<kc5tja>
hmm...I genuinely don't know what to do about this failure message that _whitenotifier-c is producing. Even after merging with upstream master, I'm still getting a net reduction in code coverage. But, all I'm touching is documentation.
<kc5tja>
Anyone know what's going on and how to fix this?
<kc5tja>
whitequark: Also, related, I've updated my most recent PR. If you could perhaps give another doc review at your leisure?
kc5tja has quit [Quit: bed-time]
rohitksingh_work has quit [Ping timeout: 250 seconds]
<hartytp>
sb0: in the Thermostat quote, why did you suggest float32/int64 as data types for the numerics and not float64?
<hartytp>
is int64 easier to implement in embedded rust than double-precision floats?
<whitequark>
there's only softfloat and double-precision softfloat is slower than single-precision
<whitequark>
rust has i32 and i64, and just like them, f32 and f64
<hartytp>
ack
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
<hartytp>
but, for a control loop that runs at 5Hz and only needs to do some basic maths (maybe a log, but that is over-kill really) that speed won't be an issue at all will it?
<hartytp>
sb0: need to tidy up my thermostat sims then looking at getting sayma in a rack
<hartytp>
if you have any advice on that one let me know!
<whitequark>
for code that runs at 5 Hz f64 will work just fine
<hartytp>
out of curiosity, how long does it take to calculate a log in soft doubles on a micro like the one in ion pak?
Xark has quit [Ping timeout: 246 seconds]
<hartytp>
ms?
Xark has joined #m-labs
<whitequark>
ionpak has hardfloat
<whitequark>
oh wait, i assumed we were talking about the softcpu here
<hartytp>
sorry, no, this is for a temperature controller using the same micro as ion pak
<whitequark>
so you have 32-bit hardfloats and 64-bit softfloats
<hartytp>
yes, but my assumption is that a simple control loop uising double precision maths will take negligible time compared to the 200ms ADC sample rate
<hartytp>
just wanted to double check the orders of magnitude involved.
<whitequark>
well for one this depends on the frequency you're running it at
<whitequark>
if you're running it at all 120 MHz it's definitely fine
<hartytp>
thanks
<hartytp>
64 bit ints should be fine too, but makes the maths a little uglier
hartytp has quit [Quit: Page closed]
<sb0>
I added DDMTD cores to various transceiver clock pins to debug Sayma
<sb0>
one interesting thing is, the phase between REFCLK and TXCLKOUT is random, as I feared from trying to understand the docs
<sb0>
right now this gets absorbed into siphaser, but it may break the WR PLL if not designed taking this into account
<sb0>
this also affects DRTIO masters, i.e. everything is referenced to TXCLKOUT (which is connected to the FPGA internal RTIO clock), not the REFCLK pin
<sb0>
those things are really frustrating, I don't see why they would not reference the phase aligner to REFCLK instead of the PMA divider output
<sb0>
sigh
<sb0>
well i guess you can put this thing into manual mode and do the alignment yourself, if needed
sb0 has quit [Ping timeout: 245 seconds]
sb0 has joined #m-labs
<sb0>
_florent_: re. this problem, are the lanes properly aligned in multi-channel mode?
<sb0>
well nothing seems unusual with the transceiver clock phases when the 180-degree bug kicks in :/
proteusguy has quit [Remote host closed the connection]
cr1901_modern1 has quit [Quit: Leaving.]
cr1901_modern has joined #m-labs
<d_n|a_>
sb0/whitequark: The aiohttp conda package on the m-labs channel is too old for artiq_influxdb (0.17; there have been API changes since). How to go about fixing this?
<keesj>
is conda required if I already install lm32 linker/toolchain when using litex
<keesj>
I was able to use migen for misoc without conda
rohitksingh_work has quit [Read error: Connection reset by peer]
<d_n|a_>
Conda isn't required by any of the tools, beyond providing a means of installing them
d_n|a_ is now known as d_n|a
sb000 has joined #m-labs
<sb000>
dna: are you using Linux or windows?
<sb000>
it should be in the conda-recipes repository, but the problem with conda is you can very easily break other things or trigger bugs when touching any package
<sb000>
if you need a quick fix and are running Linux I suggest you use nix instead
<sb000>
(m-labs conda-recipes)
<sb000>
also check if someone else packaged it in the meantime
sb000 has quit [Ping timeout: 256 seconds]
hartytp has joined #m-labs
<hartytp>
sb0: how easy did you find it to get your Sayma AMC into the rack
<hartytp>
mine took a lot of insertion force (I'm not confident that it's a reversible operation)
<hartytp>
not sure if that was because of damage that board has suffered in its travels around the world, or if there is a mechanical issue here
rohitksingh has joined #m-labs
sb000 has joined #m-labs
<sb000>
probably a mechanical issue
<sb000>
it wasn't particularly hard to fit
<sb000>
I don't know if you can really blame problems on boards being shipped around
<hartytp>
well, now it's in now. not sure it's ever coming out in once piece, but it's in
<sb000>
besides the broken clock input sma (due mostly to cables pulling it afaict) and the broken rtm transistors (design issue), is there clearly identified damage? d
<hartytp>
none that I saw
<hartytp>
well, the fp was clearly a bit bent but other than that it looked ok
<hartytp>
the RTM went in fine
<sb000>
what was the problem exactly?
<hartytp>
sb000 where were your notes on getting power in
<sb000>
you just need to make sure you have the latest firmware
<hartytp>
just took a silly amount of insertion force. I think the issue was getting it to mate with the amc connector
<sb000>
the inrush current limiter
<hartytp>
latest MMC firmware?
<hartytp>
or MCH?
<sb000>
that's all
<sb000>
and the shorted rtm power transistors
<sb000>
yes
<sb000>
mmc
<hartytp>
how do I check the version of that?
<sb000>
i haven't seen actual mch firmware issues, besides general crappy design
<hartytp>
ok, how do I check the MMC firmware rev?
sb000 has quit [Ping timeout: 256 seconds]
<sb0>
hartytp: you'd have to ask Greg, but afaict there is no version string or anything
rohitksingh has quit [Ping timeout: 272 seconds]
proteusguy has joined #m-labs
rohitksingh has joined #m-labs
<sb0>
d_n|a: ^
jevinskie has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<_whitenotifier-c>
[nmigen] jfng opened issue #35: Unassigned signals do not assume their reset values - https://git.io/fh9G5
<d_n|a>
sb0: Thanks - aiohttp exists on both `conda-forge` and `anaconda`, but there is an ancient version in the m-labs channel
<sb0>
d_n|a: okay, I can try putting that ancient version to a "trashbin" label - then let me know if things still work...
<sb0>
well there's already a "obsolete" label on it
proteusguy has quit [Remote host closed the connection]
proteusguy has joined #m-labs
kc5tja has joined #m-labs
proteusguy has quit [Remote host closed the connection]
<kc5tja>
Greetings, and apologies for lengthy absence. Holidays and such. I've pushed the latest version of the hdl.rst file in the docs, and was wondering if someone can review it and offer guidance on how further to proceed.
<kc5tja>
My plan is to next work on the chapter on simulation, but to basically stop there for this PR. Under a separate PR (to make reviews easier), I would include the API reference (if I can easily generate one) or other sections of the documentation.
<kc5tja>
Thoughts?
proteusguy has joined #m-labs
massi_ has joined #m-labs
proteusguy has quit [Remote host closed the connection]
proteusguy has joined #m-labs
kc5tja has quit [Quit: off to work]
rohitksingh has quit [Ping timeout: 268 seconds]
wolfspraul has quit [Quit: leaving]
wolfspraul has joined #m-labs
wolfspraul has quit [Client Quit]
wolfspraul has joined #m-labs
rohitksingh has joined #m-labs
proteusguy has quit [Remote host closed the connection]
proteusguy has joined #m-labs
rohitksingh has quit [Remote host closed the connection]
<_whitenotifier-c>
[nmigen] whitequark commented on issue #35: Unassigned signals do not assume their reset values - https://git.io/fh9le
<_whitenotifier-c>
[nmigen] whitequark closed issue #35: Unassigned signals do not assume their reset values - https://git.io/fh9G5
proteusguy has quit [Remote host closed the connection]
proteusguy has joined #m-labs
massi_ has quit [Remote host closed the connection]