tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
tokamak has quit [Remote host closed the connection]
tokamak has joined #yosys
FabM has joined #yosys
FabM has quit [Changing host]
FabM has joined #yosys
Xark has quit [Ping timeout: 268 seconds]
Xark has joined #yosys
Xark has quit [Ping timeout: 246 seconds]
Xark has joined #yosys
ec has quit [Ping timeout: 255 seconds]
ec has joined #yosys
ec has quit [Ping timeout: 255 seconds]
ec has joined #yosys
rowanG337 has joined #yosys
<rowanG337>
Hello everyone. I'm a master student under Myrtle.
<rowanG337>
My thesis is about implementing user - timing constraint in nextpnr. Currently I'm looking into how I can best support this from a file format.
<rowanG337>
One of the things mentioned in the yosys issuetracker is it might be helpfull for yosys itself to understand timing constraints as well to make certain synthesis decisions.
<rowanG337>
So I think one of the best ways would be to get yosys to ingest sdc files. Yosys could then output a netlist with timing constraint information.
<rowanG337>
This way SDC handling is isolated to yosys which I think is nicer then requiring the same SDC file to be passed to both yosys and nextpnr.
<tnt>
Might also help with all the net renaming that goes into synthesis ...
<rowanG337>
Yep. It might hard to ensure some nets aren't mangled but I'm not that into yosys yet that I know for sure :p
<rowanG337>
So basically I want to get some feedback from the yosys people on what their view is. I'd hate to waste a lot of time on implementing something like this and it ends up not beeing the right approach.
kristianpaul has joined #yosys
kristianpaul has quit [Client Quit]
kristianpaul has joined #yosys
crzwdjk has quit [Quit: Client closed]
FabM has quit [Quit: Leaving]
<whitequark>
the net renaming should be handled already if you look at the \hdlname attribute
<whitequark>
even flattening keeps hierarchy in that attribute
tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
rowanG337 has quit [Remote host closed the connection]
crzwdjk has joined #yosys
so-offish has joined #yosys
tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
<so-offish>
hey every1
<so-offish>
I have a Nextpnr question - am I in the right place to ask?
<so-offish>
Technically Yosys is the synthesizer and nextpnr is (one of) the routers, so I guess not, but I figured it's mostly the same people (
<whitequark>
it is mostly the same people
tokamak has quit [Quit: ZNC 1.8.2+deb2build5 - https://znc.in]
tokamak has joined #yosys
<so-offish>
I figured; if I'm understanding correctly I need to write a simplified packer, placer, and router for the ECP5; I believe I can achieve this with Nextpnr using its Python API. Can anyone provide hints about how to go about this? I think I will start by simply "placing", "Packing", and "routing" a single inverter as a demonstration - does that exist somewhere?
<so-offish>
Sorry if that's a little long and has poor structure.
<tnt>
huh ? nextpnr already supports ECP5 so why would you do that ?
<so-offish>
Because my routing isn't driven by timing.
<so-offish>
There's an example in there of the antithesis of Cad tools - selecting the longest possible wire route, which I think is hilarious - but it also shows something similar to what I need to do.
<so-offish>
Sounds brilliant - how do I get cracking.