Forums - When will Matmul be supported?

2 posts / 0 new
or Register
Last post
When will Matmul be supported?
lewis_bo
Join Date: 27 Mar 22
Posts: 4
Posted: Fri, 2022-05-06 04:36
Top

When I run the  "./snpe-onnx-to-dlc -i best.onnx -o best.dlc" command to convert the ONNX model to DLC, the following warning is generated:

/opt/snpe/lib/python/qti/aisw/converters/backend/ir_to_dlc.py:1037: RuntimeWarning: error_code=106; error_message=Incorrect/unsupported input dimensions is set. Layer MatMul_133: mismatch between 330 (400) and dimension of 1350 (1); error_component=System Configuration; line_no=1983; thread_id=140014041274176
  node.output_names[0])
/opt/snpe/lib/python/qti/aisw/converters/backend/ir_to_dlc.py:341: RuntimeWarning: info_code=802; message=Layer parameter value is invalid in GPU. Layer 1351 : output width = 512, depth = 512 width * depth (packed) = 65536 exceeds maximum image width 16384 for Adreno A650; component=GPU Runtime; line_no=981; thread_id=140014041274176
  node.op.quantizable)
/opt/snpe/lib/python/qti/aisw/converters/backend/ir_to_dlc.py:1037: RuntimeWarning: error_code=106; error_message=Incorrect/unsupported input dimensions is set. Layer MatMul_136: mismatch between 334 (400) and dimension of 1351 (1); error_component=System Configuration; line_no=1983; thread_id=140014041274176
  node.output_names[0])
/opt/snpe/lib/python/qti/aisw/converters/backend/ir_to_dlc.py:341: RuntimeWarning: info_code=802; message=Layer parameter value is invalid in GPU. Layer 1352 : output width = 512, depth = 512 width * depth (packed) = 65536 exceeds maximum image width 16384 for Adreno A650; component=GPU Runtime; line_no=981; thread_id=140014041274176
  node.op.quantizable)
/opt/snpe/lib/python/qti/aisw/converters/backend/ir_to_dlc.py:1037: RuntimeWarning: error_code=106; error_message=Incorrect/unsupported input dimensions is set. Layer MatMul_137: mismatch between 334 (400) and dimension of 1352 (1); error_component=System Configuration; line_no=1983; thread_id=140014041274176
  node.output_names[0])
And the conversion was successful
INFO - INFO_CONVERSION_SUCCESS: Conversion completed successfully
 
But when I use best.dlc in CPU Runtime, I get an error message:
     Caused by: com.qualcomm.qti.snpe.SnpeError$NativeException: Unable to create network! Cause: error_code=106; error_message=Incorrect/unsupported input dimensions is set. error_code=106; error_message=Incorrect/unsupported input dimensions is set. Layer MatMul_133: mismatch between 330 (400) and dimension of 1350 (1); error_component=System Configuration; line_no=1983; thread_id=528295382192; error_component=System Configuration; line_no=267; thread_id=540330472696
        at com.qualcomm.qti.snpe.SnpeError.getSnpeNativeError(SnpeError.java:35)
        at com.qualcomm.qti.snpe.internal.NativeNetwork.<init>(NativeNetwork.java:83)
        at com.qualcomm.qti.snpe.SNPE$NeuralNetworkBuilder.build(SNPE.java:415)
        at com.lixiang.dip.ai.encryptpic.model.SnpeModelLoader.load(SnpeModelLoader.java:89)
 
  • Up0
  • Down0
Re: When will Matmul be supported?
SahilBandar
Join Date: 23 May 18
Posts: 37
Posted: Tue, 2022-05-10 08:28
Top

Hi lewis_bo,

There is support for MatMul is already available in SNPE.
Please refer the URL given for the same, Snapdragon Neural Processing Engine SDK: Supported Network Layers (qualcomm.com)

Please try to cross-verify with the input shape you passed while converting the model to ONNX. or try to give input shape explicitly while converting.

Thanks & Regards,
Sahil Bandar

  • Up0
  • Down0
or Register

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries (“Qualcomm”). The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.