lekernel changed the topic of #m-labs to: Mixxeo, Migen, MiSoC & other M-Labs projects :: fka #milkymist :: Logs http://irclog.whitequark.org/m-labs
mumptai has quit [Remote host closed the connection]
jpeg has joined #m-labs
fengling has joined #m-labs
fengling_ has joined #m-labs
fengling has quit [Ping timeout: 245 seconds]
<mithro> sb0: what would your "dream" FPGA expansion connector look like?
<GitHub46> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/Tmm_Qg
<GitHub46> artiq/master 2954c28 Sebastien Bourdeauducq: manual/writing_a_driver: add more details according to recent discussions
<sb0> mithro, for what?
<sb0> FMC isn't bad, and it's already there
<sb0> it's good enough for many applications
<mithro> sb0: FMC isn't open and quite expensive connector wise
<sb0> it isn't open?
<mithro> sb0: to get the specification you have to pay to join it
<mithro> sb0: at least that is what I've found so far
<mithro> sb0: maybe I'm wrong?
<sb0> ah, it seems to be the case
<sb0> we should pirate it then
<mithro> The altera HSMC spec seems pretty good. http://www.altera.com/literature/ds/hsmc_spec.pdf
<sb0> this is stupid
<sb0> also the fmc pinouts are shown in every devkit
<mithro> sb0: which is stupid?
<sb0> fmc
<mithro> sb0: ha, that was my opinion so far too :)
<mithro> sb0: I'm doing up an idea for a FPGA expansion interface based around PCI-Express connectors
<mithro> sb0: which is probably a fools errand - http://xkcd.com/927/
<sb0> yes. that's why we should pirate fmc.
<sb0> or rather "analyze it from existing development boards for interoperability purposes", to use more diplomatic language
<mithro> sb0: even if we do that, FMC is still quite expensive thanks to the connectors from what I can see
<mithro> sb0: so I'm looking at trying to do something which is similar but on a cheaper PCI-Express connector
<mithro> sb0: hence I wondered is that anything you have though "this is retarded that the interface doesn't include XXXX"
<sb0> PCIe will require you to use a special cut board of a certain thickness
<sb0> and there won't be off the shelf adapters like http://www.samtec.com/images/standards/debug.jpg
<sb0> and you can't mount your card on a xilinx devkit for testing
Bertl has joined #m-labs
<GitHub140> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/TrrVng
<GitHub140> artiq/master 0088f48 Sebastien Bourdeauducq: manual/writing_a_driver: logging
Bertl is now known as Bertl_zZ
mumptai has joined #m-labs
fengling__ has joined #m-labs
fengling_ has quit [Ping timeout: 240 seconds]
_florent_ has joined #m-labs
_florent_ has quit [Client Quit]
_florent_ has joined #m-labs
mumptai has quit [Remote host closed the connection]
bentley` has quit [Remote host closed the connection]
nengel has joined #m-labs
<GitHub140> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/XT4Y5A
<GitHub140> artiq/master 8904d96 Sebastien Bourdeauducq: gui: add explorer window + rough prototype of control widgets defined with glade
Bertl_zZ is now known as Bertl
<GitHub110> [artiq] sbourdeauducq pushed 1 new commit to master: http://git.io/_VsTQA
<GitHub110> artiq/master db3114f Sebastien Bourdeauducq: serve GUI and Glade file from master
sb0 has quit [Quit: Leaving]
methril has joined #m-labs
mumptai has joined #m-labs
_florent_ has quit [Ping timeout: 256 seconds]
<rjo> looks like the minimal size for ISE is below 200 MB... That is 1% of the full install size.
<larsc> I think I watched a talk recently were somebody traced all the files that were accessed during a synthesis run for his particular FPGA and then just removed everything else and he managed to end up with ~200MB
<larsc> and things still worked
<rjo> thats what i did.
_florent_ has joined #m-labs
<rjo> makes it work nicely on travis-ci: https://travis-ci.org/nist-ionstorage/artiq
<larsc> but you weren't giving the talk, were you?
_florent_ has quit [Ping timeout: 276 seconds]
<rjo> i don't think so. that might have been mithro who is doing that as well.
<rjo> but he is using a custom box to do the build IIRC.
<larsc> also using travis
<rjo> travis only as trigger last i heard.
<rjo> the actual build was somewhere else.
<ysionneau> rjo: hi! about conda, Do you know in the meta.yaml file, when listing the build/run time dependencies, if I can say "I need this package, BUT from this source" instead of just "I need this package"
<ysionneau> for instance, we need special gbulb and pycairo3 versions, with our m-labs patches (on github) so I did my packaging of those on my binstar
<ysionneau> for now I had to name the packages "<name>-artiq" so that there is no conflict. eg. gbulb-artiq
<ysionneau> but it would be nice if I could just name it gbulb and in the artiq meta.yaml say "gbulb from m-labs binstar repository"
<ysionneau> or maybe I need to play with the "feature" notion, but I didn't get how it works
<rjo> ysionneau: oh btw. there are a few patched repos in nist-ionstorage for those.
<rjo> ysionneau: with these "channels"? or with binstar?
<ysionneau> I would like to specify the channel in the meta.yaml yes
<rjo> ysionneau: i don't know exactly. but if it understands the pip-style requirements, then just like https://github.com/nist-ionstorage/artiq/blob/ci/requirements.txt
<ysionneau> I would like to say "use gbulb from https://binstar.org/fallen"
<ysionneau> that's not an issue for now because there is no gbulb package at all in anaconda distrib or in binstar for the moment
<ysionneau> so there is no alias problem
<ysionneau> but it might happen
<ysionneau> (for now my own is the only one)
<ysionneau> ah good you added the cairo dependency in cairoplot3
<rjo> from the looks of it i would guess that "fallen/gbulb" in meta.yaml might work because that is how you resolve it with conda.
<ysionneau> I thought about that, but did not test yet
<ysionneau> will do :)
<rjo> or gbulb [artiq]"
<rjo> as a feature.
<rjo> or through the version. gbulb 1.2.3-artiq
<rjo> but looks klike you are ahead on this anyway ;)
<ysionneau> I just fixed by putting a -artiq in the package name
<ysionneau> but I could use the version string yes good idea also
<ysionneau> I still don't understand the "feature" stuff though
<rjo> i think this might be needed for artiq itself as well. artiq [gui] vs artiq [development] vs artiq [controller]
<rjo> i only know this from setup.py style feature-requirements.
<rjo> and i guess it's similar with conda.
<ysionneau> ah maybe ok
<rjo> how does conda feel?
<ysionneau> feels really nice :)
<ysionneau> was really cool to use so far
<rjo> just linux so far?
<ysionneau> except when I realized "shit I need to package this which does not exist" ... and then "oh shit, this new package I just did needs those stuff which are not packaged either" etc :p
<ysionneau> rjo: oh , no, good finding! I made my own for llvmlite but I can maybe improve mine by looking at those :)
<rjo> ysionneau: welcome to the world of distributions. i remember packaging ardour for debian 13 years ago...
<rjo> ysionneau: they look like they work on windows.
<ysionneau> hum I just realized I forgot to publish my recipe for llvmlite-or1k
<ysionneau> rjo: good point, because so far I only built conda packages for linux
<ysionneau> I need to try conda in windows
<ysionneau> 21:08 < rjo> just linux so far? < yes, I only tried Linux so far
<ysionneau> rjo: about https://github.com/nist-ionstorage/cairoplot3/commit/363b419f0ad696f9449fb6e45de2331258e59318 , isn't the dependency of cairoplot3 "pycairo"?
<ysionneau> I mean you also need libcairo in your system (and the -dev package to build pycairo) but I think the python binding package is pycairo, right?
<rjo> yeah. might be broken. i ripped it out in the actual artiq build again because there are (a) not gui tests yet and (b) they will require a bit of re-jiggering anyway.
<ysionneau> ok, I'll try it out anyway :)
<larsc> lowrisc even got a FOSDEM main-track presentation
cfelton has quit [Remote host closed the connection]
mithro has quit [Remote host closed the connection]
mithro has joined #m-labs
cfelton has joined #m-labs
nicksydney has quit [Remote host closed the connection]
[florian1 is now known as [florian]
[florian] has quit [Changing host]
[florian] has joined #m-labs
_florent_ has joined #m-labs
_florent_ has quit [Remote host closed the connection]
_florent_ has joined #m-labs