IRC logs for #openrisc Monday, 2016-10-10

--- Log opened Mon Oct 10 00:00:50 2016
wallento_franck_ are you sure you have the most recent HEAD? I had this too and olofk fixed it02:41
_franck__wallento: I'll try to test it again this morning02:46
_franck__btw: -> error 40402:47
olofk_franck_: It's the stupid python package managment tools that messes things up03:24
olofkOne of the dependencies wants attrs >= 15.2 and another one wants attrs < 16.2 (or something like that)03:25
olofkpip/distutils/setuptools or whatever it's called these days, just happily install the latest version without caring about the second requirement03:25
olofkWhich is also a bit ironic that FuseSoC's internal package manager handles that, but you can't install fusesoc because python's package manager is too stupid03:26
olofkOh well03:26
olofkThe workaround is to manually install attrs 16.0.003:27
olofkI think 15.2.something also works03:27
olofkIf you got pip, you can run pip install attrs == 16.0.003:27
olofkOtherwise....good luck.03:27
_franck__olofk: actually I don't want to install fusesoc, just run it from the sources :)03:34
olofk_franck_: I think you can still do that, but you will have to call now instead03:35
olofkAnd you still need to get the dependencies03:35
olofkAnd if you go into the fusesoc repo and run "pip install -e .", then it will basically install a fusesoc link that points to the sources03:36
olofkI started updating the install instructions on the flight home yesterday03:48
olofkAnd it was good to see you again _franck_03:50
_franck__olofk: I was pleased to see you again. Too bad I couldn't stay longer03:56
_franck__my Ubuntu virtual machine broke after an upgrade so I can't test fusesoc now03:57
olofkThat sucks03:59
-!- Netsplit *.net <-> *.split quits: shorne, pecastro04:54
_franck__fusesoc is not what it used to be :)07:48
_franck__olofk: could you list somewhere what the dependencies are ?07:49
oleg-nenashevshorne_: seems you are in the international terminal, right?07:53
oleg-nenashev_franck_: There is a Docker image for fusesoc, which contains installation steps for Ubuntu08:03
oleg-nenashev_franck__: ^^08:03
oleg-nenashevSee oleg-nenashev/fusesoc-docker on github.
olofk_franck_:  python package management is horrible. :/08:21
olofkBut you need simplesat, which is likely only packaged with pip08:22
olofk_franck_: And I haven't got a clue what the error is saying08:28
_franck__thanks, let's see what is a Docker image08:32
_franck__ok I see all steps are described in the script08:34
oleg-nenashev_frank__  Yep. Not the best installation guide, but should be helpful :-)08:38
* oleg-nenashev cannot into IRC from his phone08:39
_franck__python install did the trick10:34
_franck__shorne_: you were right.10:36
_franck__thanks oleg-nenashev10:36
olofk_franck_: You got it working?12:41
SMDnoteHi, guys13:04
kc5tjaBTW, in case anyone missed it:  and
SMDnotekc5tja: Could you repeat, please, how many loots does your cpu consumes?13:08
kc5tjaSMDnote: The core alone, without any peripherals or bus bridges, currently takes up around 4200 LUTs with all the CSRs in place.  If I remove the CSRs, I can drop it down to around 3400.13:19
SMDnotekc5tja: I've just got mystorm board, and ppl claim it has ~7500-8000 luts in it, so I wonder what I could fit into it using yosis13:20
kc5tjaIt should fit.  It still lacks interrupt support, but that shouldn't be too hard to add.  However, it should fit.13:22
kc5tjaMaking it fit with peripherals will be the biggest concern though: it does take up more than half of the available LUTs.13:22
SMDnoteand you have pipelined cpu?13:22
kc5tjaNo.  My last attempt at making a pipelined CPU resulted in it taking up almost 20,000 LUTs.13:23
kc5tjaThis CPU is slow (average 4 or 5 CPI), and it does not support user- or supervisor-modes, so it will also not run Linux.13:24
kc5tjaIt's literally best to think of this CPU in the same way as a 64-bit wide Z-80.13:24
_franck_olofk: yes it worked. I think (as wallento said) I didn't have the most recent HEAD when I first tried "python install"13:29
SMDnoteolofk: Are you going to publish a list of people attended to orconf somewhere?13:32
SMDnoteolofk: Also I wonder has anyone asked intel for sponsorship for orconf?14:39
olofkSMDnote: I think we plan to send out the list of attendees to the people who attended, but I need to confirm this with the rest of the team15:05
olofkAnd it was great meeting you :)15:05
olofkRegarding Intel, they (or at least Altera. Not sure if they had merged yet) did sponsor last year15:06
SMDnoteolofk: I mean I could ask them if they don't have solid denial position on funding orconf15:23
SMDnoteolofk: It was great meeting you too!15:24
olofkSMDnote: That would be fantastic. For us it has just been that we haven't been able to find the right people to ask15:35
--- Log closed Tue Oct 11 00:00:51 2016

Generated by 2.15.2 by Marius Gedminas - find it at!