<tpb>
Title: WIP: Tilegrid generation for Zynq 7020 by tmichalak · Pull Request #1162 · SymbiFlow/prjxray · GitHub (at github.com)
<litghost>
xi53: Changing "settings/artix7.sh" part and re-run should generate a 200tfbg676 database.
<litghost>
xi53: If you are inclined, we do need a way to run the 005-tilegrid and 074-dump_all fuzzer against different parts, to generate tilegrid and tileconn for different fabrics
<litghost>
hackerfoo: Most of those failures are in the bad direction
<litghost>
k4_N8_topology-0.85sL2-0.15gL4-on-cb-off-sb_22nm_22nm.xml/mkDelayWorker32B.v/common place_time: golden = 6.97 result = 151.71
<litghost>
In general though the nightly/weekly QoR results are not super interesting to us
<litghost>
We mostly care about the QoR / runtime from the VTR + symbiflow run
<hackerfoo>
The only changes I made were adding the Cost_Entry fill field back and removing an unused struct definition, so I'm not sure what happened.
<hackerfoo>
Wait, I'm not sure that test finished before.
<hackerfoo>
Yeah, it failed in the same way on earlier commits.
<_whitenotifier-5>
[vtr-verilog-to-routing] HackerFoo opened issue #341: Penalize base_cost outside of lookahead cost map - https://git.io/JeD2S
mohamed has quit [Quit: Connection closed for inactivity]
<bunnie>
litghost: mithro: thanks for the explanations, really appreciate it. It would be the case that any mods that activate an previously unknown feature would come out in a diff at least of the bitstreams, which is probably good enough....
bunnievorpal[m] has joined #symbiflow
bunnie has left #symbiflow [#symbiflow]
bunnievorpal[m] is now known as bunnie[m]
nrossi has quit [Quit: Connection closed for inactivity]