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
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
<sb0> rjo, why did you add "self.platform.add_period_constraint(self.rtio.cd_rsys.clk, 8.)"?
<sb0> self.rtio.cd_rsys.clk is just the system clock, it should already have a 8ns constraint from misoc. is that not the case?
<sb0> it's an implicit constraint through the PLL, maybe that failed?
<sb0> (the 125MHz clock is generated from clk200, migen adds a 5ns constraing on clk200, and vivado/ise are supposed to make that 8ns after it goes through the PLL)
<sb0> *constraint
<sb0> as for yqis, it looks like a pretty general conference. didn't you say you wanted to talk to phd students/postdocs?
<sb0> it's also reasonably priced for once, unlike the usual 3- or 4-digit fees of "prestigious" academic conferences
rohitksingh_work has joined #m-labs
rohitksingh_wor1 has joined #m-labs
rohitksingh_work has quit [Ping timeout: 258 seconds]
mumptai has joined #m-labs
ssk1328 has joined #m-labs
<ssk1328> Hi everyone I am GSoC student with TimVideos, I had question about defining multiple sinks for a pipelined actor?
<ssk1328> This is how I am defining multiple sinks for a FloatAddRGB unit, and further down how I am using it http://paste.ubuntu.com/21109705/
<mithro> ssk1328: Just ask the question and people will reply if they can. Try and be as detailed as possible and link to code that people can read
<mithro> ssk1328: Linking to full code on github is better then snippets like that
<ssk1328> A more detailed description of problem, link to code that defines the PipelinedActor https://github.com/ssk1328/HDMI2USB-misoc-firmware/blob/float-arithmetic/gateware/float_arithmetic/floatadd.py#L207
<ssk1328> And then this is how I use this PipelinedActor for connections with another module https://github.com/ssk1328/HDMI2USB-misoc-firmware/blob/float-arithmetic/gateware/hdmi_out/phy.py#L248
<ssk1328> This is the terminal output of error I get when I try to compile this http://paste.ubuntu.com/21109623/
<ssk1328> I sense that I am not defining the two sinks correctly, and there is a correct way to do it
jaeckel has quit [Ping timeout: 276 seconds]
rohitksingh_wor1 has quit [Read error: Connection reset by peer]
rohitksingh_work has joined #m-labs
ssk1328 has quit [Ping timeout: 250 seconds]
ssk1328 has joined #m-labs
swivel has quit [Ping timeout: 240 seconds]
mumptai has quit [Read error: Connection reset by peer]
jaeckel has joined #m-labs
<rjo> sb0: where is that line?
<rjo> sb0: iirc i went to yqis a couple years ago. the reason you don't get your knee jerk "too expensive" reaction is that it is more of an invitation basis.
<rjo> and when i went, i would not have wanted companies to be there. and there are generally none.
<_florent_> ssk1328: you probably need to connect your signals manually for this case (at least part of), connecting a source to two sink is not correct.
<GitHub106> [artiq] jordens pushed 1 new commit to phaser: https://git.io/vKd74
<GitHub106> artiq/phaser d928455 Robert Jordens: dsp/tools: update satadd mixin
<sb0> rjo, there's is d-wave. and why not?
<sb0> http://yqis16.icfo.eu/#register how is that more invitation-based than other conferences?
<rjo> sb0: i remember it being a workshop, less of a conference. and that is directly reflected in the phrasing of their "mission statement"
<rjo> maybe the layout changed.
<rjo> i wouldn't want to compare us with d-wave.
swivel has joined #m-labs
<rjo> oh. i was at another conference, not that one. but my concern still stands: that conference is for young researchers by young reserchers.
<sb0> they have Chris Monroe and Sandu Popescu speaking there
<sb0> so, not completely. and as i said, didn't you mention you wanted to advertise to phd students and postdocs?
<rjo> the format is described in last year's program. the invited speakers seem to be doing tutorials.
<rjo> and those are the invited speakers.
<rjo> the ones filling out the registrations and submitting abstracts are young postdocs and phd students.
<rjo> yes. but imho the forum to speak to those is dpg or damop. or maybe a more specialized workshop that is not that much targetted at getting young researchers into their fields.
<rjo> and the cost of the conference registration always seems small compared to the flights and the accommodation, wether it's 60 eur or 200 eur.
<rjo> anyway. i can give it a shot. barcelona is always nice.
ssk1328 has quit [Quit: Connection closed for inactivity]
<GitHub191> [artiq] sbourdeauducq pushed 4 new commits to master: https://git.io/vKFId
<GitHub191> artiq/master a89f96e Sebastien Bourdeauducq: runtime: support boards without LEDS
<GitHub191> artiq/master 7928ee4 Sebastien Bourdeauducq: runtime: support boards without RTIO CRG
<GitHub191> artiq/master 8fab789 Sebastien Bourdeauducq: runtime: support RTIO configurations without address (e.g. all simple TTL out)
<bb-m-labs> build #840 of artiq is complete: Failure [failed conda_build] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/840 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
<sb0> simplified drtio is almost done, remaining items are clock cleanup with si5324 + fixing any bugs
<sb0> bb-m-labs, force build artiq
<bb-m-labs> build forced [ETA 25m10s]
<bb-m-labs> I'll give a shout when the build finishes
<whitequark> sb0: what happens if I NAK a wishbone tx?
<sb0> whitequark, what is NAK? not asserting ack?
<sb0> it blocks until you assert ack
<sb0> rjo, I have printed artiq t-shirts, do you want some?
<whitequark> oh
<larsc> a more upstream master might decide to generate a timeout, not sure if the lm32 does that though
<sb0> are timeouts part of the spec?
<sb0> afaik lm32 doesn't do that, it just crashes
<larsc> I don't think so
<larsc> stb and ack are the handshaking signals, if you want to indicate e.g. an error you have to assert the err signal
<larsc> but still use stb/ack normally
<larsc> oh, whishbone is special
<larsc> you have to either assert ack or err
<larsc> or rty
<larsc> if neither is set it just blocks
<larsc> which is bad
<larsc> don't build blocking cores
<larsc> unless it is for memory mapped devices
<bb-m-labs> build #566 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/566
<whitequark> sb0: does migen just group everything into one massive combinatorial and one massive synchronous block?
<whitequark> well, the latter per clock domain, I guess
<bb-m-labs> build #841 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/841
<rjo> sb0: sure. 2.0 logo?
<rjo> whitequark: it emits individual combinatorial blocks. but yes.
<sb0> rjo, yes
<GitHub154> [artiq] sbourdeauducq pushed 1 new commit to master: https://git.io/vKF3x
<GitHub154> artiq/master 78366ed Sebastien Bourdeauducq: runtime: add missing include
<bb-m-labs> build #567 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/567
fengling has quit [Ping timeout: 240 seconds]
<bb-m-labs> build #842 of artiq is complete: Failure [failed python_unittest_1] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/842 blamelist: Sebastien Bourdeauducq <sb@m-labs.hk>
ssk1328 has joined #m-labs
<whitequark> sb0: >cannot load kernel: ELF object has an unresolved symbol: dds_batch_exit
<cr1901_modern> larsc: "unless it is for memory mapped devices" Isn't that the use case for most wishbone cores?
<larsc> cr1901_modern: I mean something like DDR or SPI flash
<larsc> bad wording, sorry
<larsc> real memory that is mapped
<larsc> not registers
<cr1901_modern> Ahhh, that makes more sense. And makes it imperative that you have cache sitting between your CPU and WB interface since SPI flash is slow as molasses :P
<ssk1328> _florent_: That was what I tried at first, but there are some other signals like ack and stb which needs to be connected in this case, and I was not sure about them
<ssk1328> _florent_: Plus here connected source to two sinks was for testing, (floatmult.source is multiplying by 0.5, and further connecting it to both floadadd.sinks should give me output same as initial input)
<GitHub184> [artiq] sbourdeauducq pushed 1 new commit to master: https://git.io/vKFWQ
<GitHub184> artiq/master 9233802 Sebastien Bourdeauducq: runtime: RTIO_DDS_COUNT -> CONFIG_RTIO_DDS_COUNT
rohitksingh_work has quit [Read error: Connection reset by peer]
<bb-m-labs> build #568 of artiq-kc705-nist_clock is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-kc705-nist_clock/builds/568
<whitequark> amazing. pulseview just segfaults if I try to put a vcd into it
<sb0> never tried that one, I use gtkwave
<sb0> it's far from perfect, but it does the job
<whitequark> I'm already using gtkwave, but gtkwave doesn't do decoding
<whitequark> the waveforms look sane to me but if I know one thing I'm awful at, it's tracking minor errors in them
<larsc> there have been a couple of fixes for vcd recently in case you are not using the current git
fengling has joined #m-labs
<whitequark> does this look reasonable or does it have gross bugs in wishbone interface?
<bb-m-labs> build #281 of artiq-win64-test is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq-win64-test/builds/281
<bb-m-labs> build #843 of artiq is complete: Success [build successful] Build details are at http://buildbot.m-labs.hk/builders/artiq/builds/843
fengling has quit [Ping timeout: 240 seconds]
rohitksingh has joined #m-labs
rohitksingh1 has joined #m-labs
rohitksingh has quit [Ping timeout: 250 seconds]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
rohitksingh1 has quit [Ping timeout: 276 seconds]
rohitksingh has joined #m-labs
ssk1328 has quit [Quit: Connection closed for inactivity]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
ssk1328 has joined #m-labs
<cr1901_modern> whitequark: To fix the segfault, someone needs to add support for VCD vectors in Pulseview. sigrok team really doesn't care for VCD.
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
rohitksingh1 has joined #m-labs
rohitksingh has quit [Ping timeout: 258 seconds]
<sb0> whitequark, do not do combinatorial wishbone feedback. it is almost guaranteed to fail timing.
<sb0> besides that, the wishbone signals look ok to me
<sb0> note that the code you want is:
<sb0> sync += ack.eq(0), If(~ack & cyc & stb, ack.eq(1), perform_transaction)
<sb0> regular wishbone doesn't do pipelining so each transaction (without comb feedback) will take 2 cycles
<sb0> 1st cycle cyc=stb=1, ack=0
<sb0> 2nd cycle cyc=stb=1, ack=1
<sb0> 3rd cycle cyc=stb=0 unless there's another transaction starting
rohitksingh1 has quit [Ping timeout: 252 seconds]
rohitksingh has joined #m-labs
<whitequark> I see
<whitequark> I'll fix that and test it on real hardware tomorrow, then
<sb0> kk. later there should be a trivial microcontroller that plays transactions from a list in block RAM, to set up the si5324 for 62.5MHz clock cleanup
<sb0> by trivial microcontroller I mean some sort of two-instruction FSM-based processor
mumptai has joined #m-labs
<larsc> I did something like that for SPI a while ago (not in migen though). https://wiki.analog.com/resources/fpga/peripherals/spi_engine/engine
rohitksingh1 has joined #m-labs
rohitksingh has quit [Ping timeout: 250 seconds]
fengling has joined #m-labs
rohitksingh1 has quit [Ping timeout: 252 seconds]
fengling has quit [Ping timeout: 240 seconds]
<larsc> the hardware only knows the field number
<larsc> at some point, before reaching the userspace application, the kernel has to do the mapping from number to order
<larsc> wrong channel
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
<ssk1328> _florent_: More detailed description of the problem I am facing in the first two pages of this doc. https://docs.google.com/document/d/1g1c2IwCVxVzSHWdXbZ746HP-fnM4y1WqFuBZNkLi5mw/edit
<ssk1328> _florent_: I need to find the lines I need to add for the equivalent of Record.connect()
fengling has joined #m-labs
fengling has quit [Ping timeout: 240 seconds]
fengling has joined #m-labs