Thursday, 2019-06-13

*** tpb has joined #symbiflow00:00
*** futarisIRCcloud has joined #symbiflow01:37
*** proteusguy has joined #symbiflow01:58
*** Bertl is now known as Bertl_zZ03:49
*** proteusguy has quit IRC04:21
*** proteusguy has joined #symbiflow04:39
*** OmniMancer has joined #symbiflow06:58
*** auscompgeek has left #symbiflow08:30
*** _whitelogger has quit IRC08:53
*** _whitelogger has joined #symbiflow08:56
mithroduck2: ping?09:54
*** proteusguy has quit IRC10:03
*** ZipCPU has quit IRC11:06
*** ZipCPU has joined #symbiflow12:42
duck2mithro: hello12:42
mithroduck2: When you have time, happy to discuss "I don't know what to do about the discrepancy between the XSD schema and VPR's internal representation." issue12:45
mithroduck2: Did you see my comment about lowering the single element structures too?12:46
*** Bertl_zZ is now known as Bertl12:50
mithromkurc: I was thinking that for we create a separate git repo which generates the verilog and then just add the verilog to prjxray?12:59
tpbTitle: Litex SoC minitest by mkurc-ant · Pull Request #881 · SymbiFlow/prjxray · GitHub (at
sf-slack2<mkurc> @mithro But the point there was in having bitstreams that we can process with bit2fasm and see what bits we are missing. So we would need bitstream in the repo13:04
sf-slack2<mkurc> @mithro And the LiteX has that neat feature that it takes care of invoking the bitstream generation tools13:04
mithromkurc: I'm suggesting that we don't add deps on the litex+migen into prjxray -- instead we just generate the verilog + shell script once and then commit that13:05
sf-slack2<mkurc> @mitrho That makes sense. However generating all the stuff was suggested by @litghost. I think it would be good to consult with him on that matter. I can do that later today.13:10
sf-slack2<tmichalak> @mkurc: litghost mentioned he won't be around Thursday, Friday I think13:12
sf-slack2<tmichalak> @mkurc: I also don't think he specified the structure of the minitest in such detail, we do need two bitstreams, one synhtesized with the yosys frontend and another with vivado along with respective fasm files13:16
sf-slack2<mkurc> @tmichalak: Anyways, the solution by @mitrho seems reasonable, we can proceed with it.13:17
sf-slack2<tmichalak> @mkurc: exactly my point, let's proceed with the verilog + shell script approach13:19
*** proteusguy has joined #symbiflow13:21
mithromkurc: I think that is what we did for picosoc / muraxsoc?13:35
sf-slack2<mkurc> @mitrho: The picosoc was verilog source code which needed some tweaking. I can't tell about the murax.13:36
mithromkurc: Murax is generated from SpinalHDL using jvm14:08
mithromkurc: I mean, I love LiteX -- built I'm worried about prjxray growing too many dependencies14:09
sf-slack2<mkurc> @mithro I understand your concern. I'll add sources and build scripts.14:10
*** alexhw has quit IRC14:10
*** alexhw has joined #symbiflow14:21
sf-slack2<mkurc> @mithro I added sources generated by LiteX together with Makefiles. This time the only new external dependency is Yosys.
tpbTitle: Litex SoC minitest (sources) by mkurc-ant · Pull Request #883 · SymbiFlow/prjxray · GitHub (at
*** OmniMancer has quit IRC15:14
*** Bertl is now known as Bertl_oO15:19
duck2mithro: I thought about the issue and I think it's just fine generating a struct like the current one. After all, XSD is supposed to only define the form of the document. The post-processor will still be there, but it won't need strcmps to fill in enums, vectors can just be copied over etc.21:51
duck2mithro: yep, it's a good idea to emit the single-element structs as typedefs, thanks21:52

Generated by 2.13.1 by Marius Gedminas - find it at!