search

gatecat / prjoxide

Documenting Lattice's 28nm FPGA parts
ISC License
143 stars 15 forks source link

main router loop running indefinitely #39

Open slagernate opened 1 year ago

slagernate commented 1 year ago

It seems to never end:

Info:     iter=20608 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20609 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20610 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20611 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20612 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20613 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20614 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20615 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20616 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20617 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20618 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20619 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20620 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20621 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20622 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20623 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20624 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20625 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20626 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20627 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20628 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20629 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20630 wires=8032 overused=1 overuse=1 archfail=NA
Info:     iter=20631 wires=8032 overused=1 overuse=1 archfail=NA
^Cmake: *** [Makefile:31: ppm_uart_bridge.fasm] Interrupt

I can provide a reproducible example momentarilly...

gatecat commented 1 year ago

Thanks, this is unfortunately a bit of a long-running issue in nextpnr, new causes seem to keep popping up. A testcase would be appreciated (although, it's worth noting that small changes will usually tip the balance to or from not routing).

slagernate commented 1 year ago

Seems to happen when using amaranth as well. Can you point me in the right direction if I were to look into nextpnr to fix this? I am curious to not only resolve this, but to have a better understanding of the tools.