| *** tpb has joined #yosys | 00:00 | |
| *** emeb has quit IRC | 00:28 | |
| *** X-Scale` has joined #yosys | 01:04 | |
| *** dh73 has quit IRC | 01:04 | |
| *** X-Scale has quit IRC | 01:05 | |
| *** X-Scale` is now known as X-Scale | 01:05 | |
| *** nrossi has joined #yosys | 01:09 | |
| *** kraiskil has quit IRC | 02:08 | |
| *** rohitksingh has quit IRC | 02:24 | |
| *** ktemkin has joined #yosys | 02:58 | |
| *** citypw has joined #yosys | 03:18 | |
| *** X-Scale` has joined #yosys | 03:31 | |
| *** X-Scale has quit IRC | 03:31 | |
| *** X-Scale` is now known as X-Scale | 03:32 | |
| promach3 | For http://sci-hub.tw/https://ieeexplore.ieee.org/document/4751932 , why the cyclic combinational path of wavefront arbiter could not be handled by EDA tool ? | 06:22 |
|---|---|---|
| tpb | Title: Sci-Hub: устраняя преграды на пути распространения знаний (at sci-hub.tw) | 06:22 |
| whitequark | EDA tools like Yosys and nextpnr focus on synchronous logic | 06:26 |
| promach3 | whitequark: are you familiar with structure of wavefront arbiter ? | 06:28 |
| whitequark | I am not | 06:37 |
| *** dys has quit IRC | 06:54 | |
| promach3 | https://en.wikipedia.org/wiki/Wavefront_arbiter | 06:58 |
| tpb | Title: Wavefront arbiter - Wikipedia (at en.wikipedia.org) | 06:58 |
| promach3 | I do not quite understand how https://slideplayer.com/slide/13220542/79/images/20/Wavefront+arbiter+Each+element+is+a+processing+element.jpg cannot be handled by EDA tool | 07:00 |
| whitequark | I don't understand how it works, but cyclic combinatorial paths are essentially equivalent to asynchronous logic | 07:01 |
| whitequark | a lot of existing tooling chooses not to bother with anything more complex than a latch (often latches too) | 07:01 |
| whitequark | there are a few issues here. first, LUTs usually aren't glitchless, so techmapping combinatorial cycles correctly is hard | 07:02 |
| whitequark | second, you have to somehow determine the worst case time for it settling to a fixpoint to provide timing analysis | 07:03 |
| promach3 | ok | 07:06 |
| promach3 | whitequark: maybe https://images.slideplayer.com/29/9485249/slides/slide_8.jpg helps in understanding | 07:06 |
| whitequark | sure | 07:08 |
| *** FabM has joined #yosys | 07:21 | |
| *** voxadam has quit IRC | 09:06 | |
| *** voxadam has joined #yosys | 09:07 | |
| *** dys has joined #yosys | 09:41 | |
| *** captain_morgan20 has quit IRC | 10:13 | |
| *** captain_morgan20 has joined #yosys | 10:14 | |
| lukego | whitequark: much obliged for the docs update :) I'll try to be such a good citizen too. | 10:31 |
| lukego | emily: cool :) | 10:31 |
| *** FabM is now known as FabM_cave | 10:52 | |
| *** jakobwenzel has quit IRC | 12:56 | |
| *** fsasm has joined #yosys | 13:25 | |
| *** pie_ has quit IRC | 13:52 | |
| *** pie_ has joined #yosys | 13:53 | |
| *** pie_ has quit IRC | 13:58 | |
| *** dys has quit IRC | 14:10 | |
| *** indy_ is now known as indy | 14:29 | |
| *** dys has joined #yosys | 14:33 | |
| *** dh73 has joined #yosys | 15:20 | |
| *** dys has quit IRC | 15:25 | |
| *** dys has joined #yosys | 15:33 | |
| *** citypw has quit IRC | 15:34 | |
| *** tannewt has quit IRC | 15:51 | |
| *** pie_ has joined #yosys | 16:29 | |
| *** dys has quit IRC | 16:37 | |
| daveshah | lukego: twitter thread for you haha | 17:42 |
| daveshah | https://twitter.com/jangray/status/1197628580131291137 | 17:42 |
| *** FabM_cave has quit IRC | 18:15 | |
| *** fsasm has quit IRC | 18:34 | |
| *** fsasm has joined #yosys | 18:34 | |
| *** kraiskil has joined #yosys | 18:36 | |
| *** nrossi has quit IRC | 19:07 | |
| *** elGamal has quit IRC | 20:21 | |
| *** Jybz has joined #yosys | 20:22 | |
| *** elGamal has joined #yosys | 20:23 | |
| *** Jybz has quit IRC | 20:37 | |
| *** Jybz has joined #yosys | 20:42 | |
| *** Jybz has quit IRC | 20:49 | |
| *** dh73 has quit IRC | 21:41 | |
| *** dh73 has joined #yosys | 21:45 | |
| *** bobzoidting has joined #yosys | 21:49 | |
| *** Stary has quit IRC | 22:14 | |
| hackerfoo | Does Yosys perform any numerical analysis e.g. optimize out any bits higher than the low N+1 of the result of adding two N bit inputs? | 22:35 |
| *** Stary has joined #yosys | 22:38 | |
| mwk | yes | 23:07 |
| mwk | and that particular case is definitely handled | 23:07 |
| *** bobzoidting has quit IRC | 23:07 | |
| daveshah | Several more complex cases aren't though | 23:07 |
| daveshah | eg a+b-b may well not be optimised away | 23:08 |
| mwk | (it's not) | 23:09 |
| daveshah | I think abc9 will usually work it out as it can see through the carry chains, in practice | 23:09 |
| daveshah | Old abc won't if it is mapped to carries though | 23:09 |
| mwk | hmm | 23:10 |
| mwk | it doesn't optimize even for plain "synth" to gates | 23:11 |
| daveshah | synth uses some weird abc script that disables a lot of optimisations | 23:11 |
| daveshah | I think if you do a normal "abc" after synth it optimises | 23:11 |
| mwk | oh, yeah | 23:12 |
| *** fsasm has quit IRC | 23:16 | |
Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!
