<benzn1989>
Any experts on nextpnr here? Trying to manually place some things and don't really see how this is supposed to work given that the HEaP manually rips up everything on every iteration.
<benzn1989>
Perhaps I can figure out how the SB_IOs are constrained
<benzn1989>
(since they definitely aren't allowed to move around)
s_frit_ has quit [Remote host closed the connection]
s_frit has joined #yosys
<benzn1989>
Aha, it rips up everything that's not the root of a carry chain
oldtopman has joined #yosys
futarisIRCcloud has quit [Quit: Connection closed for inactivity]
X-Scale` has joined #yosys
X-Scale has quit [Ping timeout: 268 seconds]
X-Scale` is now known as X-Scale
show1 has quit [Ping timeout: 240 seconds]
futarisIRCcloud has joined #yosys
mirage335 has quit [Ping timeout: 260 seconds]
dys has quit [Ping timeout: 265 seconds]
show1 has joined #yosys
mirage335 has joined #yosys
dys has joined #yosys
vidbina_ has joined #yosys
bubble_buster has quit []
bubble_buster has joined #yosys
vidbina_ has quit [Ping timeout: 268 seconds]
mithro has quit []
mithro has joined #yosys
attie has joined #yosys
attie has quit [Ping timeout: 265 seconds]
X-Scale` has joined #yosys
X-Scale has quit [Ping timeout: 265 seconds]
X-Scale` is now known as X-Scale
fsasm has joined #yosys
vidbina_ has joined #yosys
marex-cloud has quit []
marex-cloud has joined #yosys
s_frit has quit [Remote host closed the connection]
s_frit has joined #yosys
rohitksingh has joined #yosys
benzn1989 has quit [Remote host closed the connection]
dys has quit [Ping timeout: 248 seconds]
attie has joined #yosys
captain_morgan20 has joined #yosys
captain_morgan20 is now known as captain_morgan
MoeIcenowy has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
MoeIcenowy has joined #yosys
dys has joined #yosys
rjeli has quit []
rjeli has joined #yosys
captain_morgan has quit [Read error: Connection reset by peer]
attie has quit [Ping timeout: 268 seconds]
captain_morgan has joined #yosys
dh73 has joined #yosys
fsasm has quit [Ping timeout: 265 seconds]
dh73 has quit [Read error: Connection reset by peer]
Jybz has joined #yosys
s_frit has quit [Remote host closed the connection]
s_frit has joined #yosys
tecepe has quit []
tecepe has joined #yosys
rohitksingh has quit [Ping timeout: 260 seconds]
emilazy has quit []
emilazy has joined #yosys
jakobwenzel has quit [Quit: jakobwenzel]
attie has joined #yosys
rohitksingh has joined #yosys
citypw has quit [Ping timeout: 268 seconds]
develonepi3 has joined #yosys
s_frit_ has joined #yosys
attie has quit [Ping timeout: 240 seconds]
s_frit has quit [Ping timeout: 265 seconds]
rohitksingh has quit [Ping timeout: 265 seconds]
emeb has joined #yosys
rohitksingh has joined #yosys
rohitksingh has quit [Ping timeout: 265 seconds]
dys has quit [Ping timeout: 245 seconds]
show1 has quit [Ping timeout: 265 seconds]
rohitksingh has joined #yosys
show1 has joined #yosys
rohitksingh has quit [Ping timeout: 265 seconds]
cr1901_modern has quit [Quit: Leaving.]
attie has joined #yosys
cr1901_modern has joined #yosys
attie has quit [Client Quit]
rohitksingh has joined #yosys
vidbina_ has quit [Ping timeout: 268 seconds]
rohitksingh has quit [Ping timeout: 265 seconds]
<awygle>
that's odd. i ran the same build of yosys against the same verilog file with synth_ecp5 under windows (cygwin) and under linux and got different results.
<daveshah>
There are some known issues in that respect
<daveshah>
Related, iirc, to something like destructor ordering
<awygle>
i see
<awygle>
also, -abc9 does not work for synth_ecp5 under cygwin (fails to read the lut file). i don't know if that matters to anyone but me tho
<daveshah>
That's definitely worth filing an issue for
Jybz has quit [Quit: Konversation terminated!]
<awygle>
i suppose i should also include "you have to delete -fPIC to compile under cygwin" as well then
rohitksingh has joined #yosys
<ZirconiumX>
I think it's interesting to note how the primary driver of nextpnr performance is bigger target chips.
<ZirconiumX>
e.g. ECP5 resulted in HeAP placement because SA takes a good while
<ZirconiumX>
And now Xilinx targeting has router2
<daveshah>
router2 is nothing to do with Xilinx being bigger fwiw
<daveshah>
It doesn't actually perform much better even on big ECP5 designs right now
<ZirconiumX>
Hmm, I got the impression of that from your Patreon post
<daveshah>
But on ECP5 sized Xilinx designs it copes much better because it handles the resultant congestion better
<ZirconiumX>
Ah, okay
<daveshah>
The hope was that it would scale better but the most pressing concern was dealing with the congestion from having fewer routing resources
<ZirconiumX>
I had a look at nextpnr-xilinx and noted there was an option for OpenMP acceleration in the CMakeLists
<daveshah>
There is in normal nextpnr too
<daveshah>
It speeds up the sparse matrix solving part of HeAP slightly
<awygle>
ZirconiumX and I were just talking about this out of band but is nextpnr likely to have an issue routing two independent designs in the same, uh, "project"? i.e. two sections of the chip being totally unconnected and on separate clock domains?
<daveshah>
Overheads mean it is only worth it for big designs
<daveshah>
awygle: unlikely I'd say
<daveshah>
Router definitely should care
<daveshah>
*shouldn't
<ZirconiumX>
I mean, one of the designs has a target Fmax of a few megahertz
<daveshah>
So long as each part has some external IO, placement shouldn't be a problem either
<ZirconiumX>
So it shouldn't have problems in meeting timing
<daveshah>
nextpnr should already optimise each clock domain somewhat separately anyway
<awygle>
lol: "Info: Max frequency for clock 'ap_clk': 192.57 MHz (PASS at 2.00 MHz)"
<ZirconiumX>
I feel like my chess move generator is going to be an excellent example of an overcongested design >.>
<awygle>
nextpnr is so fast ;_; it finished in less than a second, diamond can't even start that fast
<ZirconiumX>
I mean, it depends on the design, naturally
<awygle>
sure, but still
<ZirconiumX>
...I still wish I could use Quartus on out of context designs...