Monday, 2019-07-01

*** tpb has joined #yosys00:00
*** cr1901_modern has joined #yosys00:12
*** brasilino has joined #yosys00:14
*** gsi__ has joined #yosys00:44
*** gsi_ has quit IRC00:47
*** brasilino has quit IRC01:57
*** lutsabound has quit IRC01:57
*** vonnieda has quit IRC02:23
*** PyroPeter has quit IRC02:42
*** PyroPeter has joined #yosys02:55
*** s_frit has quit IRC03:04
*** s_frit has joined #yosys03:05
*** citypw has joined #yosys03:23
*** citypw has quit IRC03:29
*** citypw has joined #yosys03:41
*** vonnieda has joined #yosys03:43
*** dys has quit IRC03:44
*** dys has joined #yosys04:15
*** rohitksingh_work has joined #yosys05:11
*** rohitksingh_work has quit IRC05:40
*** proteusguy has quit IRC05:46
*** dys has quit IRC05:51
*** proteusguy has joined #yosys06:24
*** Nazara has quit IRC06:24
*** Nazara has joined #yosys06:28
*** Nazara has quit IRC06:32
*** rohitksingh has joined #yosys06:34
*** Nazara has joined #yosys06:47
*** Nazara has quit IRC06:48
*** Nazara has joined #yosys06:49
*** citypw has quit IRC06:49
*** citypw has joined #yosys07:02
*** Nazara has quit IRC07:03
*** Nazara has joined #yosys07:05
*** Nazara has joined #yosys07:09
*** jakobwenzel has joined #yosys07:11
*** emeb_mac has quit IRC07:32
*** kraiskil has joined #yosys07:33
*** pie_ has quit IRC07:59
*** pie__ has joined #yosys07:59
*** rohitksingh has quit IRC08:42
*** rohitksingh has joined #yosys09:19
*** citypw has quit IRC10:16
*** FabM has joined #yosys10:19
*** _whitelogger has quit IRC10:32
*** _whitelogger has joined #yosys10:35
*** rohitksingh has quit IRC11:32
*** rohitksingh has joined #yosys11:40
*** rohitksingh has quit IRC12:27
*** rrika has quit IRC12:33
*** rrika has joined #yosys12:35
*** dys has joined #yosys12:35
*** m4ssi has joined #yosys12:55
*** rohitksingh has joined #yosys13:09
*** celadon has joined #yosys13:15
*** proteusguy has quit IRC13:16
*** s_frit has quit IRC13:16
*** s_frit has joined #yosys13:17
*** rohitksingh has quit IRC13:22
*** rohitksingh has joined #yosys13:49
*** vonnieda has quit IRC13:55
*** kristianpaul has quit IRC14:06
*** vonnieda has joined #yosys14:11
*** kristianpaul has joined #yosys14:13
*** brasilino has joined #yosys14:19
*** kristianpaul has quit IRC14:28
*** Cerpin has joined #yosys14:29
*** Cerpin has quit IRC14:40
*** kristianpaul has joined #yosys14:47
*** rohitksingh has quit IRC14:49
*** rohitksingh has joined #yosys14:49
*** maikmerten has joined #yosys14:54
*** celadon has quit IRC15:34
*** FabM has quit IRC15:42
*** celadon has joined #yosys15:52
*** emeb has joined #yosys15:54
*** m4ssi has quit IRC16:10
somlodaveshah: I think I'm either doing something wrong, or hitting some sort of regression; with the latest yosys/trellis/nextpnr I built over the weekend, the bitstream I'm sending to my ecp5 5g versa just doesn't work16:19
somloso I'm going to go back and build the versions I had from mid-May, and try to work my way back up to date from there16:19
somlobut wanted to check with you in case this rings a bell... :)16:20
*** gsi__ is now known as gsi_16:33
*** kraiskil has quit IRC16:33
*** dys has quit IRC17:03
*** vonnieda has quit IRC17:16
*** vonnieda has joined #yosys17:22
*** dys has joined #yosys17:24
somlodaveshah: got lucky, only had to downgrade yosys (to the prevous snapshot I had used, c907899) and my stuff started working again18:11
somlodidn't have to go back to May with trellis and nextpnr, whew! :)18:12
somlois yosys bisect-friendly? I could probably try that if I'm not likely to land on a bunch of intermediate commits where I can't test because of breakage unrelated to what I'm hunting for...18:12
daveshahsomlo: There are two possible causes.18:22
daveshahThe first is https://github.com/YosysHQ/yosys/commit/a0d3d2bb41cd11b5b33620e16cb67c6568b16847 which should be easy to revert18:22
tpbTitle: ecp5: Improve mapping of $alu when BI is used · YosysHQ/[email protected] · GitHub (at github.com)18:22
daveshahThe second is the ABC9 changes, which I wouldn't attempt to revert but instead try the 0.9 release candidate branch which excludes them18:23
daveshahIf neither of those are the problem them bisection would be needed18:23
daveshahI'm not sure how bisect will go with the big ABC9 PR18:23
somlothat's what I was wondering about, chances of landing on random commits where I'll end up encountering unrelated trouble to what I'm hunting for :)18:24
benreynwarZipCPU: Is there a way to add an 'assume' into my VHDL code?  Currently I'm putting them in the top SV wrapper, but it'd be nice to be able to able to put them in the design files too.18:24
somloso, low hanging fruit first -- $alu mapping, then go from there :)18:25
somloI'll report back on how that went...18:25
ZipCPUHello, benreynwar!18:26
ZipCPUThere is no way to place either assertion nor assumption into any VHDL code.18:26
benreynwarUsing standard VHDL assertions seems to work, but I hadn't found a way to use assumptions yet.18:27
ZipCPUOk, assertions work ... I had a faint memory of one working while the other did not ...18:28
benreynwarZipCPU: I kind of figured it wasn't possible, but thought I should ask before going to the effort of making a workaround.  Thanks!18:30
ZipCPUThat's why, in the VHDL course I teach, we do all the formal verification work itself in Verilog18:31
benreynwarI was thinking of replacing submodules that have been verified already with stubs that just contain assumptions corresponding to the verified assertions.  It would have been nice to be able to write those stubs in VHDL since the SV/VHDL interfaces get messy.  I'll probably just convert the assumptions into output bits that get sent to the top-level SV and then write the assumption there.18:42
ZipCPUThat'd work19:08
*** cr1901 has joined #yosys19:10
*** cr1901 has quit IRC19:10
*** kraiskil has joined #yosys19:21
*** fsasm has joined #yosys19:23
*** kraiskil has quit IRC20:06
*** maikmerten has quit IRC20:58
*** rohitksingh has quit IRC21:19
*** fsasm has quit IRC21:33
daveshahbenreynwar: I think we looked into the possibility of setting an attribute on a VHDL assert to turn it into an assume for formal, but hit a limitation somewhere22:20
daveshahThis is still something that I hope we'll get working at some point, it should preserve correct simulation behaviour for assumes too22:21
daveshahHierarchical references into the VHDL from SV should work, if you want to avoid adding outputs22:22
benreynwardaveshah:  That's a good idea! Thanks.22:26
*** vonnieda has quit IRC23:02

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!