1 , 256 , 32 , 64 , 301989888 , 11, TIDL_PoolingLayer , pool4 1, 1 , 1 , 10 , x , x , x , x , x , x , x , 11 , 1 , 256 , 32 , 64 , 1 , 256 , 32 , 64 , 524288 , 12, TIDL_ConvolutionLayer , res5a_branch2a 1, 1 , 1 , 11 , x , x , x , x , x , x , x , 12 , 1 , 256 , 32 , 64 , 1 , 512 , 32 , 64 ,2415919104 , 13, TIDL_ConvolutionLayer , res5a_branch2b 1, 1 , 1 , 12 , x , x , x , x , x , x , x , 13 , 1 , 512 , 32 , 64 , 1 , 512 , 32 , 64 ,1207959552 , 14, TIDL_ConvolutionLayer , out5a 1, 1 , 1 , 13 , x , x , x , x , x , x , x , 14 , 1 , 512 , 32 , 64 , 1 , 64 , 32 , 64 , 301989888 , 15, TIDL_Deconv2DLayer , out5a_up2 1, 1 , 1 , 14 , x , x , x , x , x , x , x , 15 , 1 , 64 , 32 , 64 , 1 , 64 , 64 , 128 , 2097152 , 16, TIDL_ConvolutionLayer , out3a 1, 1 , 1 , 7 , x , x , x , x , x , x , x , 16 , 1 , 128 , 64 , 128 , 1 , 64 , 64 , 128 , 301989888 , 17, TIDL_EltWiseLayer , out3_out5_combined 1, 2 , 1 , 15 , 16 , x , x , x , x , x , x , 17 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 524288 , 18, TIDL_ConvolutionLayer , ctx_conv1 1, 1 , 1 , 17 , x , x , x , x , x , x , x , 18 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 301989888 , 19, TIDL_ConvolutionLayer , ctx_conv2 1, 1 , 1 , 18 , x , x , x , x , x , x , x , 19 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 301989888 , 20, TIDL_ConvolutionLayer , ctx_conv3 1, 1 , 1 , 19 , x , x , x , x , x , x , x , 20 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 301989888 , 21, TIDL_ConvolutionLayer , ctx_conv4 1, 1 , 1 , 20 , x , x , x , x , x , x , x , 21 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 301989888 , 22, TIDL_ConvolutionLayer , ctx_final 1, 1 , 1 , 21 , x , x , x , x , x , x , x , 22 , 1 , 64 , 64 , 128 , 1 , 8 , 64 , 128 , 37748736 , 23, TIDL_Deconv2DLayer , out_deconv_final_up2 1, 1 , 1 , 22 , x , x , x , x , x , x , x , 23 , 1 , 8 , 64 , 128 , 1 , 8 , 128 , 256 , 1048576 , 24, TIDL_Deconv2DLayer , out_deconv_final_up4 1, 1 , 1 , 23 , x , x , x , x , x , x , x , 24 , 1 , 8 , 128 , 256 , 1 , 8 , 256 , 512 , 4194304 , 25, TIDL_Deconv2DLayer , out_deconv_final_up8 1, 1 , 1 , 24 , x , x , x , x , x , x , x , 25 , 1 , 8 , 256 , 512 , 1 , 8 , 512 , 1024 , 16777216 , 26, TIDL_ArgMaxLayer , argMaxOut 1, 1 , 1 , 25 , x , x , x , x , x , x , x , 26 , 1 , 8 , 512 , 1024 , 1 , 1 , 512 , 1024 , 8388608 , Total Giga Macs : 8.8442 1 file(s) copied. Processing config file .\tempDir\qunat_stats_config.txt ! 0, TIDL_DataLayer , 0, -1 , 1 , x , x , x , x , x , x , x , x , 0 , 0 , 0 , 0 , 0 , 1 , 3 , 512 , 1024 , 1, TIDL_BatchNormLayer , 1, 1 , 1 , 0 , x , x , x , x , x , x , x , 1 , 1 , 3 , 512 , 1024 , 1 , 3 , 512 , 1024 , 2, TIDL_ConvolutionLayer , 1, 1 , 1 , 1 , x , x , x , x , x , x , x , 2 , 1 , 3 , 512 , 1024 , 1 , 32 , 256 , 512 , 3, TIDL_ConvolutionLayer , 1, 1 , 1 , 2 , x , x , x , x , x , x , x , 3 , 1 , 32 , 256 , 512 , 1 , 32 , 128 , 256 , 4, TIDL_ConvolutionLayer , 1, 1 , 1 , 3 , x , x , x , x , x , x , x , 4 , 1 , 32 , 128 , 256 , 1 , 64 , 128 , 256 , 5, TIDL_ConvolutionLayer , 1, 1 , 1 , 4 , x , x , x , x , x , x , x , 5 , 1 , 64 , 128 , 256 , 1 , 64 , 64 , 128 , 6, TIDL_ConvolutionLayer , 1, 1 , 1 , 5 , x , x , x , x , x , x , x , 6 , 1 , 64 , 64 , 128 , 1 , 128 , 64 , 128 , 7, TIDL_ConvolutionLayer , 1, 1 , 1 , 6 , x , x , x , x , x , x , x , 7 , 1 , 128 , 64 , 128 , 1 , 128 , 64 , 128 , 8, TIDL_PoolingLayer , 1, 1 , 1 , 7 , x , x , x , x , x , x , x , 8 , 1 , 128 , 64 , 128 , 1 , 128 , 32 , 64 , 9, TIDL_ConvolutionLayer , 1, 1 , 1 , 8 , x , x , x , x , x , x , x , 9 , 1 , 128 , 32 , 64 , 1 , 256 , 32 , 64 , 10, TIDL_ConvolutionLayer , 1, 1 , 1 , 9 , x , x , x , x , x , x , x , 10 , 1 , 256 , 32 , 64 , 1 , 256 , 32 , 64 , 11, TIDL_PoolingLayer , 1, 1 , 1 , 10 , x , x , x , x , x , x , x , 11 , 1 , 256 , 32 , 64 , 1 , 256 , 32 , 64 , 12, TIDL_ConvolutionLayer , 1, 1 , 1 , 11 , x , x , x , x , x , x , x , 12 , 1 , 256 , 32 , 64 , 1 , 512 , 32 , 64 , 13, TIDL_ConvolutionLayer , 1, 1 , 1 , 12 , x , x , x , x , x , x , x , 13 , 1 , 512 , 32 , 64 , 1 , 512 , 32 , 64 , 14, TIDL_ConvolutionLayer , 1, 1 , 1 , 13 , x , x , x , x , x , x , x , 14 , 1 , 512 , 32 , 64 , 1 , 64 , 32 , 64 , 15, TIDL_Deconv2DLayer , 1, 1 , 1 , 14 , x , x , x , x , x , x , x , 15 , 1 , 64 , 32 , 64 , 1 , 64 , 64 , 128 , 16, TIDL_ConvolutionLayer , 1, 1 , 1 , 7 , x , x , x , x , x , x , x , 16 , 1 , 128 , 64 , 128 , 1 , 64 , 64 , 128 , 17, TIDL_EltWiseLayer , 1, 2 , 1 , 15 , 16 , x , x , x , x , x , x , 17 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 18, TIDL_ConvolutionLayer , 1, 1 , 1 , 17 , x , x , x , x , x , x , x , 18 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 19, TIDL_ConvolutionLayer , 1, 1 , 1 , 18 , x , x , x , x , x , x , x , 19 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 20, TIDL_ConvolutionLayer , 1, 1 , 1 , 19 , x , x , x , x , x , x , x , 20 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 21, TIDL_ConvolutionLayer , 1, 1 , 1 , 20 , x , x , x , x , x , x , x , 21 , 1 , 64 , 64 , 128 , 1 , 64 , 64 , 128 , 22, TIDL_ConvolutionLayer , 1, 1 , 1 , 21 , x , x , x , x , x , x , x , 22 , 1 , 64 , 64 , 128 , 1 , 8 , 64 , 128 , 23, TIDL_Deconv2DLayer , 1, 1 , 1 , 22 , x , x , x , x , x , x , x , 23 , 1 , 8 , 64 , 128 , 1 , 8 , 128 , 256 , 24, TIDL_Deconv2DLayer , 1, 1 , 1 , 23 , x , x , x , x , x , x , x , 24 , 1 , 8 , 128 , 256 , 1 , 8 , 256 , 512 , 25, TIDL_Deconv2DLayer , 1, 1 , 1 , 24 , x , x , x , x , x , x , x , 25 , 1 , 8 , 256 , 512 , 1 , 8 , 512 , 1024 , 26, TIDL_ArgMaxLayer , 1, 1 , 1 , 25 , x , x , x , x , x , x , x , 26 , 1 , 8 , 512 , 1024 , 1 , 1 , 512 , 1024 , 27, TIDL_DataLayer , 0, 1 , -1 , 26 , x , x , x , x , x , x , x , 0 , 1 , 1 , 512 , 1024 , 0 , 0 , 0 , 0 , Layer ID ,inBlkWidth ,inBlkHeight ,inBlkPitch ,outBlkWidth ,outBlkHeight,outBlkPitch ,numInChs ,numOutChs ,numProcInChs,numLclInChs ,numLclOutChs,numProcItrs ,numAccItrs ,numHorBlock ,numVerBlock ,inBlkChPitch,outBlkChPitc,alignOrNot 2 72 72 72 32 32 32 3 32 3 1 8 1 3 16 8 5184 1024 1 3 40 34 40 32 32 32 8 8 8 4 8 1 2 16 8 1360 1024 1 4 40 34 40 32 32 32 32 64 32 6 8 1 6 8 4 1360 1024 1 5 40 34 40 32 32 32 16 16 16 6 8 1 3 8 4 1360 1024 1 6 40 34 40 32 32 32 64 128 64 6 8 1 11 4 2 1360 1024 1 7 40 34 40 32 32 32 32 32 32 6 8 1 6 4 2 1360 1024 1 9 40 34 40 32 32 32 128 256 128 6 8 1 22 2 1 1360 1024 1 10 40 34 40 32 32 32 64 64 64 6 8 1 11 2 1 1360 1024 1 12 40 20 40 32 16 32 256 512 256 8 8 1 32 2 2 800 512 1 13 40 36 40 32 32 32 128 128 128 5 8 1 26 2 1 1440 1024 1 14 40 24 40 32 16 32 256 32 256 8 8 1 32 2 2 960 512 1 16 40 34 40 32 32 32 64 32 64 6 8 1 11 4 2 1360 1024 1 18 40 34 40 32 32 32 64 64 64 6 8 1 11 4 2 1360 1024 1 19 40 40 40 32 32 32 64 64 64 5 8 1 13 4 2 1600 1024 1 20 40 40 40 32 32 32 64 64 64 5 8 1 13 4 2 1600 1024 1 21 40 40 40 32 32 32 64 64 64 5 8 1 13 4 2 1600 1024 1 22 40 34 40 32 32 32 64 8 64 6 8 1 11 4 2 1360 1024 1 Processing Frame Number : 0 Layer 1 : Out Q : 254 , TIDL_BatchNormLayer , PASSED #MMACs = 1.57, 1.57, Sparsity : 0.00 Layer 2 : Out Q : 6097 , TIDL_ConvolutionLayer, PASSED #MMACs = 314.57, 209.19, Sparsity : 33.50 Layer 3 : Out Q : 5378 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 122.16, Sparsity : 59.55 Layer 4 : Out Q : 13609 , TIDL_ConvolutionLayer, PASSED #MMACs = 603.98, 162.66, Sparsity : 73.07 Layer 5 : Out Q : 8489 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 123.86, Sparsity : 58.98 Layer 6 : Out Q : 9235 , TIDL_ConvolutionLayer, PASSED #MMACs = 603.98, 153.91, Sparsity : 74.52 Layer 7 : Out Q : 11745 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 103.55, Sparsity : 65.71 Layer 8 :TIDL_PoolingLayer, PASSED #MMACs = 0.26, 0.26, Sparsity : 0.00 Layer 9 : Out Q : 16199 , TIDL_ConvolutionLayer, PASSED #MMACs = 603.98, 142.94, Sparsity : 76.33 Layer 10 : Out Q : 14923 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 76.61, Sparsity : 74.63 Layer 11 :TIDL_PoolingLayer, PASSED #MMACs = 0.52, 0.52, Sparsity : 0.00 Layer 12 : Out Q : 25787 , TIDL_ConvolutionLayer, PASSED #MMACs = 2415.92, 500.01, Sparsity : 79.30 Layer 13 : Out Q : 6224 , TIDL_ConvolutionLayer, PASSED #MMACs = 1207.96, 221.44, Sparsity : 81.67 Layer 14 : Out Q : 10770 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 79.45, Sparsity : 73.69 Layer 15 : Out Q : 6165 , TIDL_Deconv2DLayer, PASSED #MMACs = 0.52, 0.52, Sparsity : 0.00 Layer 16 : Out Q : 10642 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 96.17, Sparsity : 68.15 Layer 17 : Out Q : 4658 , TIDL_EltWiseLayer, PASSED #MMACs = 1.05, 1.05, Sparsity : 0.00 Layer 18 : Out Q : 13159 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 131.50, Sparsity : 56.46 Layer 19 : Out Q : 15034 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 125.67, Sparsity : 58.39 Layer 20 : Out Q : 16267 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 128.61, Sparsity : 57.41 Layer 21 : Out Q : 11247 , TIDL_ConvolutionLayer, PASSED #MMACs = 301.99, 122.65, Sparsity : 59.39 Layer 22 : Out Q : 3107 , TIDL_ConvolutionLayer, PASSED #MMACs = 37.75, 14.84, Sparsity : 60.68 Layer 23 : Out Q : 1583 , TIDL_Deconv2DLayer, PASSED #MMACs = 0.26, 0.26, Sparsity : 0.00 Layer 24 : Out Q : 1600 , TIDL_Deconv2DLayer, PASSED #MMACs = 1.05, 1.05, Sparsity : 0.00 Layer 25 : Out Q : 1604 , TIDL_Deconv2DLayer, PASSED #MMACs = 4.19, 4.19, Sparsity : 0.00 Layer 26 :TIDL_ArgMaxLayer, PASSED #MMACs = 4.19, 4.19, Sparsity : 0.00 End of config list found ! C:\PROCESSOR_SDK_VISION_03_07_00_00\ti_components\algorithms\REL.TIDL.01.01.03.00\modules\ti_dl\test\testvecs\config\import> same thnk i asked please tell me sir Optimization I have told u many times U have to understand where on tda2 the DL network is running what operation are happening in each network layer and then check for possibility of optimization of those low level operation

Caffe based model convolution layer output optimization issue on board when I am running the use case on the tda2x board please reply sir thanks Reshma

sir please reply sir its very urgent sir

>> Caffe based model convolution layer output optimization issue on board when I am running the use case on the tda2x board please reply sir What optimization issue you are using facing ? could you please explain more about the issue ? Thanks, Praveen

sir actually i am not understanding what I need to do please tell me sir how to optimize the network thanks Reshma M

Hi Reshma, I am also really not understanding what you need to do here, because as I told you couple of times that this network is already fully optimized and no further optimization possible here. So, please understand your requirement correct first and try to come out with proper questions and don't ask same questions again and again. We are happy to help but not for the questions you also don't have clarity. So, first get the clarity on this from the person who assigned this to you or who asked to do this optimization. Thanks Praveen