Visible to Intel only — GUID: qes1677540728783
Ixiasoft
2.5.2.1. Parameter Group: Global Parameters
2.5.2.2. Parameter Group: activation
2.5.2.3. Parameter Group: pe_array
2.5.2.4. Parameter Group: pool
2.5.2.5. Parameter Group: depthwise
2.5.2.6. Module: softmax
2.5.2.7. Parameter Group: dma
2.5.2.8. Parameter Group: xbar
2.5.2.9. Parameter Group: filter_scratchpad
2.5.2.10. Parameter Group: input_stream_interface
2.5.2.11. Parameter Group: output_stream_interface
2.5.2.12. Parameter Group: config_network
2.5.2.13. (Early Access) Parameter Group: layout_transform_params
Visible to Intel only — GUID: qes1677540728783
Ixiasoft
2.6.4.1. Multiple Input Graphs
For graphs with more than one input, each tensor is structured as described in the previous section. The multiple input tensors must be packed together at address offsets as reported by the FPGA AI Suite compiler.
The compiler generates CSV files that describe the input and output tensor, unless you specify the --fno-transform-tables option. Each row of the CSV file gives information about one input. For more details, refer to the FPGA AI Suite Compiler Reference Manual .
For multiple inputs, the inputs are stored by batch and then by input number. For example, for 3 inputs and 2 batches, the input tensors would be stored as follows:
- input 1, batch 1
- input 1, batch 2
- input 2, batch 1
- input 2, batch 2
- input 3, batch 1
- input 3, batch 2