Check | Expected artifact |
|---|---|
Lint | Zero errors; reviewed warnings/waivers |
CDC/RDC | Crossing report with approved structures, reset epochs and containment |
SDC | Primary/generated clocks, IO delays, crossing-specific exceptions and effective bundled-data/Gray bounds; no broad group silently overrides required paths |
Synthesis | Storage mapping (flop array, register file or macro, as CP-BUF chose), area and unconstrained-path report |
STA | Setup/hold by mode/corner, interstage timing and clock-gating checks |
Power | Clock gating/activity report and mode assumptions |
DFT | Scan-ready lint, test clock/reset hooks, memory inventory |
Physical | Hierarchy, floorplan pins, congestion/fanout review |
CP-CLOCK gives two unrelated clocks. The fragment below declares them and bounds every path between them instead of cutting those paths.
# not-run: SDC is not executable here
# CP-CLOCK: clk_io 100 MHz, clk_dp 125 MHz, no phase relationship
create_clock -name clk_io -period 10.000 [get_ports clk_io]
create_clock -name clk_dp -period 8.000 [get_ports clk_dp]
# Every crossing (Gray pointers, handshake req/ack, config bundle,
# response record) gets a datapath-only bound. 8.000 ns is the shorter
# period; derive the real bound from each FIFO and handshake design.
set_max_delay -datapath_only 8.000 -from [get_clocks clk_io] -to [get_clocks clk_dp]
set_max_delay -datapath_only 8.000 -from [get_clocks clk_dp] -to [get_clocks clk_io]
# Not used: set_clock_groups -asynchronous would also remove the bounds above.
# IO delays for the AXI-Lite and stream ports follow, per interface.Read it as two declarations and two bounds. The bound keeps Gray-pointer bits and bundled data from arriving so late, or so skewed, that a synchroniser samples a mix of old and new values.
These reports use the exact parameter set and clock modes tested in stage 6. A manifest binds the elaborated design, parameter revisions, constraints, libraries, tools, and report extraction scripts so the signoff review can reproduce every headline result.
The synthesis memory inference/mapping report, supported by area/cell counts and the expected storage instance in the netlist. Record the chosen resource in CP-BUF so the report has a target to match.