| *** tpb has joined #timvideos | 00:00 | |
| -_whitenotifier-3- [yosys] pull[bot] opened pull request #542: [pull] master from YosysHQ:master - https://git.io/JUmmC | 01:42 | |
| *** GNUmoon has quit IRC | 03:32 | |
| *** GNUmoon has joined #timvideos | 03:34 | |
| -_whitenotifier-3- [yosys] pull[bot] closed pull request #542: [pull] master from YosysHQ:master - https://git.io/JUmmC | 04:42 | |
| *** GNUmoon has quit IRC | 07:29 | |
| *** GNUmoon has joined #timvideos | 08:16 | |
| *** shenki has quit IRC | 11:05 | |
| *** shenki has joined #timvideos | 11:30 | |
| *** ChanServ sets mode: +v shenki | 11:30 | |
| *** cr1901_modern has quit IRC | 14:29 | |
| *** cr1901_modern has joined #timvideos | 14:31 | |
| *** FFY00 has quit IRC | 15:43 | |
| *** FFY00 has joined #timvideos | 15:43 | |
| *** FFY00 has quit IRC | 15:47 | |
| *** FFY00 has joined #timvideos | 15:47 | |
| -_whitenotifier-3- [yosys] pull[bot] opened pull request #543: [pull] master from YosysHQ:master - https://git.io/JUmND | 17:07 | |
| *** futarisIRCcloud has joined #timvideos | 17:07 | |
| *** cr1901_modern1 has joined #timvideos | 17:11 | |
| *** cr1901_modern has quit IRC | 17:12 | |
| *** CarlFK has quit IRC | 17:44 | |
| *** shorne has quit IRC | 17:44 | |
| *** titanbiscuit has quit IRC | 17:44 | |
| *** CarlFK has joined #timvideos | 17:46 | |
| *** shorne has joined #timvideos | 17:46 | |
| *** titanbiscuit has joined #timvideos | 17:46 | |
| *** kornbluth.freenode.net sets mode: +v CarlFK | 17:46 | |
| *** titanbiscuit has quit IRC | 17:46 | |
| *** titanbiscuit has joined #timvideos | 17:47 | |
| *** cr1901_modern1 has quit IRC | 19:04 | |
| *** cr1901_modern has joined #timvideos | 19:04 | |
| shorne | mithro: ok, I have it working, as you suggested earlier it would be good to have a linux-on-mor1kx project, ill also look at that | 19:17 |
|---|---|---|
| *** futarisIRCcloud has quit IRC | 19:26 | |
| mithro | shorne: Hi! | 19:31 |
| mithro | shorne: Long time no chat! | 19:31 |
| mithro | shorne: I would eventually like us to get back to having a "linux-on-litex" without a CPU in the name which supports rv32/64, ppc32/64, sh2/4 and mor1kx | 19:33 |
| mithro | shorne: But other people think that is "to complex", so it'll be a slow process | 19:36 |
| daveshah | Does seem like good idea long term | 19:38 |
| daveshah | Might as well add OpenSPARC to that list to round things off :) | 19:38 |
| mithro | Does OpenSPARC fit on anything? | 19:38 |
| mithro | I guess FPGAs have gotten really big since OpenSPRAC was a thing.... | 19:39 |
| daveshah | Definitely it would fit on UltraScale stuff | 19:39 |
| daveshah | I think I looked at the smallest core config at was 2x ECP5 ish | 19:39 |
| daveshah | so high Artix-7 territory | 19:39 |
| daveshah | but the fun multithreading etc wasn't included in that, that was many 100ks LUTs | 19:39 |
| mithro | Yeah - ASIC designs, specially ASIC super-scalar CPU designs with things like many ported register files, map very poorly to FPGAs | 19:40 |
| daveshah | Indeed, my only interest in these is as PnR/synthesis torture tests, not anything practical :D | 19:42 |
| shorne | mithro: I thought that was what we already had with litex-buildenv. There was build-linux.sh that just worked. | 19:43 |
| mithro | shorne: Yes, but litex-buildenv is showing it's age with a huge mishmash of Python, shell, Make, etc | 19:44 |
| daveshah | Did that ever work for all arches? | 19:44 |
| mithro | daveshah: It works for rv32 and mor1k | 19:44 |
| daveshah | I remember looking at it pre-linux-on-litex-vexriscv | 19:44 |
| daveshah | I see | 19:44 |
| mithro | Even did a tutorial about doing it -> https://github.com/timvideos/litex-buildenv/wiki/HowTo-LCA2018-FPGA-Miniconf | 19:45 |
| tpb | Title: HowTo LCA2018 FPGA Miniconf · timvideos/litex-buildenv Wiki · GitHub (at github.com) | 19:45 |
| cr1901_modern | I wanted to replace litex-buildenv's logic with meson, but that went by the wayside for now | 19:45 |
| mithro | The Antmicro team has updated that tutorial to use VexRISCV and Renode (rather than mor1kx and QEMU) | 19:45 |
| cr1901_modern | does lm32 still work? | 19:46 |
| daveshah | Ah nice, that was probably after I looked | 19:46 |
| shorne | Ah | 19:46 |
| mithro | daveshah: The LCA2018 FPGA Miniconf was a big driver of getting all the mor1kx Linux stuff on LiteX working | 19:46 |
| shorne | lm32 worked last I checked, but not the linux port | 19:46 |
| mithro | I'm not sure I can convince them (Antmicro) to add support for mor1kx into Renode, but I have insisted that they keep the build-linux.sh script working with mor1kx | 19:47 |
| cr1901_modern | well yea I don't expect the linux port to work, Yann hasn't done anything w/ it for a while | 19:47 |
| shorne | mithro: and it still works, I booted linux with the latest kernel patches | 19:47 |
| shorne | however, the branch used in litex-buildenv does not boot | 19:47 |
| shorne | hence I was asking if its still recommended | 19:48 |
| shorne | Ill figure out what is wrong with the linux branch used in litex-buildenv. | 19:48 |
| mithro | shorne: Lots of people still have success with litex-buildenv, but it is not everyone's cup of tea | 19:49 |
| mithro | And I'm the first to say it needs to be rewritten | 19:49 |
| daveshah | Rewritten, or work put into making LiteX itself more accessible instead? | 19:49 |
| daveshah | I would personally prefer to see more work on the latter | 19:49 |
| cr1901_modern | I would like to see all the cores use nmigen, even if only w/ the compat layer. | 19:50 |
| daveshah | Yeah, that's been a somewhat longstanding discussion | 19:51 |
| shorne | I think the purpose of litex-buildenv was to make litex more accessible | 19:51 |
| daveshah | My feeling is the best way to do that is to improve the main litex repo rather than have a separate thing, that risks causing ecosystem confusion | 19:51 |
| shorne | i.e. it installs all of the python dependencies and gets it working, but I guess it hide a lot of the complexity, so when you really want to develop in litex you may not know where to start | 19:52 |
| daveshah | There is now three-way duplication of platforms between litex, litex-boards and litex-buildenv | 19:52 |
| daveshah | That doesn't seem ideal imo | 19:52 |
| cr1901_modern | daveshah: I haven't worked on litex a lot since early 2019, but the PLAN was at the time for most litex boards to come straight from omigen | 19:53 |
| cr1901_modern | and subclassed if necessary to add litex-specific functionality | 19:53 |
| cr1901_modern | litex-boards didn't exist back then, but I think the idea's the same >> | 19:53 |
| daveshah | I think things have diverged quite a bit since then, particularly with oMigen development slowing down | 19:53 |
| cr1901_modern | litex targets have to be annotated with more info than the main repo boards | 19:54 |
| cr1901_modern | Dunno a good way to do that besides subclassing or duplication | 19:54 |
| daveshah | between buildenv and litex? or between litex and migen | 19:54 |
| cr1901_modern | between litex and migen | 19:55 |
| daveshah | I think in that case duplication isn't too evil | 19:55 |
| cr1901_modern | daveshah: Actually wait... it MIGHT'VE been between litex-buildenv and migen (sic) | 19:55 |
| daveshah | The problem I have is mostly the duplication between litex-buildenv and litex/litex-boards | 19:56 |
| cr1901_modern | I'll go through my old PRs in a bit and give you a better answer | 19:56 |
| daveshah | It feels like most of the litex-buildenv stuff should end up in litex-boards | 19:56 |
| cr1901_modern | the litex-buildenv board files defined some C constants that were required to init the board | 19:57 |
| cr1901_modern | like SPIflash number of cycles to delay | 19:57 |
| mithro | @daveshah We have a goal of excising platforms from litex-buildenv | 19:57 |
| daveshah | Great! | 19:58 |
| daveshah | That will already improve things a lot | 19:58 |
| mithro | litex-buildenv predates litex-boards | 19:58 |
| mithro | Infact litex-buildenv predates LiteX IIRC | 19:58 |
| cr1901_modern | to use an analogy, if nmigen were to organically grow such functionality, those constants would probably be provided by nmigen-soc | 19:58 |
| cr1901_modern | Yes, litex-buildenv descends from HDMI2USB build system | 19:59 |
| cr1901_modern | which in ancient times used misoc | 19:59 |
| mithro | cr1901_modern: I'm hoping that LiteSPI removes the need for a bunch of the constants | 19:59 |
| * cr1901_modern nods | 19:59 | |
| cr1901_modern | I'm just thinking out loud about how nmigen-soc would handle the same thing | 19:59 |
| shorne | ] | 20:25 |
| shorne | sorry, screen froze | 20:25 |
| -_whitenotifier-3- [yosys] pull[bot] closed pull request #543: [pull] master from YosysHQ:master - https://git.io/JUmND | 20:58 | |
| *** GNUmoon has quit IRC | 21:54 | |
| *** GNUmoon has joined #timvideos | 22:44 | |
| *** levi has joined #timvideos | 23:01 | |
| *** cr1901_modern has quit IRC | 23:25 | |
| *** cr1901_modern has joined #timvideos | 23:41 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!