Visible to Intel only — GUID: lle1711747388961
Ixiasoft
Visible to Intel only — GUID: lle1711747388961
Ixiasoft
1.1. Programmed Input/Output Design Example
The GTS AXI Streaming PLD clock runs up to 500MHz for PCIe* 4.0 x8 in the Agilex™ 5 D-Series device, 350MHz for PCIe* 4.0 x4 in the Agilex™ 5 E-Series device, and 300MHz in the Agilex™ 3 device with the user interface data width of 512 bits, 256 bits, and 128 bits, respectively. The user application in the design example is using the PLD clock source from the p<n>_coreclkout_hip_toapp to drive the AXI Stream interface or the p<n>_axi_st_clk input clock.
- Generated GTS AXI Streaming IP as Endpoint Variant (DUT)
- Programmable I/O Application (PIO)
- On-Chip Memory (MEM)
The GTS AXI Streaming IP design under test (DUT) is configured as an Endpoint receiving the PIO transactions from the root complex over the PCIe* link and transferring them to the PIO application module. The TLP received in the PIO application module is decoded and converted into Avalon® memory-mapped interface format. Depending on the received instruction, the on-chip memory is the targeted space to store or read data.
The PIO design example automatically creates the required files for simulation and compilation in the Quartus® Prime software. However, it does not cover all possible GTS AXI Streaming IP parameterizations.
- There is no support on the back-to-back TLP packets from the host processor.
- TLP prefix is not used and intended with Single Physical Function (PF).
- There is no requirement for error message, interrupt or status bit toggling handling.
- The backpressure mechanism of the DUT is handled through the ready signal. Besides ready, receive (RX) signals can also be backpressured through p0_app_ss_st_rx_tuser_halt. There is also the (transmit) TX credit mechanism of back pressure.
- Does not include the full features of the GTS AXI Streaming IP.