The IO Adapter subgraph delivers the \(\textbf{A}\) matrix from the input PLIO to the QRD subgraph. Buffers are used for I/O for all downstream MUSIC subgraphs. All of these subgraphs may use a single I/O buffer read over a high bandwidth memory interface, moving from tile to tile in a linear fashion. This is clarified in more detail below. No bandwidth limitations are encountered downstream due to this use of the 256-bit AI Engine memory interface. However, the design must be fed by two PLIO streams @ 32-bits for 1250 MHz to achieve a 1 \(\mu s\) throughput overall. The IO Adapter subgraph sinks two input PLIO streams and combines them into a single output buffer containing the input \(\textbf{A}\) matrix to be processed by the first QRD subgraph. Two streams are required because the \(128\times 8\) elements of \(\textbf{A}\) cannot be transferred over a single PLIO in 1 \(\mu s\). The following block diagram shows the AI Engine physical array view for the IO Adapter subgraph.