Huawei uses machine translation combined with human proofreading to translate this document to different languages in order to help you better understand the content of this document.
Note: Even the most advanced machine translation cannot match the quality of professional translators.
Huawei shall not bear any responsibility for translation accuracy and it is recommended that you refer to the English document (a link for which has been provided).
Preparing Input and Output Data for Model Inference
Figure 4-11 Input and output data preparation flowchart
![]()
Data of type aclmdlDesc is used to describe the model, for example, the number of inputs or outputs, names, data types, formats and dimensions. Data of type aclmdlDataset is used to describe the input and output data. A model may have multiple inputs and outputs, and data of the aclDataBuffer type is used to describe the memory address and memory size of each input or output.
Figure 4-12 Relationship between aclmdlDataset and aclDataBuffer
![]()
Call acl.rt.memcpy (synchronous mode) or acl.rt.memcpy_async (asynchronous mode) to implement data transfer from the host to the device through memory copy.
For model with multiple inputs and outputs, when adding aclDataBuffer to the aclmdlDataset, you are advised to obtain the names of the inputs and outputs first, and add aclDataBuffer in accordance with the corresponding index sequence.
In fixed-batch scenarios with batchSize >1, before creating data of type aclDataBuffer, acl.mdl.get_input_size_by_index needs to be called to obtain the required memory size using the index of the input.