| *** tpb has joined #symbiflow | 00:00 | |
| *** wallacejohn has quit IRC | 00:13 | |
| *** lopsided98 has quit IRC | 00:24 | |
| *** HEGAZY has quit IRC | 00:25 | |
| *** lopsided98 has joined #symbiflow | 00:26 | |
| wavedrom | Hi All. I am new to channel. | 00:30 |
|---|---|---|
| wavedrom | Is it a good place to discuss ideas for GSOC? | 00:30 |
| ZirconiumX | Sure | 00:35 |
| ZirconiumX | Everybody else has :P | 00:35 |
| ZirconiumX | wavedrom: ^ | 00:35 |
| *** wallacejohn has joined #symbiflow | 00:40 | |
| wavedrom | does SymbiFlow has official UI of some sort? | 00:41 |
| *** balmlake has joined #symbiflow | 00:44 | |
| wavedrom | I am kinda Visual guy. I like all sort of diagrams, schematics, charts | 00:44 |
| wavedrom | I was even thinking about some sort of IDE-EDA https://github.com/drom/atom-ide-eda | 00:48 |
| tpb | Title: GitHub - drom/atom-ide-eda: Atom as IDE for EDA (at github.com) | 00:48 |
| duck2 | i think even the UI-less flow is unofficial right now | 00:57 |
| ZirconiumX | wavedrom: Depends what you want as a UI. | 01:05 |
| ZirconiumX | nextpnr has a GUI | 01:05 |
| ZirconiumX | So you can explore a design while routing it | 01:05 |
| ZirconiumX | Yosys does not, but...it's kinda questionable how it would benefit from it | 01:06 |
| wavedrom | well, Synplify Pro was my favorite FPGA synthesis tool just because of good RTL view https://electronix.ru/forum/uploads/monthly_01_2014/post-45001-1391180774.png | 01:11 |
| wavedrom | and netlist view too | 01:13 |
| wavedrom | this is a good start https://github.com/nturley/netlistsvg | 01:18 |
| tpb | Title: GitHub - nturley/netlistsvg: draws an SVG schematic from a JSON netlist (at github.com) | 01:18 |
| -_whitenotifier-3- [ideas] drom opened issue #41: RTL Schematic View - https://git.io/JvP9G | 01:38 | |
| *** citypw has joined #symbiflow | 01:40 | |
| ZirconiumX | wavedrom: nextpnr can produce something like it | 01:44 |
| wavedrom | can you send a link? | 02:31 |
| ZirconiumX | https://github.com/daveshah1/nextpnr-xilinx | 02:31 |
| tpb | Title: GitHub - daveshah1/nextpnr-xilinx: Experimental flows using nextpnr for Xilinx devices (at github.com) | 02:31 |
| wavedrom | screenshot? | 02:34 |
| ZirconiumX | https://twitter.com/q3k/status/1024623710165237760 | 02:36 |
| ZirconiumX | https://twitter.com/gojimmypi/status/1090097876707926016 | 02:36 |
| *** Bertl_oO is now known as Bertl_zZ | 03:26 | |
| *** balmlake has quit IRC | 03:27 | |
| *** wallacejohn has quit IRC | 04:30 | |
| *** Vonter has quit IRC | 04:50 | |
| *** balmlake has joined #symbiflow | 04:58 | |
| *** Vonter has joined #symbiflow | 05:47 | |
| *** balmlake has quit IRC | 06:27 | |
| wavedrom | it is not exactly a schematic | 06:48 |
| wavedrom | it is more like topology | 06:48 |
| *** OmniMancer has joined #symbiflow | 07:30 | |
| *** OmniMancer1 has joined #symbiflow | 07:31 | |
| *** OmniMancer has quit IRC | 07:34 | |
| *** wavedrom has quit IRC | 07:39 | |
| *** clay_1 has joined #symbiflow | 08:22 | |
| clay_1 | Hello! | 08:45 |
| *** Bertl_zZ is now known as Bertl | 09:12 | |
| *** _whitelogger has quit IRC | 10:06 | |
| *** _whitelogger has joined #symbiflow | 10:08 | |
| *** rjordans has joined #symbiflow | 10:20 | |
| clay_1 | litghost Sorry for yet another late reply. I have done the following trying to tackle the issue: | 11:51 |
| clay_1 | I open a terminal and try to issue the PYTHONPATH like in the test but without the CMAKE -E env doing the following | 11:53 |
| clay_1 | `PYTHONPATH= symbiflow-arch-defs/build/env/conda/lib/python3.7/site-packages: symbiflow-arch-defs/third_party/prjxray: symbiflow-arch-defs/third_party/prjxray/third_party/fasm: symbiflow-arch-defs/xc7: symbiflow-arch-defs/utils symbiflow-arch-defs/build/env/conda/bin/python3` | 11:53 |
| ZirconiumX | Hi clay_1 | 11:53 |
| clay_1 | ZirconiumX heyy :) | 11:54 |
| clay_1 | This opens a python3 environment (if I am not mistaken) | 11:54 |
| clay_1 | and there I type the following | 11:54 |
| clay_1 | `-mfasm2bels --connection_database /home/michailm/phd/symbiflow/symbiflow-arch-defs/build/xc7/archs/artix7/devices/xc7a50t-basys3-roi-virt/channels.db--db_root /home/michailm/phd/symbiflow/symbiflow-arch-defs/third_party/prjxray-db/artix7 --part xc7a35tcpg236-1--fasm_file /home/michailm/phd/symbiflow/symbiflow-arch-defs/top.bit.fasm--top top | 11:56 |
| clay_1 | /home/michailm/phd/symbiflow/symbiflow-arch-defs/top_bit.v /home/michailm/phd/symbiflow/symbiflow-arch-defs/top_bit.v.tcl` | 11:56 |
| clay_1 | By putting all the arguements considered as necessary in the readme | 11:56 |
| *** rjordans has quit IRC | 11:57 | |
| clay_1 | this gives me `IndentationError: unexpected indent` | 11:58 |
| clay_1 | I was tried removing arguements that the error was coming fro untill It reached a point it was pointing to `-mfasm2bels` | 11:59 |
| clay_1 | Any Idea what I might be doing wrong ? | 11:59 |
| *** rjordans has joined #symbiflow | 11:59 | |
| sf-slack | <tmichalak> clay_1: this error indicates an indentation problem in the fasm2bels.py script or any scripts used by it. Do you have a more complete error message? | 12:15 |
| clay_1 | Is there any way I can get more complete error message ? | 12:17 |
| clay_1 | It sais ` File "<stdin>", line 1` and then an arrow pointing at `mfasm2bels` | 12:18 |
| clay_1 | but that was after removing some arguements. when I had them all on it was the same error on the first letter of the address of the `.tcl` file declaration | 12:19 |
| *** OmniMancer1 has quit IRC | 12:19 | |
| sf-slack | <acomodi> clay_1: One thing I suggest is to use the complete flow. Take an example from the `xc7/tests` (such as counter for example) and add your test there | 12:20 |
| clay_1 | `xc7/tests/counter` looks to me like having design source files in verilog and a constraint file. How will that help me run fasm2bels? So far by litghost's help I have run the `make dram_test_64x1d_bit_v` and from there I saw how it used the fasm2bels | 12:28 |
| clay_1 | but since then I had no sucess | 12:29 |
| sf-slack | <acomodi> clay_1: Yeah, I mean, you could reproduce the test you are trying to make with your own test. | 12:30 |
| *** az0re has quit IRC | 12:30 | |
| clay_1 | If I make a folder with my hdl + constraint file and then do a `make my_test` ? | 12:31 |
| sf-slack | <acomodi> exactly | 12:31 |
| clay_1 | Thank you, thats a nice idea! | 12:32 |
| sf-slack | <acomodi> You can for instance copy the counter test folder, name it as you want, change the hdl, CMakeLists and constraints file | 12:32 |
| clay_1 | The remaining problem will be that I will still not be able to ultimately go from a bitstream to bels and thats my actual goal | 12:33 |
| sf-slack | <acomodi> I see... the fact is that fasm2bels was meant to be a verification method of the whole flow, and mainly for debugging purposes. It is possible to use it for your goals for instance, but it may result in troubles you have experienced | 12:40 |
| clay_1 | oh that makes a lot of sense actually | 12:44 |
| clay_1 | First I was trying to use the `fasm2pips.py` from the project xray but it turned out to have a bug and thats why i turned to `fasm2bels` since it felt like its a superset of it | 12:46 |
| *** HEGAZY has joined #symbiflow | 12:47 | |
| clay_1 | I just realized that the conda environment used was not the one created by the `make all_conda`. I think I fixed that and now I will try everything all over | 13:32 |
| *** mkru has joined #symbiflow | 14:04 | |
| *** HEGAZY has quit IRC | 14:26 | |
| clay_1 | I finally had some progress! | 15:13 |
| sf-slack | <acomodi> clay_1: great to hear that | 15:14 |
| clay_1 | I think that I am able to run fasm2bels for the dram_test_64x1d independently of the make | 15:14 |
| clay_1 | then i tried to run with with only the arguements in the fasm2bels readme but I got errors which this time are much more constructive. I started tackling them from teh bottom | 15:15 |
| clay_1 | the error was `in append_ibuf_iostandard_params if "SSTL135" in iosettings["IOSTANDARD"]:` | 15:16 |
| clay_1 | so i added the arguement `--iostandard LVCMOS33` | 15:16 |
| clay_1 | and this got me the following error | 15:17 |
| clay_1 | `IOSTANDARD+DRIVE+SLEW settings provided for IOB_X0Y2 do not match their counterparts decoded from the fasmRequested: IOSTANDARD=LVCMOS33, DRIVE=NoneCandidates are: IOSTANDARD | DRIVE | SLEW |-------------------|--------|------| LVTTL | 16 | SLOW | LVTTL | 12 | SLOW | LVCMOS33 | 16 | SLOW | | 15:17 |
| clay_1 | LVCMOS33 | 12 | SLOW |` | 15:17 |
| sf-slack | <acomodi> If I recall correctly that is not an error, but a warning | 15:17 |
| clay_1 | oh that would be great ! | 15:19 |
| clay_1 | I will check the results and see if they make sense | 15:19 |
| *** Bertl is now known as Bertl_oO | 15:24 | |
| *** rjordans has quit IRC | 15:39 | |
| *** citypw has quit IRC | 16:01 | |
| *** az0re has joined #symbiflow | 16:13 | |
| *** balmlake has joined #symbiflow | 16:41 | |
| *** abeljj[m] has quit IRC | 16:49 | |
| *** abeljj[m] has joined #symbiflow | 16:49 | |
| *** Degi has joined #symbiflow | 19:04 | |
| Degi | Hi, I'm interested in the Summer of Code and have a PCIe interface for the ECP5 in nMigen in mind, though I'm open for other projects too | 19:06 |
| *** wallacejohn has joined #symbiflow | 19:15 | |
| *** clay_1 has quit IRC | 19:29 | |
| ZirconiumX | Degi: https://github.com/whitequark/yumewatari | 19:30 |
| tpb | Title: GitHub - whitequark/Yumewatari: 妖刀夢渡 (at github.com) | 19:30 |
| Degi | Yes I think I'd start with that as a base, though that project had no updates since 16 months | 19:32 |
| ZirconiumX | Useful reference if nothing else | 19:33 |
| Degi | (Also its in migen, I'd rather use nMigen, but yes, I learned a bunch from reading through the source code of that project a while ago) | 19:34 |
| ZirconiumX | Migrating from migen to nmigen is not that difficult as such | 19:59 |
| Degi | Yes, there's also litepcie which does the TLP | 20:09 |
| *** OmniMancer has joined #symbiflow | 20:15 | |
| mithro | Degi: working on a PCIe interface is not a *small* task, so any student taking that on will need to have to show a pretty strong understanding of what needs to be done -- the fact you have already seen yumewatari and talking about LitePCIe and TLP is a good sign... | 20:46 |
| *** miek has joined #symbiflow | 20:52 | |
| *** balmlake has quit IRC | 20:52 | |
| Degi | Hm I've read parts of the PCIe spec and to me what seems missing is the data link layer and I guess Yumewatari needs some work on (and translation to nMigen) | 20:56 |
| sorear | afaik the problem with people picking that up is that debugging anything that goes wrong will be a challenge without a very expensive oscilloscope | 20:59 |
| Degi | Hm you could use a second FPGA to analyze the data sent out by the SERDESs | 21:00 |
| -_whitenotifier-3- [ideas] drom opened issue #42: render LUT diagram - https://git.io/JvXBp | 21:01 | |
| *** balmlake has joined #symbiflow | 21:02 | |
| sorear | probably "put a LED on the board, and get very experienced with forming specific hypotheses / translating those hypotheses into HDL that turns on the LED if you're right/wrong" | 21:03 |
| Degi | I mean you could hook the second FPGA with UART or so up to a PC and log the data that arrives | 21:04 |
| Degi | Although I've done that a few times too, the dev board has 8 LEDs and its kinda OK for basic feedback | 21:09 |
| daveshah | One of the problems with yumewatari was tuning the analog parameters of the serdes (which have very limited documentation) | 21:28 |
| Degi | Hm could that be figured out with error rate tests and trial and error? | 21:29 |
| daveshah | Yes, but that could be quite a bit of work | 21:31 |
| daveshah | It would likely need several different motherboards to test too | 21:31 |
| Degi | Well I have 3 here | 21:33 |
| Degi | Hm could the values be reverse-engineered from something that lattice diamond creates? (I've never used that program yet) | 21:33 |
| *** balmlake has quit IRC | 21:33 | |
| daveshah | Yes, that provides a starting point, I think that's what wq did | 21:34 |
| Degi | I guess that's where all the undocumented values are from | 21:34 |
| *** HEGAZY has joined #symbiflow | 22:12 | |
| *** HEGAZY has quit IRC | 22:36 | |
| *** mkru has quit IRC | 22:47 | |
| *** futarisIRCcloud has quit IRC | 22:48 | |
| *** wallacejohn has quit IRC | 22:53 | |
| mithro | Degi: _florent_ in #litex has a lot of experience getting the high speed transceivers in FPGAs to play nice | 22:54 |
| mithro | Degi: He worked on https://github.com/enjoy-digital/usb3_pipe which should work on both ECP5 and Artix 7 | 22:55 |
| tpb | Title: GitHub - enjoy-digital/usb3_pipe: USB3 PIPE interface for Xilinx 7-Series / Lattice ECP5 (at github.com) | 22:55 |
| mithro | Degi: azonenberg has been doing a lot of work with getting things like eye diagrams out of devices too | 22:58 |
| Degi | Thanks! florent seems to be currently not in the channel though. | 23:09 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
