sb0 changed the topic of #m-labs to: ARTIQ, Migen, MiSoC, Mixxeo & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
_whitelogger has joined #m-labs
<cr1901_modern> sb0: Did you get back to olofk or vice versa re: fusesoc? We were discussing integration w/ Migen last night and I think we came up w/ a decent idea (Migen user imports fusesoc, fusesoc returns metadata on the verilog project, migen uses metadata and auto-creates an instance)
<sb0> cr1901_modern, ok
<cr1901_modern> sb0: Wait, you agree :o? This implies adding a parser for fusesoc metadata to migen (at least that's how olofk) wants to do it. Not difficult, but... it is extra code/bloat.
<sb0> why does this need to be inside migen?
<cr1901_modern> B/c of how olofk has been refactoring fusesoc: https://gitter.im/librecores/fusesoc?at=59ca32c5f7299e8f53731588
<cr1901_modern> sb0: (Of course there are other ways to do this. I just asked you for feedback on the off chance you'd agree :P)
<sb0> what is the end goal of this, exactly?
<sb0> and I still don't see why this can't be put into a package separate from migen and fusesoc
<sb0> also if fusesoc is a package manager/build tool, it would make sense to do things the other way around, i.e. fusesoc calls migen
<cr1901_modern> sb0: Goal is automate migen being able to build designs which rely on external verilog code using a common package manager (fusesoc).
<cr1901_modern> I mean, I think migen's build process is superior for reasons
<cr1901_modern> but yes that also works
<sb0> I don't particularly like the migen build process itself, it's more that you can have all sorts of tweaks using python
<sb0> for a multi-dependency project, something like cargo would be nice
<sb0> but ok. the problem you are trying to solve is: you have a project which is built using python in the "migen/misoc" way, and that project uses some fusesoc-wrapped cores, and you want those to be downloaded and instantiated automatically?
<cr1901_modern> Yes.
<sb0> okay, well I think having that "glue" code in a separate repository is doable
<cr1901_modern> Okay cool, I'll relay that back then...
<sb0> and you can get started that way, if popular, clean and well-maintained then it can be moved into a folder in migen ..,
<cr1901_modern> Fair enough
<cr1901_modern> Just so I know, what is wrong w/ the migen build process in your eyes, and what behavior would you in theory prefer?
<sb0> it's not much of a build process, it's just a python script that does things
<sb0> and also the code in migen/build is messy and ad-hoc
<sb0> the platform api is also not super nice
<cr1901_modern> Really? I mean, I almost never want to touch Verilog when I can get migen to make me a top level (even for Verilog modules I'm just routing platform signals thru)
<cr1901_modern> sb0: "for a multi-dependency project, something like cargo would be nice" <-- you may wish to look into Meson, which has FPGA support in progress. I could just like migen.build b/c I'm used to it. But >>
<cr1901_modern> I _do_ find the platform api valuable b/c you can write "universal top levels", dependent only on peripherals and adding a board to the platform api
<cr1901_modern> ^That's why I did the blog post in the first place
<cr1901_modern> Code reuse/portability is good (TM)
<cr1901_modern> I don't have great social media reach, but I do hope more ppl use migen and port more boards :P
<sb0> sure, it works, but it's a bit dirty
<cr1901_modern> sb0: Yes, agreed; in at least one case, I've had to resort to fragments and do_finalize() to get a "universal top level" and I don't like it at all. But at present, it was still less work than anything that required me to write Verilog. Maybe I'll think of a better solution in time.
<cr1901_modern> sb0: In your experience, have you seen a lot (i.e. more than one) migen user who creates standalone Modules and creates a repo for it (say, for example, a standalone non-MiSoC UART meant to be used in other code)?
<cr1901_modern> If so, I think I can see why you want something like cargo...
rohitksingh_work has joined #m-labs
<sb0> cr1901_modern, no
<sb0> but having something like cargo would be beneficial to python in general and not just migen projects
<sb0> the current package managers are very crappy
sb0 has quit [Quit: Leaving]
bb-m-labs_ has joined #m-labs
rjo1 has joined #m-labs
rjo has quit [*.net *.split]
bb-m-labs has quit [*.net *.split]
rohitksingh_work has quit [Ping timeout: 260 seconds]
rohitksingh_work has joined #m-labs
sb0 has joined #m-labs
<sb0> rjo1, boards rebooted and connected to relay
<sb0> relay is on /dev/ttyACM0
<rjo1> sb0: ack
<rjo1> sb0: i guess R253 is not there on the saymas you have?
rjo1 is now known as rjo
<sb0> rjo, where is it located on the pcb?
<sb0> do you already know?
sb0 has quit [Quit: Leaving]
_whitelogger has joined #m-labs
<rjo> sb0: my guess would be the left one of the two above the "T27", underside top left.
rohitksingh_work has quit [Read error: Connection reset by peer]
llopez has joined #m-labs
llopez has quit [Ping timeout: 260 seconds]
llopez has joined #m-labs
<rjo> _florent_: about getting sines out of the dacs...
<rjo> _florent_: i am integrating jesd into current artiq but that may take a while to be useful to greg's allaki testing because of ethernet etc.
<rjo> _florent_: we could just strap a BRAM or a cordic into a channel in your sayma_test repo.
<rjo> _florent_: what's your take?
<rjo> _florent_: and is the migen jesd204b repo up to date with your litex jesd204b?
<GitHub> [conda-recipes] jordens pushed 1 new commit to master: https://github.com/m-labs/conda-recipes/commit/418fd58498d0a95a0cc1a7f843d694eb48953975
<GitHub> conda-recipes/master 418fd58 Robert Jordens: jesd204b 0.4
<rjo> bb-m-labs_: force build --props=package=jesd204b conda-lin64
<bb-m-labs_> build forced [ETA 10m53s]
<bb-m-labs_> I'll give a shout when the build finishes
<bb-m-labs_> build #326 of conda-lin64 is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/326
<rjo> bb-m-labs_: force build --props=package=jesd204b conda-lin64
<bb-m-labs_> build forced [ETA 10m53s]
<bb-m-labs_> I'll give a shout when the build finishes
<bb-m-labs_> build #327 of conda-lin64 is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/conda-lin64/builds/327
sb0 has joined #m-labs
mumptai has joined #m-labs
<whitequark> sb0: rjo: re #830
<whitequark> I'm being stupid, but I can't figure a way to turn a remainder operator into a modulo one
<rjo> ;)
<rjo> whitequark: how is remainder defined?
<rjo> whitequark: is that an llvm op?
<sb0> whitequark, if n < 0: d-remainder(-n, d)
<sb0> python also defines negative d ...
<rjo> yes. i'd consider deviating from python if the IR/assembly becomes troublesome.
<sb0> d-remainder(-n,d) also fails if n is a multiple of d
<sb0> hm yes it's annoying
<GitHub1> [migen] jordens pushed 1 new commit to master: https://git.io/vdmua
<GitHub1> migen/master 5a175ec Robert Jordens: sayma_amc: add dac jesd pins
<GitHub65> [artiq] jordens pushed 4 new commits to sinara: https://github.com/m-labs/artiq/compare/2091c7696a24...38388d113610
<GitHub65> artiq/sinara 2604806 Robert Jordens: sayma_rtm: make build dir
<GitHub65> artiq/sinara 2821f50 Robert Jordens: conda: jesd204b 0.4
<GitHub65> artiq/sinara 5e3cc83 Robert Jordens: sayma_amc: SAWG (untested)
<bb-m-labs_> build #189 of migen is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/migen/builds/189
<whitequark> rjo: deviating from python seems very bad
<whitequark> i can do whatever math you tell me to do
<whitequark> just, please, don't make me think about it
<whitequark> llvm op, yes
<whitequark> no smod or anything
<whitequark> not sure why
<sb0> what does the cpu have in hardware?
kristianpaul has quit [Quit: Lost terminal]
<sb0> whitequark, surely there are other languages with a modulo operator like python, and they have the proper way to do that with llvm ir?
<whitequark> ok, I will look at it
<sb0> what does numba do?
<rjo> sb0: i can't get the second flash chip to work on any of the three AMCs. could you check whether they are wired like florent's? the photos i uploaded are from florent's.
halfr has quit [Ping timeout: 255 seconds]
halfr has joined #m-labs
kristianpaul has joined #m-labs
<travis-ci> batonius/smoltcp#92 (default_route - 07f0431 : Egor Karavaev): The build passed.
<travis-ci> batonius/smoltcp#93 (default_route - 0412399 : Egor Karavaev): The build passed.
<GitHub108> [smoltcp] batonius commented on issue #44: Added `FromStr` implementation for `IpCidr`. https://git.io/vdmQw
mumptai has quit [Remote host closed the connection]
<_florent_> rjo: no problem to integrate bram or cordic into sayma_test if this can speed up things, just provide me the data generator and I'll prepare something
<_florent_> rjo: I'll also integrate the last jesd changes in jesd repo