mkdir has quit [Remote host closed the connection]
Jybz has joined ##openfpga
gsi__ is now known as gsi_
Jybz has quit [Quit: Konversation terminated!]
azonenberg_work has quit [Remote host closed the connection]
bibor has quit [Ping timeout: 252 seconds]
bibor has joined ##openfpga
lutsabound has joined ##openfpga
Miyu has joined ##openfpga
<keesj>
mkdir I sometimes directly program the extenral flash
s_frit has quit [Remote host closed the connection]
s_frit has joined ##openfpga
azonenberg_work has joined ##openfpga
azonenberg_work has quit [Ping timeout: 245 seconds]
Asu has joined ##openfpga
cr1901_modern has quit [Read error: Connection reset by peer]
Bob_Dole has quit [Read error: Connection reset by peer]
Asu` has joined ##openfpga
Asu has quit [Ping timeout: 258 seconds]
azonenberg_work has joined ##openfpga
Asu has joined ##openfpga
Asu` has quit [Ping timeout: 246 seconds]
lutsabound has quit [Quit: Connection closed for inactivity]
lutsabound has joined ##openfpga
scream has joined ##openfpga
adamgreig has quit [Ping timeout: 258 seconds]
adamgreig has joined ##openfpga
cr1901_modern has joined ##openfpga
<tnt>
whitequark: wrt to https://github.com/YosysHQ/yosys/pull/1183/ wouldn't it be better to ignore the flag ? So as not to break every script that now has -relut option in it ...
lutsabound has quit [Quit: Connection closed for inactivity]
Bob_Dole has joined ##openfpga
Miyu has quit [Read error: Connection reset by peer]
Miyu has joined ##openfpga
vonnieda has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
dh73 has joined ##openfpga
dh73 has quit [Remote host closed the connection]
dh73 has joined ##openfpga
vonnieda has joined ##openfpga
wpwrak has quit [Ping timeout: 258 seconds]
wpwrak has joined ##openfpga
<_whitenotifier-3>
[whitequark/Boneless-CPU] whitequark pushed 4 commits to master [+0/-0/±6] https://git.io/fjPoF
<_whitenotifier-3>
[whitequark/Boneless-CPU] whitequark 3697884 - gateware.core_v3: optimize MEM/EXT data mux.
mumptai has joined ##openfpga
<whitequark>
abc is really not being cooperative
<whitequark>
i perturb the design a bit and the LUT count jumps by like 5%
<whitequark>
literally just changing the encoding of internal control signals should not materialize or disappear 20 LUTs...
<whitequark>
it's especially annoying when i *remove* logic and the LUT count *rises*
<whitequark>
for example, remove a 3'b1?? case and replace it with default: in two places and that adds 24 LUTs. wtf.
<whitequark>
and of course it shreds all evidence of its misbehavior by making debug info useless
<whitequark>
i hate abc so much
<tnt>
That's been my experience too (back when I was writing decoding logic).
<tnt>
For something that's supposed to be good at optimizing combinatorial logic, it wasn't hard to beat it by hand.
<whitequark>
because it uses fucking AIGs
<whitequark>
fuck AIGs
<whitequark>
there's a new switch -abc9 which makes it work *very slightly* less bad
<whitequark>
the commits above remove 78 LUTs without -abc9 and ~55 LUTs with -abc9
<whitequark>
except the smaller abc9 design is much more bloated than the larger no-abc9 design... so it's also useless
<whitequark>
if the abc9 designs were a little bit larger, they would be competitive with my flowmap pass (in a bad way, that is)
show has quit [Quit: WeeChat 2.4]
s_frit has quit [Remote host closed the connection]
s_frit has joined ##openfpga
show has joined ##openfpga
Miyu has quit [Ping timeout: 252 seconds]
solo1 has joined ##openfpga
dh73 has quit [Remote host closed the connection]
Asu` has joined ##openfpga
Asu has quit [Ping timeout: 268 seconds]
solo1 has quit [Ping timeout: 248 seconds]
dh73 has joined ##openfpga
<whitequark>
abc just managed to find a way to blow up a 4-bit state machine--a glorified counter--to 64 LUTs
<whitequark>
it actually works much better if I add -noflatten
<whitequark>
because it doesn't try to do stupid shit
Asu` has quit [Remote host closed the connection]
mumptai has quit [Quit: Verlassend]
<zignig>
whitequark: have you actually got v3 running on FGPA ?
<whitequark>
zignig: nope
<whitequark>
never tried
<duck2>
how? a4 lut should cover every boolean fn with 4 inputs and an output. so it should be 4 luts really. maybe it _flattens_ to one-hot encoding? then 64 luts makes sense- 4 luts for every state bit.
<whitequark>
duck2: no that doesn't make sense
<whitequark>
why the fuck would you flatten a 4-bit counter to one-hot
<zignig>
I'm back home tomorrow , I'll give it a try for S&G.
<duck2>
whitequark: uh, because you aren't passed -noflatten :p
<whitequark>
duck2: that's not what -noflatten means.
<whitequark>
you should probably take a look at the synth_ice40 script.
Bike has joined ##openfpga
GenTooMan has joined ##openfpga
vonnieda has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
dh73 has quit [Ping timeout: 260 seconds]
<_whitenotifier-3>
[whitequark/Boneless-CPU] whitequark pushed 2 commits to master [+0/-0/±4] https://git.io/fjPSO