<FelixVi>
oh wow, that is quite different from misoc
<cr1901_modern>
FelixVi: Well, I was intending to make my example simpler, but no time to prepare it b/c I'm multiplexing a number of life events. Badly.
<FelixVi>
cr1901_modern: I'll look around there
<FelixVi>
cr1901_modern: My feeling is that litex might work while misoc is broken
<FelixVi>
That would explain why you were saying it works all along - and I think litex probably works
<cr1901_modern>
FelixVi: Litex isn't that different from misoc in that respect (well anymore)
<cr1901_modern>
So I would _not_ go as far as to say misoc is broken while litex works
<FelixVi>
cr1901_modern: Let me test that out first
<FelixVi>
I'll probably get to play with it a some more tomorrow
<FelixVi>
cr1901_modern: The DDR issue I had with one of the platforms was due to a faulty board btw
<cr1901_modern>
The Saturn?
<cr1901_modern>
And that's good :D... for some metric of good
<FelixVi>
cr1901_modern: Yeah, it was a lot easier with another board - turned out one of the address pin was shot on the board I have at home
<cr1901_modern>
So you should be able to make a PR for migen/misoc then for that board
<cr1901_modern>
It's a good starter board for lv45
<cr1901_modern>
which is a medium-size FPGA within hobbyist reach
<FelixVi>
Yeah, I think so - still haven't tested things out
<mithro>
FelixVi: heyo
<FelixVi>
It's hard to find LX75 boards, but that's the biggest webpack does anyways
<FelixVi>
mithro: hey - sorry I was ignorant ;) should have tried litex from the beginning... I know you told me so
<cr1901_modern>
mithro: Why do you think misoc/litex sufficiently diverge?
<mithro>
FelixVi: My eventual goal is to get less divergence between misoc/litex
<cr1901_modern>
litex afaict is mostly a superset of misoc w/ different paths
<FelixVi>
cr1901_modern: I don't know for sure - but if litex recently worked on S6 it appears to be different
<cr1901_modern>
err pkg hierarchy*
<FelixVi>
I can boot from BRAM and execute code from BRAM
<FelixVi>
but just no luck with flash
<FelixVi>
so I'll look if that's where litex and misoc might have differences
<cr1901_modern>
mithro: Also, something I forgot to mention. Still working on TinyFPGA (and balancing life poorly). See DM
<mithro>
FelixVi: we don't actually execute stuff from flash most of the time -- the BIOS copies the stuff from flash into DDR and then we execute from there
<FelixVi>
mithro: Yeah, but I couldn't get a bios prompt from flash
<mithro>
FelixVi: We just run the BIOS from BRAM on everything apart from the MimasV2
<FelixVi>
maybe I am confused, but I think you should be able to put bios and executable code into flash and then execute from DDR
<FelixVi>
mithro: Oh, that is good to know - I made a bios version that is much smaller to preserve BRAM
<FelixVi>
but I figured ideally everything would live in flash
<FelixVi>
So you typically put the bios into BRAM and executables into flash memory?
<cr1901_modern>
FelixVi: You don't want that unless you absolutely have to
<cr1901_modern>
And if you absolutely have to, you _really_ need an icache, otherwise your core is going to be unacceptably slow most likely
<cr1901_modern>
(like 1/32th speed. "You'll never keep the pipeline filled" slow)
<mithro>
FelixVi: What board are you working on? The AC701 should have *plenty* of resources
<FelixVi>
I'm on a XC6SLX45, so it should be fine
<FelixVi>
so, bios in bram and you load code into ddr is the preferred way?
<FelixVi>
I'll try it that way
<FelixVi>
mithro: are the timvideos repositories the stable ones and mithro is for dev?
<mithro>
FelixVi: pretty muxh
<FelixVi>
mithro: Cool, so I'll play with the stable ones
rohitksingh_work has joined #m-labs
rohitksingh_work has quit [Client Quit]
<rjo1>
sb0, mntng: are you using the fmc-lvds-32ch-dio for novogorny/zotino?
<rjo1>
whitequark: congrats to the 5000th artiq commit.
ashafir has joined #m-labs
<rjo1>
whitequark: what are the next steps on ethernet?
rjo1 is now known as rjo
<rjo>
ashafir: hello
<rjo>
ashafir: you should put your conda installation into /srv/scratch/ashafir
<rjo>
that has plenty of space and the backup machiner doesn't get bogged down by it.
<rohitksingh>
let me try to get this sorted out myself
<rjo>
rohitksingh: what was the issue?
<rohitksingh>
rjo: Hi! I wanted to know how to generate or1kx executable from rust code. Any pointers would be very helpful!
<rjo>
rohitksingh: i am looking into it.
<rohitksingh>
rjo: Thanks! :) I'm also trying to get some way with this
<rjo>
rohitksingh: for now do: conda install -c m-labs/label/dev misoc
<rjo>
ashafir: ^^
sb0 has quit [Ping timeout: 260 seconds]
<rohitksingh>
rjo: Thanks! This one has fixed/bypassed the compile error in `artiq/firmware/runtime/ethmac.rs` for now...I'll proceed with current setup. Thank you! :)