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.

TDA4VM: Edge AI : Unsupported version for slice layer

Part Number: TDA4VM

Tool/software:

I am trying to run our custom model which has slice layer with the following details given in the image ,

Edgeai : 927

Model opset _ir version : 18 _ 8 

when i try to compile the model i get the below message with error 


Slice layer : Unsupported onnxOpSetVersion 18 -- Slice_nf_0

INFORMATION: [TIDL_ResizeLayer] Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
INFORMATION: [TIDL_ResizeLayer] Any resize ratio which is power of 2 and greater than 4 will be placed by combination of 4x4 resize layer and 2x2 resize layer. For example a 8x8 resize will be replaced by 4x4 resize followed by 2x2 resize.
****************************************************
** 2 WARNINGS 0 ERRORS **
****************************************************
In TIDL_runtimesPostProcessNet 4
************ in TIDL_subgraphRtCreate ************
The soft limit is 2048
The hard limit is 2048
MEM: Init ... !!!
MEM: Init ... Done !!!
0.0s: VX_ZONE_INIT:Enabled
0.15s: VX_ZONE_ERROR:Enabled
0.38s: VX_ZONE_WARNING:Enabled
0.4168s: VX_ZONE_INIT:[tivxInit:190] Initialization Done !!!
0.56525s: VX_ZONE_ERROR:[ownTensorCheckSizes:119] Invalid view parameter(s) in dimension: 0
0.56556s: VX_ZONE_ERROR:[tivxUnmapTensorPatch:758] map address is null
0.56563s: VX_ZONE_ERROR:[tivxUnmapTensorPatch:762] map size is equal to 0
0.101492s: VX_ZONE_ERROR:[tivxMemBufferAlloc:130] size is 0
0.101539s: VX_ZONE_ERROR:[ownAllocReferenceBufferGeneric:349] Memory allocation failed
0.101550s: VX_ZONE_ERROR:[ownGraphAllocateDataObject:1003] Memory allocation for data reference failed
0.101561s: VX_ZONE_ERROR:[vxVerifyGraph:2104] Memory alloc for data objects failed
0.101571s: VX_ZONE_ERROR:[vxVerifyGraph:2213] Graph verify failed
TIDL_RT_OVX: ERROR: Verifying TIDL graph ... Failed !!!
TIDL_RT_OVX: ERROR: Verify OpenVX graph failed

and the model doenot complie , can you let us knw the solution for this

  • Hi,

    1) Suggested to always use latest tools (in case of 9.2 sdk its 9.2.9.0) the allowlisting mechanism is matured in latest tools, also we have done 10.0 tools public release, can you share the logs on latest tools 10.0

    Here are the limitation on the slice layers from TIDL side, please do diligence for the same

    13 Slice/Split TIDL_SliceLayer
    • Must have 4 inputs into the operator, where one is the variable (Must be <=4 dimensions) input while the other 3 are constant/initializers & 1D
    • Only batch size = 1 is supported
    • Non-one stride is not supported individually (Only supported in Patch Merging)

    Also please share attach the toy model (not the complete but sharded one where issues is visible)

  • Hello , 

    Thanks for response  please find the below logs for latest tools 10.0 

    due to the production development from customer we pefer to execute in the environment of  edgeai 9.2.7 and tidltools 9.2.0 can you please provide alternate solutions for this 

    PKڄY1W log_topkonly_fp16_c.txtUT
    E�fl�fj�fux���\{s�8�ߟ����j7l �uB����dw����T1�ז�05�Z�cH6�����JhI-�[���/�"����Q�X�z�vPkU'lJ��Wɣ3
    }WYq;��	�,���,����x6g��Y%���k\��0���4D.`#�2���BN�����{����'@��p�#vO=���ys�={�I+����Ȋ6����jU��a���0I�L��İ��,�|' �>��+gD���:ޘ�=�p΢���`N�m�+�ۅ`�	��		�`D"�է5�Q���ԫ�b[��*�jK�
    ��9�\q'��>!�
    8�ˆ�O0����qJ�ۡ�'��9�J�`G�#��=b>jr{G�JU}�k��;�VwQY5��1��!���!�0<2LƶO�Q��<'�#H��.N���	'[SCI�3��|���Ylߙ���)y"��[Y�g�
    �,�֊�]e�ø�O��\}��I丳
    �?��F��,�[bFJg���^�4��Rs����o���G�ٸ"\�za�,º}�R�O���%�M=�ò�=���ȶ�?'��g�B,0.{�����=t�м�Zđ�lkO�xbG$f~2�iިJ���C����ES�&��d�"�<�s�c�
    �'�YXh�R�,�a��qx�����1�v��˘�����������Q.�a������ν���b
    (T]/�сa$#mu+JuؐD�hI��6B� �C�}g��tPC�P�=Ƿi�z�H���f	9��h�2�6�$4��MܨX�	��#�-l�2��f���CH���6hv�cdP�Θ�-ì@߉�DV~�.3�l�+�>������5� S͠�B�˂�o��D�S2�:Ѭ����c|]L�HG�ZF�T���uy�̚]�O.�����F�'"k��g��|�`��ֲ.݊|f�lTj��ʃ7_K>�&.Ø�m����z���g�.�;-��i��k�&�!1��(z:������{��EXkJ��VV����_5��'���,�	������پV�o����z#WTW�n>�z��o-x3%�ւ��1�����H�ݗg�XV�$�b�N2vd�6:�ݐ��H���nVD5����Vv=]�,�M�tp��‡�Շ~�rp�9k�;!/�$K'�h�Rg�u�t��h���$�}@�2�SE�ecc��y���ZFϊ�{Ь�o7��*T��A�+��dE$Yf#��](���B��I��[��t�mY?�\������0�]�FE��s�a�#{iǘ��m�}=��XE��\䋚�+��l�E`9�ɋ%�il�1Q>���I��k���>��2T��'�uE�q�սٳ��񄏅�W�߷��珗"C:��U���~�e�Ǩ�c�*N0��=���\�۟E_�= ��x�?���8��g4�J����^�����{������c(NKY��!�d{^b2r#�Le8Q�<	�4�i���K|OdЂ\Ӗ}a�m��2��c�)?͚��L��v�^��3��v{��Z]Z�K����t���@ari����ZS�^Ja�Vv[T'�����G�\��X�u��Fˈ)>)�����˗�q�'�P��du��!
    ���눭W`Kbn�Vq�N�5���v���9Y�E��̩͞���FZ(ɢ~�� ��ܥ������^M�n\��X�w�, ��`/�sť�\wEz���x���=�s��nb@�Hp�ЙSi‚4F�s'��p<���L����A��z7�А�?�-%|�*.t�K���IF�gF���tU��
    �HdLKdZ����^V*9��Y)�6%)��6�Y/+U�Etq8Y��>�N�js�F����VU6N��?K���l�f�rI�4kV�f��y����HT�Tয়~Z)9�`Q�Ϗ[?����u�n�{��0��4���`p9X�>�V�+/Ӡn6͵�qr�(di�ǵ[)����lK:�W�'ڮ�je.7㒻��y�A��MW7�gw�Ɍ�n���s��D[ͦ���4����`3���,}|ܬ7[�yl�'�)Z�5��6����F;|�hG���m�h�+�0��e�7���鞏&��H!��n�\��{w$
    ��3] ���PӲ��IZ�M�>
    ɜ]p�����
    �M�,��N�6#�Q��3"��2/��[��wi'!FrL��#:+�ӢO媇O�Op/�w<E���
    	i�x��@r�����i�"�9u�,��]cH�-�尒���2M^�Vک��ɥ�v�vri��f��\Zi��N.��sXj'�V�9*��K+���ɥ�v�o=��4�^�1K���J;e6�O0�wv�%�S!���������h5���?@΁�|�B�'��ڈ��%�;�:ѝ������YD毁�I��z��c��|U��|U��|U��,�����t	��%�SD�PO]B=Et	��%�SD�PO]B=Et	��%�SD����C�f�>�^?��>%���縄zV�J�g���zJ���z���)�K���.��"��z���)�K���.��"��z���)�K��������e6�P�u</~*Գ�<��4�ֳG�h�>s�D\�	�6�}�r�V�z��4��S<?�h��~�s��ϻ���-p�!�_����7&T�~��9}�c�i�E䬿��y������/��nV.*��N#��Ҕ�:M��iå9���O��"�����r
    u�#��,�8��(h�����Lu��t�h{g�Y'����w[���gg0�3�"�g�ק����_W/{:ǚw�N��kg�P_�V�Ɗ+����6�O "<�p��pд�2���nТ�իwN��K^I�?PK��g�;1WPK��Y topkonly_fp16/UT
    ��f A�fc�fux��PK��Y topkonly_fp16/tempDir/UT
    ��f A�fM�fux��PK��Yd& topkonly_fp16/tempDir/graphvizInfo.txtUT
    ��fM�fM�fux����Ak�0��}�:�hۭ�d���"!3�a�`��~֮TRF���\��^��y�����8�[gZ���ͅ�?R�{��\:����J�撽&�a�.�|�����4�f�j�k։\g(������	/�r
    aB}
    �D3�ӷ�������{'yM�R��K��c�ҍx�����2��ײT(:�>C��0L��4{�=�5$�`��wi$.r��?��F��hd¾1#(!�R�:Vno�����y<�8�$d<Z�� ��o�䯡�4�PK= �dPK��Y^60 topkonly_fp16/tempDir/runtimes_visualization.svgUT
    ��fM�fM�fux���Zko�8��_���)S�C���`�I�`;i��3X��V��d�ʫ�~�%)Y�u���8;iaJ⽗����7?<���.�β�8t���r��C���^�Y��$/���)J燷{o�q����c2���O?~8��线���]����F���:{Ĺ��Ɂ�����{A���}?M&��p�.�	�\��U#}�z�}Z�ӤJG��G2�*wٗ:^©�#�{��H�g�\fU�J]�����?����dZeWɰ���,FyR����ip�\���iy<�����]9��3�^�7��l�No�*�Ig��lv��ٗ��T�=��´<s���k�JƃI��4_WP9��]���X>:�<�~�Vu�``H���>��a�[����FDZ��bG�I6:tT��a��f��j���rzs�̆I�~�{E�e�����X�AD�R�
    mV��o��N~�jo{o&e�8������й�Ϊ�sZ~��T�N�iZ@�Oʬ���={�I*�B=��8���q�>淳�5}���t�~sb�BL�k_3f�w"��(�O�L�p��������a_ �c����	a�q���c$t�(�vZG�90���)ͯ�[�B�
    q�0UG�c3�N�����<���j��W%����_��翑�Ogg�/��ȇ��0�')��N&����˓�D%����x��Y�PN�����)�)�{뢊�e��Ze�|P��yur��r���d�O|.�3X�{&�Î����z;�����p�S:9C�!d�7�I����UQ�'�$ԲPA-�P����r��(O
    a�k+
    )�]�#����X��V��҇��O/)����й�F��6���;�h&�rUU�*��r8s�\�?K�ٕ�2˾@�2�|���+��ej4al�ϒYu�ή�I:����f��J�V[�,�~f�t[�2�b�m��7�p��X��"`u�ÜE$C�c1��.�cE���2tP�ռ@�X��Rt���X��Vn0���]�o5�āΕ�kg�\�;�^����&�d�(��r��5hH���d���
    5���K��A�D!��8"P�*��j?�"R��}�X�~0��"d!��}*b(D4O���O�U�tc���
    �:���HF�~Z�����L��Rz�5�o��w��OXߙ���9���F6w`#���hh
    �\���PI����څ�FBw`#�Y��i
    ��pl�C�U=o��~&�GW�u��+0��j��Mu���)�.ka=���^#�[�q��F]#4�چZV[�Hlcs�‚&�^-�Y87�]�P�cA�1��7�(n���޵`�x��ie+n����{��j\�-�����Zrc����RЈn,��3�n&��x-��m@<�J�qQ�ht�L����n=� ���V���?�V��
    �@ثj�bU��X�6�K��=�(����ׇ�z%A$��QE��>jj�
    �	Ř���Q;n%�)#S�2�ڟ�T\��ڴi��p��F����=m�ky�"O�Y��5�����Wh�D��H�y��*���+�X@��A�M�A�>ѰS�X��
    :�6�E_�3�:c~\����j�@��4�xM*F���	�|�kj3��@G���Ƃ���H��"�܆:PmG��4����Dt;(kg���.ܔ�^F�A��5�i��NA�u>�)k��<���N�Q<V{�롹7h�^�z��^��8���ө�z`��zJ&��UiPǰyxbaG�_�����\�Ą�[��0�{�<��1�N��XP�|"T_�0$m
    =$�E(������6���������D��-�)���%Xų�u�IlT�����f/�yr���;��^���l6ڛ;�^�^.������_��|+#,�1���#1R���"(�Չ"
    S��n��s���������AH�?���VJ�V�u�N�P�f���/k�_�˻��j��#�]��.<_�\�>�~�k�A��_r(�aܣ�?#1|"�59(��I�"DM&��j�Y_��Ï�D�i��V�KE'�~n�l���M��l��
    P�~7"؂FK#lE��a���׈�T{ɍ1.]�ij�����([mPc�g	�J���*�����T�`[�b5����[�/v��l��W��k��L���1.�u��ߵ��ѐ=�[6e���+�6�ƒ����l-4�r�
    ӹt�	����PK-ŗ�^6PK��YF8 topkonly_fp16/tempDir/subgraph_0_tidl_net.bin_netLog.txtUT
    ��fM�fM�fux��՘]k�0����v4�֏���	�`
    ��L�&X+��
    ���X�M͗���h���=pt8�Ρ��:�'S�̠�,(U8J��	ǽN��;��:�Ǽ�v7n����[yѬ)x�V��z3O�+Z��ka�~��=�jU8xn��G�,cwYz�eӵs��k�Q�ba�\A����-0�d�98lC�?@�q�N�V�݋	b�cb�LF�Ŕ��>!���q��!pi&�bB�/eb2��;g�&��'mw�%�%���9XDL�9�L���]��}B�;&5�OЕ|BS�2�1���R�Ą�BĄ�H�X�@���21k{&�F툒��	�����Ǥ3y'�7^s�'��yI�hԑ��
    &uF�L�䝤v�=���-^js1��,RL�@qL������K�
    j)�&#�;��L��=&�O�4�)�2i��	���E1AZL���BĄ�@�$��ޙɡv����„��Hٳ]�'9���c��#d�
    �O���Gy��[����K��|��c��O����(�]?�����O=h��dD��Y�PK�쭜&FPK��Y�{a- topkonly_fp16/tempDir/subgraph_0_tidl_net.binUT
    ��fM�fM�fux����;nQ��ď�q����5P�@����]�*`l'Q�R��0�J||GN��m��IG�<�;�~�W���L�q���4���W1�kf����v�{��0�bY~����9_~O���,�����������͛�F)�j43,���I��]���*'���������l3�5���X~��D���o��yR[����]l��,��)��p���*�b9Sj:���|<{v��S��o��]�?�ߛ����3oC�[r�ߏٴ~t��ln�������q�?�ΊA�~u��+�r>�+���վ�GȞ�b�*�7�:��+{���#��|\�w�E���N>�_ ����������4�W>��|<���8��?�]Y������,��鏣�l��yn��1w�����UrW^�4��u���������t����?��	������|PKgߐ��{aPK��Y�q. topkonly_fp16/tempDir/subgraph_0_tidl_io_1.binUT
    ��fM�fM�fux����A
    �P���],p�Q$�V�Fh�ѓt]`��>��݌�W�R�9�����Nk�D?������/��Gn��(���D�S�$F�\��#�:���b���"E��B,�?7����s������<��_f�9��&3�}�n�0�����3�`�n	��kȩZCNPK�w���qPK��Y< topkonly_fp16/tempDir/subgraph_0_tidl_net.bin.layer_info.txtUT
    ��fM�fM�fux��m�I
    �0E�9�;BF��5��r��e'h����)�&h����Ho���&�ϲNݎr��K���(�,��:��f����2��d}t��HK�� �H�$E�1&)y��J����PK�Q�uPK��Y�~1 topkonly_fp16/tempDir/subgraph_0_tidl_net.bin.svgUT
    ��fM�fM�fux���]�o�����\*U[)ό�m�j�������ҭ�
    NbL.8M�����&8��r6�3c�9�c�s���L��[4��Ӥ��]����t:���~���վ�A�4LF�x�D�N2����ы��st��������߼��~���z������wq�w���u.���Y�w}}ݽ������,���O�=��^�J��k
    ��(u�>���Q��4�/��y��o�?j��tmڵ�/��Gh_�e����v{��d:��O���p��g�i:��N��h&Qz�����Cx���MG��ɻ���t��o6�4�\ij������:I�������%N�Qx͟!,:�.�u<J/�u��L;�"��/�~�:.3�oqt���߱����{)�:������g�#��k.�,+�@�%�n�q��8�?�]�s��>|���8�ϥ�A�,L�g�٤ߙ������S4��슳�*��3�v}Dmܵ�fm��%~yW�EOt����t�����x<�w�/�4�>���W����gQ®��4NRv��l�{��ڈ�}>|��u�F��uz�s�&�͐�
    _�99�P<��W��c���L�G��c�Z���5����q4a����N~�و()�pr5N��t�eׯ�{y�~x���������/g'���gg�(}�ӑx������Mt�ߘ\Ma�D�y�B#v"��[ŗ�[v����Y���#{�������t�;���e�!�x��{����a|�6NzoÛ�/�|vO���&�_��W��[�;׃x�w�(����xO����푸1�Y|�ϓh�i:-��x���}���F_�����|�
    ���I��gt3�a�g���z��6��z�����Z��	�����p|���Ǧ�|j%��LP7�0U{�n��&�����7�M��y��5w���%l&a��2+wW0�a>�|���l��0����gey�,:� 6���n{,�o�CDIקH��+_@\N	�6,v��`��z��׼a����Z	�%b4��Q�!Wk�R
    ��/
    g7it���a?LN/��~g�FcF�3u���A�����i��x̞Ɠh�7�f�<2g�����Q����y��Xt�h�gvKYo���y�\�?��Ө⍣���I�>�<�N���k�{��m����M<�;H������V;��ɭ}{˨@�����qt4�N_����밾���ɿ�����6R<�*�U���WiG4];p�Q3�$>�Pq��|�hnI��kn�-K��sˢ�z�$���+B5�|0��'��!:�5ܜ[��� �ى�e�I<|'��=:ƈ,d�����p�ڿ-�x+��s���y�H�8 q[�8��ę_
    �ѹ���Y���u^h���f��:q��G�@�g-'�&F�Зqx�U��2Q��zbe'[��T8��e�n� a���:�2|��]��Q�K*��ԥJRR��.���]O��uWN��f��1�{�.��ec�JM`镚�v�[����R�K�o�Y�:��2�l�hnK�~�wn���Z��Fe8��0�aRo��n�nE�������4�"�:�
    ��Q�O_��D����J��-?ﷄ�8+���+R��*�z�Z2A��ie��,�,�V�IfTk��X��5|�`�|�fi�e?��ۣ�\r�u08��VN^�;�8,�>N�)O�R��7Ów�??_Ž.}M���ך�VG��~1ǭ��S3�f���n����?tߴ7�m�]H{mKet
    �����"�:EۖΗ��'R�`�zU�"YK���y_%��V�T������8��fU,RӪX$��69��d%z{��/#H�c��B�+?ܧÊ��e��$.\��向�2`?���I
    ,X�p8[�pƗDt��x��d��֖����a8p�D4}�g��}�Ғ��f:Z�G�JB!��3Ҝe��bx���Ã׿
    !P�
    �u�pv9G�s3��8����e�m�A�
    �7nh�Ol�Bޣ���a"��"�Lɐ�����B�H&!ɴ�2�|o
    �
    �t�%��Q���%3��nOb�K��`>z�fv�������֧t�ȁ�ՠl��ѡl�U,�Ͳ��")�M��@7�nn\a�V`/�^��{���˖��f/k�\��4�\�8����c����z7����Hֺ�����"�#�
    ln.R8$z2��c�/��젽��.Z�SI���$㑘�lt״���w���&�,�N�.n�
    �p+K�J��P��d:���n&ULs-V싢Kec?��%��z����p-U�[�z�L�V+�E��j�5�S)Yb�,��]%;w�Z�V~zGQV��E6l���|��? Δ�B��?]�?ݬ��B~Zfi�0<x��1�c���Y��sw����[��*K��ڲ��۵ɴ�A����[���ްY|[:�d�B�m�ꑮ"ؚŷw�lߺKi����ȳ�a��.�N�)��^��n��!pC�P|���e�p``��o�}��2�������~�EL�~������s�JX����>�{�M�E_5���
    �8�&;G㚬$�b���"�A�B	V2`%c�+;J+�@B��	�q��iT~��}Yݷ}�9���x��/@_�[�Ł)�bj
    ��k
    ���^�7SzM�7k_��8�L��p���o�S�n#�)�Z�Jt="����e�\Σh4���#�?�gvZ{fo�gv3�����k����ߨ��ć�����������1�+����A!���C�Ѵ��
    �"�LaH	��"f�!%� W(I.P�P
    ��\���ٞ�dW��`�%NK5�N7�4��3|WUSzdt�"o��	�m�W{�n��oz	h�6Oo�����n�IJ��6j�@0�`��F��
    c�f��v���2��S����.EP�'�9�S����!�ة�<��P��L%�Гi��S�Fd@�����HO���>x�]��e)0h�a������L�r �b�V���v�.�U,v���W�m���ʲ��VkQ�+r��Õ<�>%y�%��Y�bWZ�]�Um��-1�?C�>;w�Z�V�H��H��J���+x��+x�^����S�c�)���,z�2!� @2�d��$���M���B�9F!v����( ̖��[@�`�����W�eX���--j�⊵B!v��à��r8�;2��Hᵎ�.G�&�e=[��&�$�I��'m�'��G[��MF�hEр��a]->�Z�,"!b�=-���KP����\v�4�f�mK����8���N4?���+靖���Q�P=���mvא���]��bCi)�!�C`��7�s�*�ߨ�����d����d�x�
     @6�SM���BL���R�A����)��=l^J�S�y�e��b����C�lܾ���{�-3a�6���A�D�3�2H`��NF����ET�L��3�g����'9e���
    �&�K�,+53L9�\���Re6W6����
    ���ը�PT�fбTc�.�P��S�3��"��qV�F
    ��I��5UH��v�
    Iܭ|7,ua����x���4��@;���h��v���h��x��U�RlV�ǬZžY��ɚ�V-�P������Z�Y�Z<�d��ɪ��A�b�R���MȮ�A�ᇉ��G� Z��}��� v��V�UҬ���f��zTFԔ�j`�j�4GT˧���K�$���*'U���4R
    T�e�(��@+�
    =��x[�B��q8?�U�Oש�b��b�X
    ��0V�9�*^��Fy-X)8��Xi/������;7�I�@z��ҳQ���� =۫�
    U!KC���"�2֥���`��q���F=n��bC��3��BO��B=�f�az�l";����BDص�PQ�i�:^�x!�B?\"��K3�ԥ����L�-A�q��C{o�	D�m]QT[��(�-B��*�l̶uE��S�b��:T5ơ�]QR[��(����'�Ri���(cvQ�Ő!�M�4�v�O��)�U�4@V��nX9ƍ�1��Jm��vN	���4�x�?�V�2�*6�=����F��g�Ʈ��Bc��u�Y��u�*��ņcVѲ�By�G7!"#"�����Ir� ��ۃ�0I�1��HN��K��+[���8��ey�n����>��[/|xV��@���e4B$�r�E��T�B����X������MD�"���v��`>���/[dg��~�*�5���x+QH�½l�C��m�o�����b|e��4�C� ��$�^�!��$�y͆Xmxm�m�9jl��1y���5
    �/ŽQ�Q8@K��u�Ya�ar=��E�V؜�6�����Xa#D�)r��K�M%h�ݚd��|#O.��9�5��ӊ���cѪ�z2�Q�ݪ2�$�-�`1�Z�_���V~	VJ0�ֱ�F��@݆�J�n×��~Y�x���}e��6��A�(Q�D�"yYW��M�7�9n�2r܈������u��F���V<���f�[��)7V�qc�8׵'�����PK̤%��~PK��Y 3 topkonly_fp16/tempDir/subgraph_0_calib_raw_data.binUT
    ��fM�fM�fux��ch�ta`(�
     �{ �D@����?bG��u�[֔�Oۛm/�{˂f��PKBe��9 PK��Y@ topkonly_fp16/tempDir/subgraph_0_tidl_io_.qunat_stats_config.txtUT
    ��fM�fM�fux����1o�0�w�
    �=u�R�J�ڴ�u�X�}�>���k��,]������;[�Z��[�,�cV
    7�ڃŐ�G1R�WO��6b��U:ó��'p�n��Qc
    �O�<��vKO�+Y�=��=x�}l
    yGNpȲn*�=�'5�lp�q�>˜юg��0����*فѭ�0��v�&^��
    ������$�{'�u'�V�e[͜Ķ4���\���)5�]&��5��C�/���N.��v��xQϓO��?�nn�R�74W��,+�� ~PKO��PK��Y�< topkonly_fp16/tempDir/subgraph_0_tidl_io__stats_tool_out.binUT
    ��fM�fM�fux��Scb�����
    X`�j��PKXZ�Y�PK��Y�< topkonly_fp16/tempDir/subgraph_0_tidl_net.bin_paramDebug.csvUT
    ��fM�fM�fux���I�L-�LQ�Q�MM�s�LKK-J�KN�+P�@i���t����7(5YE~Q:P��Baib^IfUj�/�t,��$�X�ʡ�����
    tzd�(pPK��aJd�PK��Yr= topkonly_fp16/tempDir/subgraph_0_tidl_io_.perf_sim_config.txtUT
    ��fM�fM�fux����Ao�@�����ri%'�r�CĬUT�.W�e��1�
    l��$r~}g�4q�S/��ӛy=�B>"�=$s(�0�{� �r
    ��Z��;��`
    ��q�L �J�=5������ >�\�#k�rh����" �w�iE�9���5����ŋ#,K��j��l3n�hs�E@'`���Ԡ�M��%��q5v5h�9J�D�%Ƞ�"O�d��Ak]c����<ei�2N̗w�b��|�Kx�������so���iA�8���  �{Y��(�)�W0��I!ې2�����2?���n4$^z��KƳ0e�S�����
    ˋ���J&�z��^嫜}�����S�|��Vq��g�|��V��k��߫�s��܋��~��Ā���%/��!S�n�g��R��v�[��"�}3�-�Z>�V��������u���I����zn��M��)y���\�a�}��PK�z-��rPK��Y topkonly_fp16/allowedNode.txtUT
    ��fM�fM�fux��3�2�2�2�2�PK�6#�PK��Y�  topkonly_fp16/onnxrtMetaData.txtUT
    ��fM�fM�fux���+�u/J,���OI-�5�2�*.M��$��9� N�P&�4�3��������/-�s3A2�)�:�`�KbI"���F�yi:��@
    (�_Z��,NMM)���e�9\PKb�6�h�PKڄY��g�;1W ��log_topkonly_fp16_c.txtUT
    E�fl�fj�fux��PK��Y �A�topkonly_fp16/UT
    ��f A�fc�fux��PK��Y �A�topkonly_fp16/tempDir/UT
    ��f A�fM�fux��PK��Y= �d& ��@topkonly_fp16/tempDir/graphvizInfo.txtUT
    ��fM�fM�fux��PK��Y-ŗ�^60 ���
    topkonly_fp16/tempDir/runtimes_visualization.svgUT
    ��fM�fM�fux��PK��Y�쭜&F8 ��topkonly_fp16/tempDir/subgraph_0_tidl_net.bin_netLog.txtUT
    ��fM�fM�fux��PK��Ygߐ��{a- ���topkonly_fp16/tempDir/subgraph_0_tidl_net.binUT
    ��fM�fM�fux��PK��Y�w���q. ��4topkonly_fp16/tempDir/subgraph_0_tidl_io_1.binUT
    ��fM�fM�fux��PK��Y�Q�u< ��w5topkonly_fp16/tempDir/subgraph_0_tidl_net.bin.layer_info.txtUT
    ��fM�fM�fux��PK��Y̤%��~1 ��v6topkonly_fp16/tempDir/subgraph_0_tidl_net.bin.svgUT
    ��fM�fM�fux��PK��YBe��9 3 ���Htopkonly_fp16/tempDir/subgraph_0_calib_raw_data.binUT
    ��fM�fM�fux��PK��YO��@ ���Itopkonly_fp16/tempDir/subgraph_0_tidl_io_.qunat_stats_config.txtUT
    ��fM�fM�fux��PK��YXZ�Y�< ��;Ktopkonly_fp16/tempDir/subgraph_0_tidl_io__stats_tool_out.binUT
    ��fM�fM�fux��PK��Y��aJd�< ���Ktopkonly_fp16/tempDir/subgraph_0_tidl_net.bin_paramDebug.csvUT
    ��fM�fM�fux��PK��Y�z-��r= ���Ltopkonly_fp16/tempDir/subgraph_0_tidl_io_.perf_sim_config.txtUT
    ��fM�fM�fux��PK��Y�6#� ��Otopkonly_fp16/allowedNode.txtUT
    ��fM�fM�fux��PK��Yb�6�h�  ���Otopkonly_fp16/onnxrtMetaData.txtUT
    ��fM�fM�fux��PK$kP

    the zip files constains the logs and the modelartifacts , please do remove .txt extension

  • Hi Abhilash,

    due to the production development from customer we pefer to execute in the environment of  edgeai 9.2.7 and tidltools 9.2.0 can you please provide alternate solutions for this 

    Thanks for the information, I have few questions 

    1) The latest 9.2 tools (9.2 sdk firmware) may i know what dependency are you taking here ?

    2) Regarding 10.0 tools yes its needs updated firmware, which is part of 10 sdk, are you saying there are customers in production with 9.2 sdk ?

    Also Abhilash,

    We have request to please insert zip files only on our windows machines we cant make use of .txt files, nevertheless the platform supports zip files uploading.

    Pratik

  • 1) yes , we are using 9.2.0 and the dependency is from our system level 

    2) as we are in the final stage of releasing our work migration is not feasible to 10.0 we would be expecting solution for 9.2.7 version 

  • Hi Abhilash,

    I've taken a look at the model artifacts you have provided. It doesn't seem that the ONNX toy model was attached, but based on the logs, it seems that you have included your TopK layer in this model. 

    • As previously discussed in the TopK thread, there was difficulty with this layer, are these observations from the toy model made with/without fixes from your end?
    • Can you supply the toy model? It does not appear to be in the zip file from my end. 
    • Is there a reason for a dependence on 9.2.7 versus 9.2.9? Both will work with the 9.2 SDK. Can you please share the full log file from 9.2 as txt file?

    Best,

    Asha

  • topk_only.zip

    Hello Asha, 

    1. i have attached the toy model please have a look into it 

    2. Based on the pratik and anush suggestion the topk is modified 

    3. Since we have done most of the developments in the 9.2.7 we would like stick to that version as this shall be going into production , the log files are re-shared as txt again please have alook 

    root@cf0e25348060:/home/root/examples/osrt_python/ort# python3 onnxrt_ep_cnd_topkonly.py -c
    Skipping import of model optimizer
    Available execution providers :  ['TIDLExecutionProvider', 'TIDLCompilationProvider', 'CPUExecutionProvider']
    
    Running 1 Models - ['condlanet_FNC_SafetyNoEdge_NewForkV1']
    
    
    Running_Model :  condlanet_FNC_SafetyNoEdge_NewForkV1  
    
    
    Running shape inference on model /home/root/lucid/model_onnx/topk_only.onnx 
    
    ************Started the Interpreter 
     
    tidl_tools_path                                 = /home/root/tidl_tools 
    artifacts_folder                                = ../../../model-artifacts//condlanet_FNC_SafetyNoEdge_NewForkV1/ 
    tidl_tensor_bits                                = 16 
    debug_level                                     = 4 
    num_tidl_subgraphs                              = 16 
    tidl_denylist                                   = 
    tidl_denylist_layer_name                        = 
    tidl_denylist_layer_type                        = 
    tidl_allowlist_layer_name                       = 
    model_type                                      =  
    tidl_calibration_accuracy_level                 = 64 
    tidl_calibration_options:num_frames_calibration = 3 
    tidl_calibration_options:bias_calibration_iterations = 3 
    mixed_precision_factor = -1.000000 
    model_group_id = 0 
    power_of_2_quantization                         = 2 
    ONNX QDQ Enabled                                = 0 
    enable_high_resolution_optimization             = 0 
    pre_batchnorm_fold                              = 1 
    add_data_convert_ops                            = 3 
    output_feature_16bit_names_list                 =  
    m_params_16bit_names_list                       =  
    m_single_core_layers_names_list                 =  
    Inference mode                                  = 0 
    Number of cores                                 = 1 
    reserved_compile_constraints_flag               = 1601 
    partial_init_during_compile                     = 0 
    ti_internal_reserved_1                          = 
    
    ========================= [Model Compilation Started] =========================
    
    Model compilation will perform the following stages:
    1. Parsing
    2. Graph Optimization
    3. Quantization & Calibration
    4. Memory Planning
    
    ============================== [Version Summary] ==============================
    
    -------------------------------------------------------------------------------
    |          TIDL Tools Version          |              10_00_04_00             |
    -------------------------------------------------------------------------------
    |         C7x Firmware Version         |              10_00_02_00             |
    -------------------------------------------------------------------------------
    |            Runtime Version           |            1.14.0+10000005           |
    -------------------------------------------------------------------------------
    |          Model Opset Version         |                   9                  |
    -------------------------------------------------------------------------------
    
    NOTE: The runtime version here specifies ONNXRT_VERSION+TIDL_VERSION
    Ex: 1.14.0+1000XXXX -> ONNXRT 1.14.0 and a TIDL_VERSION 10.00.XX.XX
    
    ============================== [Parsing Started] ==============================
    
    [TIDL Import]  WARNING: 'meta_layers_names_list' is not provided - running OD post processing in ARM mode
    Number of OD backbone nodes = 0 
    Size of odBackboneNodeIds = 0 
    [TIDL Import] [PARSER] UNSUPPORTED: Layer type not supported by TIDL --- layer type - TopK,  Node name - TopK_219 -- [tidl_onnxRtImport_core.cpp, 503]
    [TIDL Import] [PARSER] SUPPORTED: Layers type supported by TIDL --- layer type - Slice,  Node name - Slice_nf_0 -- [tidl_onnxRtImport_core.cpp, 524]
    [TIDL Import] [PARSER] SUPPORTED: Layers type supported by TIDL --- layer type - Slice,  Node name - Slice_nf_1 -- [tidl_onnxRtImport_core.cpp, 524]
    [TIDL Import] [PARSER] SUPPORTED: Layers type supported by TIDL --- layer type - Reshape,  Node name - Reshape_221 -- [tidl_onnxRtImport_core.cpp, 524]
    [TIDL Import] [PARSER] SUPPORTED: Layers type supported by TIDL --- layer type - Reshape,  Node name - Reshape_223 -- [tidl_onnxRtImport_core.cpp, 524]
    
    ------------------------- Subgraph Information Summary -------------------------
    -------------------------------------------------------------------------------
    |          Core           |      No. of Nodes       |   Number of Subgraphs   |
    -------------------------------------------------------------------------------
    | C7x                     |                       4 |                       1 |
    | CPU                     |                       1 |                       x |
    -------------------------------------------------------------------------------
    -------------------------------------------------------
    | Node | Node Name |              Reason              |
    -------------------------------------------------------
    | TopK | TopK_219  | Layer type not supported by TIDL |
    -------------------------------------------------------
    Running Runtimes GraphViz - /home/root/tidl_tools/tidl_graphVisualiser_runtimes.out ../../../model-artifacts//condlanet_FNC_SafetyNoEdge_NewForkV1//allowedNode.txt ../../../model-artifacts//condlanet_FNC_SafetyNoEdge_NewForkV1//tempDir/graphvizInfo.txt ../../../model-artifacts//condlanet_FNC_SafetyNoEdge_NewForkV1//tempDir/runtimes_visualization.svg 
    ============================= [Parsing Completed] =============================
    
    TIDL_createStateImportFunc Started:
    Compute on node : TIDLExecutionProvider_TIDL_0_0
      0,           Slice, 1, 1, 501_nf, 501
      1,         Reshape, 2, 1, 501, seeds_val
      2,           Slice, 1, 1, 502_nf, 502
      3,         Reshape, 2, 1, 502, seeds
    
    Input tensor name -  502_nf 
    
    Input tensor name -  501_nf 
    Output tensor name - seeds 
    Output tensor name - seeds_val 
    ****************************** Start of session *******************
    
    ***************Running_Inference Section **********
    
    This is Lucid Model for image /home/root/lucid/data_onnx/test_images/img0277.png
    In TIDL_onnxRtImportInit subgraph_name=subgraph_0
    Layer 0, subgraph id subgraph_0, name=seeds
    Layer 1, subgraph id subgraph_0, name=seeds_val
    Layer 2, subgraph id subgraph_0, name=502_nf
    Layer 3, subgraph id subgraph_0, name=501_nf
    ==================== [Optimization for subgraph_0 Started] ====================
    
    In TIDL_runtimesOptimizeNet: LayerIndex = 8, dataIndex = 6 
    [TIDL Import] [PARSER] WARNING: Requested input data convert layer is not added to the network, It is currently not optimal
    ----------------------------- Optimization Summary -----------------------------
    --------------------------------------------------------------------------
    |      Layer      | Nodes before optimization | Nodes after optimization |
    --------------------------------------------------------------------------
    | TIDL_SliceLayer |                         2 |                        2 |
    --------------------------------------------------------------------------
    
    =================== [Optimization for subgraph_0 Completed] ===================
    
    In TIDL_runtimesPostProcessNet 
    ************ in TIDL_subgraphRtCreate ************ 
     The soft limit is 10240
    The hard limit is 10240
    MEM: Init ... !!!
    MEM: Init ... Done !!!
     0.0s:  VX_ZONE_INIT:Enabled
     0.14s:  VX_ZONE_ERROR:Enabled
     0.18s:  VX_ZONE_WARNING:Enabled
     0.3151s:  VX_ZONE_INIT:[tivxInit:190] Initialization Done !!!
    ************ TIDL_subgraphRtCreate done ************ 
     ============= [Quantization & Calibration for subgraph_0 Started] =============
    
    *******   In TIDL_subgraphRtInvoke  ******** 
       0         1.00000         0.00000       255.00000 5
       2         1.00000         0.00000         0.00000 6
       1         1.00000         0.00010         0.99535 6
       3         1.00000         0.00010         0.99535 6
       4         1.00000         0.00010         0.00010 6
       5         1.00000         0.00000         0.00000 6
       6         1.00000         0.00010         0.00010 6
       7         1.00000         0.00000         0.00000 6
       8         1.00000         0.00010         0.00010 6
       9         1.00000         0.00000         0.00000 6
      10         1.00000         0.00010         0.00010 6
      11         1.00000         0.00000         0.00000 6
      13         1.00000         0.00010         0.00010 6
     Layer,   Layer Cycles,kernelOnlyCycles, coreLoopCycles,LayerSetupCycles,dmaPipeupCycles, dmaPipeDownCycles, PrefetchCycles,copyKerCoeffCycles,LayerDeinitCycles,LastBlockCycles, paddingTrigger,    paddingWait,LayerWithoutPad,LayerHandleCopy,   BackupCycles,  RestoreCycles,Multic7xContextCopyCycles,
         2,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         3,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         4,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         5,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         6,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         7,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         8,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         9,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        10,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        11,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        13,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
     Sum of Layer Cycles 0 
    Sub Graph Stats 20.000000 1847.000000 488.000000 
    *******  TIDL_subgraphRtInvoke done  ******** 
    2
    
    ***************Running_Benchmark_Section **********
    
    ****************************** Start of session *******************
    
    ***************Running_Inference Section **********
    
    This is Lucid Model for image /home/root/lucid/data_onnx/test_images/img0278.png
    *******   In TIDL_subgraphRtInvoke  ******** 
       0         1.00000         0.00000       255.00000 5
       2         1.00000         0.00000         0.00010 6
       1         1.00000         0.00010         0.99535 6
       3         1.00000         0.00010         0.99535 6
       4         1.00000         0.00010         0.00010 6
       5         1.00000         0.00000         0.00010 6
       6         1.00000         0.00010         0.00010 6
       7         1.00000         0.00000         0.00010 6
       8         1.00000         0.00010         0.00010 6
       9         1.00000         0.00000         0.00010 6
      10         1.00000         0.00010         0.00010 6
      11         1.00000         0.00000         0.00010 6
      13         1.00000         0.00010         0.00010 6
     Layer,   Layer Cycles,kernelOnlyCycles, coreLoopCycles,LayerSetupCycles,dmaPipeupCycles, dmaPipeDownCycles, PrefetchCycles,copyKerCoeffCycles,LayerDeinitCycles,LastBlockCycles, paddingTrigger,    paddingWait,LayerWithoutPad,LayerHandleCopy,   BackupCycles,  RestoreCycles,Multic7xContextCopyCycles,
         2,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         3,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         4,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         5,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         6,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         7,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         8,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         9,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        10,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        11,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        13,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
     Sum of Layer Cycles 0 
    Sub Graph Stats 11.000000 1431.000000 397.000000 
    *******  TIDL_subgraphRtInvoke done  ******** 
    2
    
    ***************Running_Benchmark_Section **********
    
    ****************************** Start of session *******************
    
    ***************Running_Inference Section **********
    
    This is Lucid Model for image /home/root/lucid/data_onnx/test_images/img0279.png
    *******   In TIDL_subgraphRtInvoke  ******** 
       0         1.00000         0.00000       255.00000 5
       2         1.00000         0.00000         0.00010 6
       1         1.00000         0.00010         0.99535 6
       3         1.00000         0.00010         0.99535 6
       4         1.00000         0.00010         0.00010 6
       5         1.00000         0.00000         0.00010 6
       6         1.00000         0.00010         0.00010 6
       7         1.00000         0.00000         0.00010 6
       8         1.00000         0.00010         0.00010 6
       9         1.00000         0.00000         0.00010 6
      10         1.00000         0.00010         0.00010 6
      11         1.00000         0.00000         0.00010 6
      13         1.00000         0.00010         0.00010 6
     Layer,   Layer Cycles,kernelOnlyCycles, coreLoopCycles,LayerSetupCycles,dmaPipeupCycles, dmaPipeDownCycles, PrefetchCycles,copyKerCoeffCycles,LayerDeinitCycles,LastBlockCycles, paddingTrigger,    paddingWait,LayerWithoutPad,LayerHandleCopy,   BackupCycles,  RestoreCycles,Multic7xContextCopyCycles,
         2,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         3,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         4,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         5,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         6,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         7,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         8,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
         9,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        10,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        11,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
        13,              0,              0,              0,              0,              0,                 0,              0,                 0,              0,              0,              0,              0,              0,              0,              0,              0,              0,
     Sum of Layer Cycles 0 
    Sub Graph Stats 36.000000 2141.000000 296.000000 
    *******  TIDL_subgraphRtInvoke done  ******** 
    In TIDL_runtimesPostProcessNet 
    
    -------- Running Calibration in Float Mode to Collect Tensor Statistics --------
    [=============================================================================] 100 %
    
    ------------------ Fixed-point Calibration Iteration [1 / 3]: ------------------
    [=============================================================================] 100 %
    
    ------------------ Fixed-point Calibration Iteration [2 / 3]: ------------------
    [=============================================================================] 100 %
    
    ------------------ Fixed-point Calibration Iteration [3 / 3]: ------------------
    [=============================================================================] 100 %
    
    ==================== [Quantization & Calibration Completed] ====================
    
    ========================== [Memory Planning Started] ==========================
    
    
    ------------------------- Network Compiler Traces ------------------------------
    Wrong data type for layer input 2, 14
    Wrong data type for layer output 2, 14
    Wrong data type for layer input 7, 38
    Wrong data type for layer output 7, 38
    Error : Error Code = <ERR_UNSUPPORTED_DATA_TYPE>
    [TIDL Import]  FATAL ERROR: Network Compiler failed to execute - Memory planning failed with return code - 65280 -- [tidl_import_core.cpp, 1002]
    [TIDL Import] Aborting