• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 101 subscribers
  • Views 238 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

CCS/PROCESSOR-SDK-DRA8X-TDA4X: tidl model import failed

shuai wang32
Genius 4770 points

Part Number: PROCESSOR-SDK-DRA8X-TDA4X

Tool/software: Code Composer Studio

Hi :

             i    am   using     psdk_rtos_auto_j7_06_02_00_21   SDK

             i       import  my   cnn   use   tidl_model_import.out     have   some   problem  ,i  do  not  know   what  to  slove  this !

            follow   is   the   log!

~/psdk_rtos_auto_j7_06_02_00_21/tidl_j7_01_01_00_10/ti_dl/utils/tidlModelImport$ ./out/tidl_model_import.out  ../../../ti_dl/test/testvecs/config/import/public/caffe/tidl_import_landmark.txt
Caffe Network File : ../../test/testvecs/models/public/caffe/landmark/mouth_model_v6_rgb.prototxt  
Caffe Model File   : ../../test/testvecs/models/public/caffe/landmark/mouth_model_v6_rgb2.caffemodel  
TIDL Network File  : ../../test/testvecs/config/tidl_models/caffe/landmark/tidl_net_landmark.bin  
TIDL IO Info File  : ../../test/testvecs/config/tidl_models/caffe/landmark/tidl_io_landmark_  

Name of the Network : resnet_128_40_v6
IPLayer,=128
WARNING: PReLU Layer fc256_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!
IPLayer,=256
WARNING: PReLU Layer fc256_2_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!
IPLayer,=80

~~~~~Running TIDL in PC emulation mode to collect Activations range for each layer~~~~~

Processing config file #0 : /home/super/psdk_rtos_auto_j7_06_02_00_21/tidl_j7_01_01_00_10/ti_dl/utils/tidlModelImport/tempDir/qunat_stats_config.txt
 ----------------------- TIDL Process with REF_ONLY FLOW ------------------------

#    0 . .. T       3.06  ... .... .....
#    1 . .. T       2.83  ... .... .....
------------------ Network Compiler Traces -----------------------------
Main iteration numer: 0....

 Life time alive buffers are with ID: 10034 ( 0), 50033 ( 1), 60033 ( 1), 40002 ( 2), 20034 ( 0), 33 ( 2), 40001 ( 2),
Preparing for memory allocation : internal iteration number: 0
Mmeory overlap issue, dataId = 6, baseAddrconstSatisfied = -1
Main iteration numer: 1....

 Life time alive buffers are with ID: 10034 ( 0), 50033 ( 1), 60033 ( 1), 40002 ( 2), 20034 ( 0), 33 ( 2), 40001 ( 2),
Preparing for memory allocation : internal iteration number: 0
successful Memory allocation

-------------------- Network Compiler : Analysis Results are available --------------------

****************************************************
**                ALL MODEL CHECK PASSED          **
****************************************************

Thanks!

Shuai

  • 0
    TI__Mastermind 22085 points

    If you are referring to these warnings. You can ignore this, these are debug information


    IPLayer,=128
    WARNING: PReLU Layer fc256_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!
    IPLayer,=256
    WARNING: PReLU Layer fc256_2_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!
    IPLayer,=80

  • 0 in reply to kumar.desappan
    Genius 4770 points

    Hi :kumar 

          but    at  next  step    i   run   ./PC_dsp_test_dl_algo.out     use    the    imported   model    the  result   is   not  right!

          so   i  think     model     import    still  have   some   problem!

    Thanks!

    Shuai

         

  • 0 in reply to kumar.desappan
    Genius 4770 points

    Hi :kumar

         There are multiple prelus in the CNN, but the last two full connections have reported problems!

        int32_t dataSize = TIDLPCLayers.layerParams.batchNormParams.numChannels;     i    check  this   is  not  right    here  dataSize=1

    follow  is  the  log!

    arcsoft check:dataSize=36 ,TIDLPCLayers.bias.bufSize=36,TIDLPCLayers.bias.ptr=0x55695ba71f90 , TIDLPCLayers.weights.ptr=0x55695ba71ef0 ,TIDLPCLayers.weights.bufSize=36 , TIDLPCLayers.slope.ptr =0x55695ba7c480 ,TIDLPCLayers.slope.bufSize=36!

    arcsoft check:dataSize=36 ,TIDLPCLayers.bias.bufSize=36,TIDLPCLayers.bias.ptr=0x55695ba7c520 , TIDLPCLayers.weights.ptr=0x55695ba76480 ,TIDLPCLayers.weights.bufSize=36 , TIDLPCLayers.slope.ptr =0x55695ba76520 ,TIDLPCLayers.slope.bufSize=36!

    arcsoft check:dataSize=36 ,TIDLPCLayers.bias.bufSize=36,TIDLPCLayers.bias.ptr=0x55695ba3fe90 , TIDLPCLayers.weights.ptr=0x55695ba32a30 ,TIDLPCLayers.weights.bufSize=36 , TIDLPCLayers.slope.ptr =0x55695ba32cd0 ,TIDLPCLayers.slope.bufSize=36!

    IPLayer,=128

    arcsoft check:dataSize=1 ,TIDLPCLayers.bias.bufSize=128,TIDLPCLayers.bias.ptr=0x55695ba68530 , TIDLPCLayers.weights.ptr=0x55695ba71bf0 ,TIDLPCLayers.weights.bufSize=128 , TIDLPCLayers.slope.ptr =0x55695ba7c010 ,TIDLPCLayers.slope.bufSize=128!

    WARNING: PReLU Layer fc256_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!

    IPLayer,=256

    arcsoft check:dataSize=1 ,TIDLPCLayers.bias.bufSize=256,TIDLPCLayers.bias.ptr=0x55695ba4dc50 , TIDLPCLayers.weights.ptr=0x55695ba82fd0 ,TIDLPCLayers.weights.bufSize=256 , TIDLPCLayers.slope.ptr =0x55695ba5f580 ,TIDLPCLayers.slope.bufSize=256!

    WARNING: PReLU Layer fc256_2_relu's bias cannot be found(or not match) in coef file, Random bias will be generated! Only for evaluation usage! Results are all random!

    IPLayer,=80

    Thanks!

    Shuai

  • 0 in reply to shuai wang32
    Genius 4770 points

    Hi :kumar

             layer levle traces are generated in the tidl/test/trace folder.

            The inference software generates two files for each data ID

            caffe_tidl_infer_eye.txt_0000_00001_00064x00064.y

            caffe_tidl_infer_eye.txt_0000_00001_00064x00064_float.bin

           here    float.bin     is     input   of   each   layer?

           and    what   is   saved   in    .y?

    Thanks!

    Shuai

  • 0 in reply to shuai wang32
    Genius 4770 points

    HI：

              from   follow   table   numChannels ==1     so   It's actually 128, so it's a warning

     

    -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
      Num|TIDL Layer Name               |Group |#Ins  |#Outs |Inbuf Ids                       |Outbuf Id |In NCHW                             |Out NCHW                            |
    -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
        0|TIDL_DataLayer                |     0|    -1|     1|  x   x   x   x   x   x   x   x |  0       |       0        0        0        0 |       1        3      128      128 |
        1|TIDL_ConvolutionLayer         |     1|     1|     1|  0   x   x   x   x   x   x   x |  1       |       1        3      128      128 |       1        4       64       64 |
        2|TIDL_BatchNormLayer           |     1|     1|     1|  1   x   x   x   x   x   x   x |  2       |       1        4       64       64 |       1        4       64       64 |
        3|TIDL_ConvolutionLayer         |     1|     1|     1|  2   x   x   x   x   x   x   x |  3       |       1        4       64       64 |       1        8       32       32 |
        4|TIDL_BatchNormLayer           |     1|     1|     1|  3   x   x   x   x   x   x   x |  5       |       1        8       32       32 |       1        8       32       32 |
        5|TIDL_ConvolutionLayer         |     1|     1|     1|  5   x   x   x   x   x   x   x |  6       |       1        8       32       32 |       1        8       32       32 |
        6|TIDL_PoolingLayer             |     1|     1|     1|  2   x   x   x   x   x   x   x |  4       |       1        4       64       64 |       1        4       32       32 |
        7|TIDL_ConcatLayer              |     1|     2|     1|  4   6   x   x   x   x   x   x |  7       |       1        4       32       32 |       1       12       32       32 |
        8|TIDL_BatchNormLayer           |     1|     1|     1|  7   x   x   x   x   x   x   x |  8       |       1       12       32       32 |       1       12       32       32 |
        9|TIDL_ConvolutionLayer         |     1|     1|     1|  8   x   x   x   x   x   x   x |  9       |       1       12       32       32 |       1       12       16       16 |
       10|TIDL_BatchNormLayer           |     1|     1|     1|  9   x   x   x   x   x   x   x | 11       |       1       12       16       16 |       1       12       16       16 |
       11|TIDL_ConvolutionLayer         |     1|     1|     1| 11   x   x   x   x   x   x   x | 12       |       1       12       16       16 |       1       12       16       16 |
       12|TIDL_PoolingLayer             |     1|     1|     1|  8   x   x   x   x   x   x   x | 10       |       1       12       32       32 |       1       12       16       16 |
       13|TIDL_EltWiseLayer             |     1|     2|     1| 10  12   x   x   x   x   x   x | 13       |       1       12       16       16 |       1       12       16       16 |
       14|TIDL_BatchNormLayer           |     1|     1|     1| 13   x   x   x   x   x   x   x | 14       |       1       12       16       16 |       1       12       16       16 |
       15|TIDL_ConvolutionLayer         |     1|     1|     1| 14   x   x   x   x   x   x   x | 15       |       1       12       16       16 |       1       12        8        8 |
       16|TIDL_BatchNormLayer           |     1|     1|     1| 15   x   x   x   x   x   x   x | 17       |       1       12        8        8 |       1       12        8        8 |
       17|TIDL_ConvolutionLayer         |     1|     1|     1| 17   x   x   x   x   x   x   x | 18       |       1       12        8        8 |       1       24        8        8 |
       18|TIDL_PoolingLayer             |     1|     1|     1| 14   x   x   x   x   x   x   x | 16       |       1       12       16       16 |       1       12        8        8 |
       19|TIDL_ConcatLayer              |     1|     2|     1| 16  18   x   x   x   x   x   x | 19       |       1       12        8        8 |       1       36        8        8 |
       20|TIDL_BatchNormLayer           |     1|     1|     1| 19   x   x   x   x   x   x   x | 20       |       1       36        8        8 |       1       36        8        8 |
       21|TIDL_ConvolutionLayer         |     1|     1|     1| 20   x   x   x   x   x   x   x | 21       |       1       36        8        8 |       1       36        4        4 |
       22|TIDL_BatchNormLayer           |     1|     1|     1| 21   x   x   x   x   x   x   x | 23       |       1       36        4        4 |       1       36        4        4 |
       23|TIDL_ConvolutionLayer         |     1|     1|     1| 23   x   x   x   x   x   x   x | 24       |       1       36        4        4 |       1       36        4        4 |
       24|TIDL_PoolingLayer             |     1|     1|     1| 20   x   x   x   x   x   x   x | 22       |       1       36        8        8 |       1       36        4        4 |
       25|TIDL_EltWiseLayer             |     1|     2|     1| 22  24   x   x   x   x   x   x | 25       |       1       36        4        4 |       1       36        4        4 |
       26|TIDL_BatchNormLayer           |     1|     1|     1| 25   x   x   x   x   x   x   x | 26       |       1       36        4        4 |       1       36        4        4 |
       27|TIDL_ConvolutionLayer         |     1|     1|     1| 26   x   x   x   x   x   x   x | 27       |       1       36        4        4 |       1       72        2        2 |
       28|TIDL_FlattenLayer             |     1|     1|     1| 27   x   x   x   x   x   x   x | 28       |       1       72        2        2 |       1        1        1      288 |
       29|TIDL_InnerProductLayer        |     1|     1|     1| 28   x   x   x   x   x   x   x | 29       |       1        1        1      288 |       1        1        1      128 |
       30|TIDL_BatchNormLayer           |     1|     1|     1| 29   x   x   x   x   x   x   x | 30       |       1        1        1      128 |       1        1        1      128 |
       31|TIDL_InnerProductLayer        |     1|     1|     1| 30   x   x   x   x   x   x   x | 31       |       1        1        1      128 |       1        1        1      256 |
       32|TIDL_BatchNormLayer           |     1|     1|     1| 31   x   x   x   x   x   x   x | 32       |       1        1        1      256 |       1        1        1      256 |
       33|TIDL_InnerProductLayer        |     1|     1|     1| 32   x   x   x   x   x   x   x | 33       |       1        1        1      256 |       1        1        1       80 |
       34|TIDL_DataLayer                |     0|     1|    -1| 33   x   x   x   x   x   x   x |  0       |       1        1        1       80 |       0        0        0        0 |
    -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
    Thanks!

    Shuai