19:09 <daveshah> Am I right in saying that based on https://github.com/SymbiFlow/prjxray/blob/master/fuzzers/074-dump_all/artix7_ignored_wires.txt ; the IOI2GCLK wires are currently broken in the xray database?

19:09 <tpb> Title: prjxray/artix7_ignored_wires.txt at master · SymbiFlow/prjxray · GitHub (at github.com)

22:19 <litghost> daveshah: One minute. I believe what you are seeing there is the tiles that are not above/below the middle of the CMT, and therefore no IOI2GCLK connection exists

22:26 <daveshah> Thanks! I was seeing some odd behaviour and what looked at a glance like a mismatch between how nextpnr was routing things and Vivado and didn't want too spend too much time on it if there was an xray issue

22:26 <daveshah> I can look into it some more tomorrow though

22:27 <litghost> daveshah: So I took a look, and I think it kind of works out

22:27 <litghost> daveshah: At a minimum the CCIO pins that are directly above and below the CMT middle are okay

22:28 <litghost> daveshah: It is possible that the CCIO pins that 2 above and below the CMT are missing their connections

22:28 <litghost> daveshah: I'll look into the connection database for the a7 and see

22:29 <litghost> daveshah: In particular the line "LIOI3_TBYTESRC_X0Y19/LIOI_I2GCLK_TOP1" may mean that the node "LIOI3_X0Y21/LIOI_I2GCLK_TOP0" is not probably preserved

22:29 <daveshah> The problematic pin was RIOI3_X43Y67 (clock input for Arty A35T)

22:30 <litghost> daveshah: That tile is not a CCIO pin, so it lacks a IOI2GCLK wire

22:30 <daveshah> oh, sorry, typo

22:30 <litghost> daveshah: Only the first two IOB33M pins above and below the CMT are CCIO pins

22:31 <litghost> daveshah: Am I missing something?

22:31 <daveshah> I copied the reset pin instead. I meant X43Y75

22:31 <daveshah> That's not in the list, so it should work fine

22:31 <daveshah> I'll have more of a look tomorrow

22:31 <daveshah> thanks for your help!

22:32 <litghost> daveshah: So that is the first IOB33M above the CMT, so I expect it should be okay?

22:32 <litghost> daveshah: I'll check my database and see

22:32 <daveshah> https://usercontent.irccloud-cdn.com/file/uqnuFRMX/Screenshot%20from%202020-01-16%2022-32-34.png

22:32 <daveshah> this is the route that Vivado uses, fwiw

22:33 <daveshah> It's not a blocking issue as using general routing seems to work fin

22:33 <litghost> I'm checking my database to see

22:37 <litghost> https://gist.github.com/litghost/78c03fccab50e16bb82a9ba31cc2ce6e

22:37 <tpb> Title: gist:78c03fccab50e16bb82a9ba31cc2ce6e · GitHub (at gist.github.com)

22:37 <litghost> So from symbiflow-arch-defs interpretation of the prjxray-db, that node is correct

22:38 <daveshah> Ack, I suspect it is something odd with how I'm importing the database. I'll check when I'm back at my desk tomorrow

22:43 <litghost> daveshah: I checked LIOI3_X0Y21/LIOI_I2GCLK_TOP0, and it looks good too: https://gist.github.com/litghost/9c65c6a726ecda42cd3ea191307dc059

22:43 <tpb> Title: gist:9c65c6a726ecda42cd3ea191307dc059 · GitHub (at gist.github.com)

22:44 <litghost> daveshah: So I believe the CCIO nodes should be okay. If you have another node you suspect is a problem, do let me know