Lofty changed the topic of #prjmistral to: Project Mistral: Yosys (and hopefully nextpnr) on Cyclone FPGAs - https://github.com/ZirconiumX/mistral - logs: https://freenode.irclog.whitequark.org/prjmistral
jopdorp has quit [Read error: Connection reset by peer]
jopdorp has joined #prjmistral
QDX45_ has joined #prjmistral
QDX45 has quit [Ping timeout: 265 seconds]
_whitelogger has joined #prjmistral
<Sarayan> Interesting design: all connections to a routing line are done in one point, on a very small rectangle (like 4x2, 4x2, that kind of size) of bits in the 2d firmware. Which means there's only one driving spot on evey metal line
<Sarayan> then there's multiple connections downstream of course. It's all 1:n
<Lofty> I guess that's one way of avoiding multiple drivers
<Sarayan> it's not n:m with a constraint of only one of the n active, there's really one one point
<Sarayan> only
<Sarayan> yeah
<Lofty> I think Xilinx has something similar, but maybe daveshah or mwk could correct me there
<Sarayan> on the spartan 2 I don't remember it being that obvious
<Sarayan> that was "4x1, 4x2" incidentally
<Sarayan> iirc the startan 2 was more single-mosfet connect-x-to-y all over the place
<mwk> starting with virtex 2 / spartan 3, that's what the xilinx routing looks like as well
<mwk> ... mostly
<mwk> there's also such a thing as long lines, which can be driven from one of 4 tiles
<mwk> which are finally gone on ultrascale
<Sarayan> so they're converging on that aspect
QDX45_ has quit [Quit: Leaving]
QDX45 has joined #prjmistral