The AIE to HLS Kernel block reformats a signal driven by a port of an AI Engine Kernel block, so that the resulting signal
matches the data type and complexity required by an HLS Kernel block. For example, if
the AI Engine's output data type is cint16 and the HLS
Kernel's input data type is ap_axis<64>. In this
case, use an AIE to HLS kernel block with parameter output type set to uint64. This block reads cint16 samples from its input and then packs pairs of subsequent cint16 samples into uint64
samples to its output.
Figure 1. AIE to HLS Kernel Block
Double-click the block symbol to see the parameters of the AIE to HLS Kernel block.
- Output Data Type
- You can select from cint16, int32, uint32, cint32, int64, uint64, sfix128, ufix128, float, cfloat.
- Output Size
- The size of the output port. The output port is a variable-sized signal. The
Output Size
parameter specifies its maximum size. The Output Size parameter has a lower
bound, calculated based on the input data type, input size, and output data
type. For example, if the input data type is
cint16, the input size is 8, and the output data type isint32. In this case, the Output Size parameter must be at least 4. To use a smaller Output Size, enable the Allow Multiple-Rate Processing parameter. - Allow Multiple-Rate Processing
- When you enable this property, the Output Size parameter has no lower bound. If you set the Output Size parameter to a value less than the original lower bound, the output sample rate is greater than the input sample rate.
- Input Data Type
- Possible values are: int8, int16, int32, int64, uint8, uint16, uint32, uint64, cint16, cint32, float, cfloat, bfloat16, cbfloat16.