jakobwenzel has quit [Remote host closed the connection]
jakobwenzel has joined #yosys
jakobwenzel has quit [Ping timeout: 260 seconds]
jakobwenzel has joined #yosys
X-Scale` has joined #yosys
jakobwenzel has quit [Ping timeout: 260 seconds]
jakobwenzel has joined #yosys
X-Scale has quit [Ping timeout: 272 seconds]
X-Scale` is now known as X-Scale
jakobwenzel1 has joined #yosys
jakobwenzel has quit [Ping timeout: 240 seconds]
jakobwenzel1 is now known as jakobwenzel
<agg>
has anyone run the prjtrellis fuzzers since the change to pybind in #154? many of them are failing for me with python type errors where wrapped classes don't have constructors or don't support assignment, and no database files are getting written out, whereas if I checkout prjtrellis before #154 it seems OK
vidbina has joined #yosys
kraiskil has joined #yosys
jakobwenzel has quit [Ping timeout: 264 seconds]
kraiskil has quit [Ping timeout: 240 seconds]
peeps[zen] has joined #yosys
peepsalot has quit [Ping timeout: 240 seconds]
<agg>
also I guess I'm far out into the weeds here but my diamond 3.11 install doesn't have 'bsdata.acd' which it seems like bstool needs to dump a bitstream, does that get generated or was it removed after 3.10 or something?
<daveshah>
The bitstream dumps aren't needed for this kind of fuzzing so you can just patch out that part of diamond.sh
<agg>
I was vaguely hoping it might tell me more about what's different in the diamond-built bitstream compared to ecpunpack
<agg>
(or use it with compare_bits.py)
<agg>
(is this the best/a good channel to ask about prjtrellis stuff, btw?)
<daveshah>
I think you might have to get Diamond 3.10 then (Lattice have an archive section)
<daveshah>
It's as good as any
peepsalot has joined #yosys
peeps[zen] has quit [Ping timeout: 265 seconds]
peeps[zen] has joined #yosys
peepsalot has quit [Ping timeout: 265 seconds]
blunaxela has quit [Read error: Connection reset by peer]
blunaxela has joined #yosys
emeb has joined #yosys
<agg>
hmm, got 3.10 installed but it's also missing bsdata.acd
<daveshah>
Sounds like something has gone wrong with the environment variables then
emeb_mac has joined #yosys
mndza has joined #yosys
<cr1901_modern>
My diamond 3.11 install on Windows works fine, fwiw
<cr1901_modern>
I don't remember how to install diamond on Ubuntu- the article I used no longer exists
<cr1901_modern>
also, some shell scripts lie about being sh compat and need to be changed to bash
<cr1901_modern>
need the shebang* to be changed to bash
<agg>
cr1901_modern: when you say works fine, do you know if that includes "bstool -d" specifically? i think you won't notice if it's broken unless you're going looking at the output
<agg>
(does your diamond install contain the file ispfpga/data/bsdata.acd ?)
<cr1901_modern>
I don't have a bsdata.acd file. Dump files are being generated fine
<agg>
huh, ok, thanks for checking!
<cr1901_modern>
I know this because if there were no dump files, I couldn't fill out the JSON for MachXO2 devices. And I only did some of the remaining members after I install diamond 3.11
<agg>
my bstool is complaining about "ERROR - baspmdevkey(476): Process ACD file "bsdata" failed. Used ACD File Name is </usr/local/diamond/3.10_x64/ispfpga/data/bsdata.acd>."
<agg>
which i assumed meant somehow that file was missing, but I guess actually means it's somehow not doing the right thing instead
<agg>
will keep poking it then
<cr1901_modern>
My diamond 3.10 doesn't have bsdata.acd either (Linux)
peeps[zen] is now known as peepsalot
xtro has joined #yosys
ayazar1 has joined #yosys
lf_ has joined #yosys
lf has quit [Ping timeout: 260 seconds]
ayazar1 has quit [Quit: ayazar1]
ayazar1 has joined #yosys
ayazar1 has quit [Quit: ayazar1]
lf_ has quit [Ping timeout: 264 seconds]
lf has joined #yosys
kraiskil has joined #yosys
alexhw has quit [Ping timeout: 268 seconds]
<cr1901_modern>
daveshah: pytrellis switched to pybind?
<daveshah>
yeah, mmicko_ did that a couple of months ago
<cr1901_modern>
This is... gonna be fun to merge
<cr1901_modern>
Must've missed it
<agg>
I'm reasonably sure it broke a bunch of fuzzers but I can probably fix it
<agg>
Though as far as I can tell the old fuzzers aren't yielding any new insight into my MAC :p
<agg>
Is it possible they're not detecting bits for connecting the MULTs to the ALUs (which are sort of fixed connections, in that the ALU MA input can only come from MULT P out), since no fuzzer cases have both instances? Not sure if the general routing fuzzer will get them.
<daveshah>
yeah, there could be something like that going on
<daveshah>
they would show up as unknown bits in the ecpunpack output from the diamond design though