Open hansfbaier opened 1 year ago
How could I debug this and get to the bottom of this issue?
@gatecat Could it be that if the placer runs out of SLICEMs that it just loops endlessly? That is what it looks like. There is a lot of LUT ram in the design and after some time placing them, it does not seem to get any valid SLICEM tiles anymore. Is there somewhere kept track of how many SLICEM's have been allocated? I guess it is not possible to know that before placing?
Anyway, the authors of the demo project said they could successfully place and route the project with both vivado and f4pga/VPR, so in principle the resources should be available.
This example SoC build well with f4pga and Vivado, but does not terminate with default settings: https://github.com/chili-chips-ba/openXC7-TetriSaraj.git 29a8501669b167c55efc5da87a8ae7402271d986 When I use
--placer-heap-cell-placement-timeout 2048
then it terminates with the error messageThis looks like something is wrong with the RAM placer constraints. Could that be a casualty of https://github.com/gatecat/nextpnr-xilinx/pull/56 ?
Update: I tried to build without
-nobram
in yosys, but I get a similar result: