clifford changed the topic of #yosys to: Yosys Open SYnthesis Suite: http://www.clifford.at/yosys/ -- Channel Logs: https://irclog.whitequark.org/yosys
matthuszagh has quit [Ping timeout: 256 seconds]
emeb has quit [Quit: Leaving.]
pacak has quit [Ping timeout: 258 seconds]
pacak has joined #yosys
Degi has quit [Ping timeout: 240 seconds]
Degi has joined #yosys
citypw has joined #yosys
flaviusb has quit [Ping timeout: 260 seconds]
FFY00 has quit [Remote host closed the connection]
flaviusb has joined #yosys
craigo has joined #yosys
craigo has quit [Remote host closed the connection]
craigo has joined #yosys
proteusguy has quit [Ping timeout: 240 seconds]
proteusguy has joined #yosys
Asu has joined #yosys
proteusguy has quit [Ping timeout: 246 seconds]
proteusguy has joined #yosys
jakobwenzel1 has joined #yosys
proteusguy has quit [Ping timeout: 256 seconds]
proteusguy has joined #yosys
xtro has quit [Ping timeout: 240 seconds]
SpaceCoaster has quit [Quit: ZNC 1.7.2+deb3 - https://znc.in]
proteusguy has quit [Ping timeout: 240 seconds]
kraiskil has joined #yosys
kraiskil has quit [Ping timeout: 258 seconds]
AdamHorden has quit [Quit: ZNC 1.8.1 - https://znc.in]
dormito has quit [Ping timeout: 256 seconds]
kraiskil has joined #yosys
dormito has joined #yosys
kraiskil has quit [Ping timeout: 265 seconds]
SpaceCoaster has joined #yosys
kraiskil has joined #yosys
proteusguy has joined #yosys
X-Scale has joined #yosys
simeonm has quit [Ping timeout: 246 seconds]
simeonm has joined #yosys
simeonm has quit [Ping timeout: 240 seconds]
simeonm has joined #yosys
FFY00 has joined #yosys
AdamHorden has joined #yosys
alexhw has joined #yosys
_whitelogger has joined #yosys
citypw has quit [Ping timeout: 240 seconds]
jakobwenzel1 has quit [Quit: jakobwenzel1]
tpb has joined #yosys
tpb has quit [Remote host closed the connection]
emeb has joined #yosys
Guest30238 is now known as gtw
xtro has joined #yosys
strongsaxophone has joined #yosys
cr1901_modern has quit [Read error: Connection reset by peer]
cr1901_modern has joined #yosys
filt3r has joined #yosys
kraiskil has quit [Ping timeout: 256 seconds]
az0re has quit [Remote host closed the connection]
kraiskil has joined #yosys
<lf> Hi i want to take a look at an ecp5 bitstream. is there a way to convert it back into a json file? ecpunpack gives me an .config that is hard to read. ideally i would like to get a verilog file with the SB_ instances to make easer to change things.
<daveshah> No that's not possible
<daveshah> for iCE40 there is a tool icebox_vlog but I never got round to doing similar for ECP5
jordigw has joined #yosys
<lf> ah ok was hopeing i could just load it into nextpnr and get an json out.
<lf> i will read up on the config format
az0re has joined #yosys
TalS has joined #yosys
craigo has quit [Ping timeout: 256 seconds]
dormito has quit [Ping timeout: 256 seconds]
Asu has quit [Remote host closed the connection]
tpearson-mobile has joined #yosys
<tpearson-mobile> Is there a way to get yosys to spit out resource usage per module?
<tpearson-mobile> I'm trying to determine why a very complex design is using a tremendous amount of resources vs. the proprietary tooling, and being able to isolate the offending module(s) would help
<daveshah> Add -noflatten to synth_ecp5 or whatever
m4ssi has joined #yosys
<tpearson-mobile> Thanks, I'll give that a go!
<daveshah> Oh, incidentally, for ECP5 the nextpnr SLICE usage isn't necessarily representative of actual LUT usage, due to the fairly sparse packing at the moment to ease routing
<daveshah> The current packer will actually 'stick' at around 95% of SLICEs for quite varied LUT and FF usages
<tpearson-mobile> Good to know, that actually explains a lot
<daveshah> This is something that is being improved over the next few months
<tpearson-mobile> however, I still need to track down an inordinate amount of FFs being generated, so the noflatten option is what I'm looking for
<daveshah> Something that the ECP5 flow doesn't do yet is pack FFs into DSPs
<tpearson-mobile> hmmm...yeah, that could be related (the resource hog is a CPU)
<daveshah> Sounds unlikely then if its a big difference
<tpearson-mobile> OK
<daveshah> That mostly affects signal processing pipelines with lots of multipliers and FFs
<tpearson-mobile> yeah I was wondering if maybe the FPU was balooning out
<daveshah> Whereas for a typical CPU I would expect that to be <<1k FFs
<daveshah> Oh, could be that yeah
<tpearson-mobile> I've got a bunch of stats to dig through now, so I'll spend a bit of time doing that and see where the problem actually resuldes
<tpearson-mobile> *resides
<daveshah> Out of curiosity, what are you comparing against? Synplify or LSE?
<tpearson-mobile> To be honest I'm not the one running the proprietary tooling, that's another group. I believe they're checking against the Lattice tooling
<tpearson-mobile> but they're indicating that the resource usage on the open flow is *way* out of line, like 3x higher
<daveshah> That sounds a lot like BRAM inference issues
<tpearson-mobile> It does, yeah
<tpearson-mobile> hence why I want to go module by module and check for the high consumption one
<Lofty> Check what the memory_bram pass says
<tpearson-mobile> hm, whole bunch of " No acceptable bram resources found."
<tpearson-mobile> that explains the I/D cache resource explosions, but I was already expecting that to be honest
<tpearson-mobile> and have the caches cut back to bare minimums as a result
<tpearson-mobile> 5000 cells seems a bit large for a two line D cache ;)
<tpearson-mobile> it's odd because it's inferring some BRAMs but obviously not all of them
<daveshah> Are they TDP RAMs?
<tpearson-mobile> I'd expect so, yes
<tpearson-mobile> the ECP5 does have TDP resources though as far as I know?
<daveshah> Yes but Yosys doesn't support them
<tpearson-mobile> oh. ouch.
<tpearson-mobile> that'd explain it
<tpearson-mobile> nearly 7k cells just for the I cache
<tpearson-mobile> Rule #1 for bram type $__ECP5_PDPW16KD (variant 1) accepted. Mapping to bram type $__ECP5_PDPW16KD (variant 1): Shuffle bit order to accommodate enable buckets of size 9.. Results of bit order shuffling: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45
<tpearson-mobile> 46 47 48 4$ Write port #0 is in clock domain \clk. Mapped to bram port A1. Read port #0 is in clock domain !~async~. Bram port B1.1 has incompatible clock type. Failed to map read port #0. Mapping to bram type $__ECP5_PDPW16KD failed.
<tpearson-mobile> Is there a roadmap / timing for TDP support?
<Lofty> Basically nobody wants to touch memory_bram
<Lofty> I think lambda volunteered to rewrite it, but yeah
<tpearson-mobile> oh. ouch.
tpearson-mobile has quit [Remote host closed the connection]
dormito has joined #yosys
m4ssi has quit [Remote host closed the connection]
kraiskil has quit [Ping timeout: 240 seconds]
kraiskil has joined #yosys
lf has quit [Ping timeout: 256 seconds]
lf has joined #yosys
kraiskil has quit [Ping timeout: 240 seconds]