lekernel changed the topic of #milkymist to: Milkymist One, Migen, Milkymist SoC & Flickernoise :: Logs: http://en.qi-hardware.com/mmlogs
jimmythehorn has quit [Quit: jimmythehorn]
gbraad has joined #milkymist
gbraad has joined #milkymist
ohama has quit [Read error: Operation timed out]
ohama has joined #milkymist
xiangfu has joined #milkymist
xiangfu has quit [Ping timeout: 255 seconds]
xiangfu has joined #milkymist
xiangfu has quit [Ping timeout: 276 seconds]
xiangfu has joined #milkymist
hypermodern_ has joined #milkymist
hypermodern_ has quit [Remote host closed the connection]
hypermodern_ has joined #milkymist
hypermodern_ has quit [Quit: hypermodern_]
hypermodern_ has joined #milkymist
hypermodern_ has quit [Remote host closed the connection]
xiangfu has quit [Quit: leaving]
xiangfu has joined #milkymist
elldekaa_ has joined #milkymist
lekernel has joined #milkymist
elldekaa_ has quit [Remote host closed the connection]
<GitHub157>
[mibuild] sbourdeauducq pushed 5 new commits to master: http://git.io/JCgg4g
<GitHub157>
mibuild/master 71c8172 Sebastien Bourdeauducq: xilinx_ise/CRG_SE: reset inversion support
<GitHub157>
mibuild/master 37d8029 Sebastien Bourdeauducq: CRG: support reset inversion
<GitHub157>
mibuild/master 2491017 Sebastien Bourdeauducq: CRG: use new Module API
<larsc>
lekernel: the issue is in the signal name. It iterates over a set of all signals, so the indices for signals with the same name end up different between different runs
<larsc>
sorting the signals, fixes it
<lekernel>
can you send a patch?
<larsc>
I'm not sure where exactly they should be sorted, so I'll let you take care of it
<lekernel>
where's the iteration you are talking about?
<larsc>
in build_namespace()
<larsc>
I just inserted a signals = sorted(signals) in the first line
<lekernel>
and that fixes it?
<larsc>
or I'm lucky
<larsc>
I did do 10 runs, no difference
<lekernel>
Signal derives from HUID, so iteration order on sets should be deterministic
<lekernel>
that's weird
<larsc>
without the patch every second one is different
<larsc>
When I build the milkymist-ng project the asmicon_cmd_chooser_stbX (or similar) signal with X either 1 or 0 tend to be swapped between runs
<lekernel>
yeh, I have that too
playthatbeat has quit [Ping timeout: 245 seconds]
playthatbeat has joined #milkymist
wpwrak has quit [Ping timeout: 245 seconds]
wpwrak has joined #milkymist
<GitHub34>
[mibuild] sbourdeauducq pushed 1 new commit to master: http://git.io/1sWw9A
<lekernel>
German bureaucracy is truly incredible. I wanted to buy a lx9 microboard to play with fpgatools, and Avnet insist on a *German* VAT number without which you can't place your order
<lekernel>
I guess I'll just order from the US where they aren't that pesky and just pay extra in shipping...
<Fallenou>
=(
<lekernel>
and don't tell me about German efficiency :) Farnell lost my company documents and after ages when they finally decided to ship my order, parts were (and still are) missing, then I can also talk about the EHSM room cancelled because of internal confusion 1 week before the event
Gurty has joined #milkymist
<GitHub8>
[migen] sbourdeauducq pushed 4 new commits to master: http://git.io/1iDW_A
<GitHub18>
[mibuild] sbourdeauducq pushed 1 new commit to master: http://git.io/XYV9BA
<GitHub18>
mibuild/master 6feb6e6 Sebastien Bourdeauducq: New clock_domain API
<GitHub8>
migen/master bd8bbd9 Sebastien Bourdeauducq: Make ClockDomains part of fragments
<GitHub8>
migen/master 9b9bd77 Sebastien Bourdeauducq: sim: compatibility with new ClockDomain API
<GitHub8>
migen/master 208e039 Sebastien Bourdeauducq: Local clock domain example
<GitHub87>
[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/QZvswg
<GitHub87>
milkymist-ng/master b2173bb Sebastien Bourdeauducq: Use new ClockDomain API
<larsc>
yeay, clockdomains!
<lekernel>
they were there before, but you could not define module-local ones
<lekernel>
now when you have a module that defines a clock domain itself, and have several instances of that module, it does the right thing
Gurty has quit [Ping timeout: 248 seconds]
<lekernel>
also you can write cd_myclock = ClockDomain() instead of cd_myclock = ClockDomain("myclock") - it also uses tracer.get_obj_var_name() now
<larsc>
and which clock is used by default?
<lekernel>
"sys"
<larsc>
makes sense, I guess
<lekernel>
if you don't want the default behaviour (have sys_clk + sys_rst ports on the generated file) you need to have a ClockDomain named "sys" in some fragment, and drive its signals
mumptai has joined #milkymist
antgreen has joined #milkymist
<GitHub105>
[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/35twjQ