_whitelogger has joined #symbiflow
_whitelogger has joined #symbiflow
_whitelogger has joined #symbiflow
BryceSchroeder has joined #symbiflow
_whitelogger_ has joined #symbiflow
_whitelogger has quit [Ping timeout: 250 seconds]
_whitelogger has joined #symbiflow
GuzTech has joined #symbiflow
octycs has joined #symbiflow
<digshadow> GuzTech: hi, this is john about the prjxray issue
<digshadow> Are you at CCC by chance? Would make it easy to troubleshoot
<GuzTech> Hi John
<GuzTech> Unfortunately no :(
<digshadow> Ah okay, no worries
<GuzTech> I couldnt find any hotel with decent prices when the ticket sales began so...
<digshadow> :(
<GuzTech> Why does it always start so late?
* digshadow shrug
<GuzTech> Ah anyway
<digshadow> Okay, so to recap: 1) you are using latest master symbiflow/prjxray 2) you are using artix7 3) you are using vivado 2017.2
<digshadow> Correct?
<GuzTech> I searched in the tilegrid.json file and the IOB33S string appears 96 times, and yes I'm using Vivado 2017.2.
<GuzTech> Let me triple checck
<GuzTech> *check
<digshadow> FWIW we just made a small change to how the vivado environment works yesterday
<tpb> Title: introduce vivado wrapper by felixheld · Pull Request #416 · SymbiFlow/prjxray · GitHub (at github.com)
<GuzTech> Yes, thanks to that I can get to the next step which is to run the fuzzers.
<GuzTech> I could do it before, but then I had to first build with cmake and *then* source the vivado settings file.
<digshadow> out of curiosity, which OS are you using?
<GuzTech> I'm on the latest master, using Vivado 2017.2 and sourced the artix7 settings file.
<GuzTech> Arch Linux
<digshadow> good to know. FWIW I'm not sure if its been run on arch before, although I don't know of any specific issues that would cause
<daveshah> I've run parts of X-Ray on Arch before without problem
<digshadow> just a sec...
<GuzTech> Especially with Python related stuff I've noticed that running the newest packages can cause issues.
<GuzTech> For example, in the virtualenv requirements.txt, the latest pyjson5 (v0.4.5) does not build here.
<GuzTech> It's a known bug and the author actually has a pull request that fixes it, so I manually applied it and installed it here.
<GuzTech> Stuff like that happens on Arch with Python stuff unfortunately.
<digshadow> Okay, noted
<digshadow> ah I just remembered
<digshadow> this is the tilegrid fuzzer, not a fuzzer using tilegrid
<digshadow> it does not use the main tilegrid.json
<digshadow> so
<GuzTech> Yeah, I don't know what the problem is exactly.
<GuzTech> If it helps, the iob/build/specimen_001/top.v file is empty.
<GuzTech> Can't really synthesize an empty file and fix the IOB.
<digshadow> sorry getting distracted
<digshadow> I need to fix something, be back
<GuzTech> np
<digshadow> GuzTech: I'm preparing a debug output top.py for you to run
<GuzTech> sure
<tpb> Title: gist:8892dcb6b9f712d39e9d95729362c810 · GitHub (at gist.github.com)
<digshadow> I've added a small bit of debug info there
<digshadow> lets see if your DB path looks reasonable, and see if you can see *any* sites
<digshadow> When you looked at tilegrid.json before, which one did you look at? The correct one is 005-tilegrid/build/basicdb/tilegrid.json
<GuzTech> Yup that's the one.
<GuzTech> I posted the output on the gist
<GuzTech> It seems that it cannot find IOB33S
<digshadow> thanks looking
octycs has quit [Ping timeout: 260 seconds]
<GuzTech> I'll be gone for like an hour. Just so you know
octycs has joined #symbiflow
<digshadow> GuzTech: since I'm at CCC prjxray table I'm getting a lot of people coming up
<digshadow> I'll keep poking as I can, but I'll probably keep getting distracted
<digshadow> GuzTech: other things that could go wrong...maybe something wrong with the ROI. Could try echoing the variables on the comment: https://gist.github.com/JohnDMcMaster/8892dcb6b9f712d39e9d95729362c810
<tpb> Title: gist:8892dcb6b9f712d39e9d95729362c810 · GitHub (at gist.github.com)
<digshadow> if all of those look good, probably the next option would be to start playing with the python code to figure out why its dropping the entries
<digshadow> could you print out a list of all sites that it has and send that to me?
<digshadow> GuzTech: I can help with that if its not easy
octycs has quit [Ping timeout: 268 seconds]
octycs has joined #symbiflow
<GuzTech> I'm back.
<GuzTech> Yeah I understand, no worries :)
<GuzTech> Ok, the environment variables are different.
<GuzTech> The part is correct.
<GuzTech> echo $XRAY_ROI -> SLICE_X0Y100:SLICE_X35Y149 RAMB18_X0Y40:RAMB18_X0Y59 RAMB36_X0Y20:RAMB36_X0Y29 DSP48_X0Y40:DSP48_X0Y59
<GuzTech> No IOBs are in there as in your output, so that's probably the reason why.
<GuzTech> And echo $XRAY_ROI_GRID_X1 gives 9 instead of 0 as in your case.
<GuzTech> Which makes sense since they are not in the artix7 settings file.
<GuzTech> Weird, it is in the repo on github and I'm on the latest master.
<GuzTech> digshadow: Ok, I've found the issue. I used the download-latest-db.sh script which points to a repo of mithro. So the settings file I'm sourcing is his file which apparently is much older and does not contain the IOBs.
<GuzTech> I'm not getting the IndexError anymore, but I now I get some file not found error. I'm looking into that right now.
<GuzTech> Ok, so it's looking for the database/artix7/xc7a50tfgg484-1.yaml file, which does not exist in my checkout.
<digshadow> GuzTech: aaaaah yeah
<digshadow> that makes sense. Hmm we need to enhance README notes about that
* digshadow pounds on keyboard
<GuzTech> Yeah, now I understand why you guys need help :P
<GuzTech> Getting started can already be a pain :P
<GuzTech> Not that I'm blaming you guys, don't get me wrong.
<GuzTech> The only project that I got working immediately and easily was the Nyuzi processor. You check it out with the compiler, it builds everything with one script, and you can even run the processor in either qemu or iverilog and play Doom on it.
<GuzTech> My mind was blown at how much it did and how easy it was to set up :O
<digshadow> hehe
<digshadow> GuzTech: just made some updates to the readme to hopefully clarify. Working on modifying download latest to be more friendly
<GuzTech> ok
<GuzTech> I was looking how to make sure everything also works under zsh (which I'm using by default).
<GuzTech> Where is the database/artix7/xc7a50tfgg484-1.yaml created?
<GuzTech> Ah in 001-part-yaml *doh*
<GuzTech> That means that developers should first run that one, then if they want the others.
<digshadow> GuzTech: I'm working on fixing "download latest database" to be more friendly
<GuzTech> And the whole cmake build stuff is already in the toplevel Makefile. You can just do "make build" :)
<digshadow> GuzTech: I pushed the README update. Would you be interested in a PR to simplify it?
<digshadow> ah no
<digshadow> you don't want to do that
<digshadow> that will take a very long time
<digshadow> because it would build *everything*
<digshadow> oh just make build
<digshadow> I see
<GuzTech> Yup, just the cmake build stuff.
<digshadow> GuzTech: would you like to submit a doc update?
<GuzTech> Sure
<digshadow> cool :)
<GuzTech> If I can get everything working first :P
<GuzTech> I just did 001-part-yaml and then 005-tilegrid, but it still complains that it cannot find the part yaml file in the database.
perillamint has joined #symbiflow
<GuzTech> If I do make pushdb in 001-part-yaml manually, then it copies the yaml file.
<GuzTech> The Makefile for 001-part-yaml seems correct, and I am using bash for this.
<GuzTech> So just running make in 001-part-yaml does not work.
<digshadow> GuzTech: you did "make run" or "make"?
<digshadow> make run will push it, make will not
<GuzTech> I ran make.
<GuzTech> Looking at the toplevel Makefile in fuzzers/ it appears that I should run "make run" like you said.
<digshadow> GuzTech: I just merged in a cleaner "download latest"
<digshadow> if you give that a try it should preserve the settings.sh
<tpb> Title: Move settings.sh out of database dir · Issue #421 · SymbiFlow/prjxray · GitHub (at github.com)
<GuzTech> ok, I'll try it right now
<GuzTech> Ok, now that I've run "make run" in 001-part-yaml, I have run 005-tilegrid successfully.
<GuzTech> I'll try the latest download version
<GuzTech> Yup, with the latest downloaded version it also works!
<GuzTech> digshadow: What would you like me to add to the documentation?
<GuzTech> What you have now works just fine.
<GuzTech> I could add the pyjson5 compiler error stuff on Arch and Fedora.
perillamint has quit [Remote host closed the connection]
perillamint has joined #symbiflow
perillamint_ has joined #symbiflow
perillamint has quit [Read error: Connection reset by peer]
perillamint__ has joined #symbiflow
perillamint_ has quit [Ping timeout: 250 seconds]
octycs has quit [Ping timeout: 268 seconds]
BryceSchroeder has quit [Remote host closed the connection]
<digshadow> GuzTech: I pr that said to run "make build"
<digshadow> instead of cmake blah blah
tpw_rules has joined #symbiflow
adamgreig has joined #symbiflow
tpw_rules has left #symbiflow ["Textual IRC Client: www.textualapp.com"]
perillamint__ has quit [Ping timeout: 250 seconds]
perillamint__ has joined #symbiflow
<GuzTech> digshadow: Ok, I've updated README.md and created a PR.
BryceSchroeder has joined #symbiflow
perillamint__ has quit [Ping timeout: 250 seconds]
<BryceSchroeder> So I just ordered an LFE5UM5G-85F dev board. I got it intending to mess around with project Trellis and maybe (as time permits) even find some way to help out. However I just found out that the Lattice Diamond license that comes with it is a time-limited subscription?! Like, what the heck, it's already bound to that one device isn't it? So my question is
<BryceSchroeder> how effective is that time-limited part
<BryceSchroeder> do I need to wait until I am sure I have plenty of free time to mess with it
<BryceSchroeder> or can I go ahead and install it now without worrying about it expiring efficaciously?
<daveshah> Yes, it's a year's license. It's not device bound though, it works for any license locked Lattice part
<daveshah> I suspect Trellis will be pretty good in a year's time though
<BryceSchroeder> That's good - I'm mainly interested in the DSP tiles
<BryceSchroeder> as something to see progress in. I'm a medical student about to start my surgery rotation though, so I can't spend too much time working with this now;
<daveshah> Sure
<BryceSchroeder> I just kinda bought it because I heard about it on a 35CC talk.
<BryceSchroeder> *C3
<daveshah> DSP are still very incomplete
<daveshah> I would love help with them
<BryceSchroeder> So I guess it would be better to go ahead and do what I can now, and count on the FOSS tools being good by the time my license expires anyway?
<BryceSchroeder> Anyway, thanks for your input. I really appreciate the good work being done by this project. I had just been ignoring FPGAs as a thing until icestorm because I couldn't stand the development tool model in the FPGA world
<BryceSchroeder> I had like one FPGA dev kit in college and after what a pain in the butt its software and licensing was
<BryceSchroeder> I was like "fine, FPGAs just haven't been invented yet for my purposes."
perillamint__ has joined #symbiflow
octycs has joined #symbiflow
perillamint__ has quit [Ping timeout: 260 seconds]
octycs has quit [Remote host closed the connection]
tpb has quit [Remote host closed the connection]
tpb has joined #symbiflow
kraiskil__ has joined #symbiflow