IRC logs for #openrisc Friday, 2014-02-14

--- Log opened Fri Feb 14 00:00:10 2014
--- Day changed Fri Feb 14 2014
blueCmd23:59 <@madrazr> blueCmd: when he/she visits any GSoC 2014 page, the existence of "My Profile" link indicates he has a profile for GSoC 201400:00
stekernmorning all00:54
poke53281good morning01:07
poke53281a little bit early01:08
stekernpoke53281: nah, it's already 08:30 here01:29
stekernwell needed vacation, avoiding the snow and cold in finland01:32
poke53281come to Vancouver. No Snow but lots of rain.01:33
poke53281I came back from Germany two days ago01:33
stekernI swept through Germany (Frankfurt) on the way here and on a business trip a week earlier (Munich)01:37
stekernit was less snow than in finland, also warmer, but not quite the climate that's here ;)01:37
poke53281Well, next week I will have a trip to Saskatoon. They have currently cozy -18┬░ which is pretty warm for February.01:39
blueCmdstekern: what's the stock market prices over there?01:44
blueCmdstekern: since it's only 17:44 here maybe I can do some good deals01:44
poke53281I am more curious if it is worth investing in Bitcoins.01:46
poke53281I have access to several cluster but I think they would kill me if I start Bitcoin mining on them.01:47
blueCmdwith ASIC mining you're dead in the water :)01:56
blueCmdstekern: do you want to mentor gsoc?02:08
blueCmdolofk: maybe?02:08
hansfbaierpoke53281: Bitcoin mining on CPUs and GPUs is pointless now02:22
hansfbaierOnlu ASICs rule there02:23
poke53281A few weeks ago my favorite magazine ct brough an article about bitcoin mining. The letters to the editors printed two weeks later were clear. "Stop this bullshit." It is such a waste of energy and resources that it should be forbidden.02:26
stekernblueCmd: what does mentoring involve? I can do it if there's a demand for it04:16
* wkoszek is actually looking for people to split $15k cost with for eXtremeMiners miner04:27
wkoszekblueCmd, stekern I'm wkoszek in melange04:32
wkoszekI think I registered now.04:32
blueCmd00:21 < wookey> blueCmd_: you need to think of a task which is small enough for a student to do in 3 months (including gettgin up to speed), and  well defined enough to measure sucess/failure04:33
blueCmd00:24 < blueCmd_> fair enough04:33
blueCmdstekern: ^^04:33
blueCmd00:29 < wookey> and be prepared to invest about as much time as it would take to do yourself (unless your student is unusually good)04:33
blueCmd00:30 < wookey> but obviously if you get someone enthused enough to stick around then that a big win in the long term. And it's fun :-)04:33
blueCmdwkoszek: yes, thanks!04:33
blueCmdolofk: since you wrote 'Every year I'm thinking about submitting stuff like this to GSoC, but I guess it would need a backing company/foundation as well as a few mentors' I'll assume you are voluntaring to be one04:38
blueCmdgreat, thanks!04:38
blueCmdwkoszek: do you want to mentor?04:46
blueCmdLoneTech: juliusb_: how about you? you want a student to care for?04:46
blueCmd(promise to feed him/her!)04:47
wkoszekI can mentor 1-2 ideas. I'll be mentoring for FreeBSD too.04:48
wkoszekWe could do FreeBSD/openrisc project.04:48
blueCmdThat would be soo sweet :)04:48
wkoszekDoes OpenRISC have LLVM backend?04:48
blueCmdwkoszek: sort of04:49
blueCmdbut that is absolutely something we could have as a project to turn into a 'absolutely!'04:50
blueCmdwkoszek: do you see 'OpenRISC' under 'My dashboard' now?04:50
wkoszekThis project is aimed at developing embedded RISC microprocessor architectures and open source RTL implementations. Since its beginning in 1999, the project has attracted anyone interested in CPU development and covers everything from ISA definition to RTL description writing and testing to simulator, tool chain and operating system porting04:52
wkoszekDid you guys get accepted previously?04:52
blueCmdI have no idea actually, I tried finding any referenses but to no avail04:53
wkoszekI'd try to make it more SW-sounding description.04:53
vxenow that crowd sourcing is trendy, are there any openrisc-based projects to back? I'm looking at the donation page on opencores wondering when it's going to happen :)04:53
wkoszekSince GSOC = SW project.04:53
blueCmdwkoszek: It was copy paste to have the application submitted, feel free to improve it!04:55
wkoszekThis project is aimed at developing embedded operating environment,05:05
wkoszekincluding OpenRISC (microprocessor), OR1KSIM (OpenRISC simulator), OS (Linux05:05
wkoszekfor OpenRISC), compiler (gcc and LLVM for OpenRISC). It includes05:05
wkoszekVerilog/VHDL IP core too. Since its beginning in 1999, the project has05:05
wkoszekattracted anyone interested in hardware/software system development and is05:05
wkoszekin great need for skilled system/embedded engineers.05:06
wkoszekblueCmd: What do you think?05:06
wkoszekI just don't know if Google is funding HW stuff.05:06
wkoszekblueCmd: Do you guys have idea list?05:10
blueCmdnow we do05:11
blueCmdwkoszek: I approve! good description05:11
wkoszekblueCmd: OK, so I put this page in Google too.05:13
wkoszekblueCmd: The list looks pretty crappy.05:14
wkoszekblueCmd: pastebin me copy&paste desc. of idea05:16
wkoszekblueCmd: I'd submit 2-3 more ideas.05:16
blueCmdwkoszek: Maybe it does. Perhaps I've grown used to how the wiki on opencores looks :)05:19
blueCmdwkoszek: what do you mean with pastebin desc. of idea?05:20
wkoszekblueCmd: I want to add stuff to Wiki, but I don't have account.05:22
blueCmdwkoszek: Ah05:24
blueCmdwkoszek: is that what you wanted?05:25
blueCmdwkoszek: nice ideas!06:02
blueCmdwkoszek: added, wiki format can only get you so far though06:07
wkoszekblueCmd: I don't know about this one. I think it's too small for GSOC>06:16
wkoszekblueCmd: All your ideas are OK, but you'd have to elaborate on them, so that people know what they'd be doing.06:24
wkoszekblueCmd: Plus Google is actually reading this stuff.06:24
blueCmdwkoszek: I will rework them, for the moment I just want them out there06:24
blueCmdwkoszek: 'Plus Google is actually reading this stuff.' hm?06:25
wkoszekblueCmd: Tomorrow is the application deadline. Not sure how it works, but I guess they'll be starting to pick orgs by reading the ideas lists and decided which are worth a slot.06:30
wkoszekjuliusb_: Are you going ot be a mentor?06:30
wkoszekjeremybennett: Are you going to be a mentor?06:30
wkoszekolofk, stekern Would be good to have more mentors to share the load. :)06:31
blueCmdwkoszek: I don't see them reading the applications over a weekend06:31
wkoszekblueCmd: Don't you have to re-paste the list tomorrow? For Google Code In you had to transfer the ideas from the website to Melange. But I don't remember at which stage you have to do this.06:34
* wkoszek -> his book06:34
blueCmdwkoszek: there was a thing where you linked a webpage for the ideas06:36
blueCmdso I don't think so06:37
olofk_blueCmd: Reading up on the backlogs. Yes, I would love to do some mentoring, but I'm a bit worried about having the time to do it07:17
blueCmdolofk_: we'll share it worst case07:17
blueCmdit's encouraged to be multiple ppl per student anyway07:17
olofk_Great. I really like this to take off07:18
blueCmdI think this won't be a problem since this channel is quite active07:18
blueCmdyes, thank wkoszek for kicking us in the behind :)07:18
olofk_Now I'm just freaking out because I have had so many ideas, but now I can't remember any of them07:20
olofk_Where can I find the current submission page?07:21
blueCmdolofk_: IRC-logs! :P07:22
olofk_ah yes. I thought all the content there had been copied to some other place07:23
blueCmdolofk_: that page is new, I moved stuff into that page07:23
olofk_Looking at it now. Looks good07:23
blueCmdwell, it doesn't really - but it's something :P07:24
poke53281blueCmd:  As far as I know Java needs libffi too, not only Python07:24
olofk_hmm.. OpenRISC System Creator and OpenRISC Package Manager both look very much like what orpsocv3 is doing07:25
blueCmdpoke53281: I wasn't sure, feel free to update the page07:25
blueCmdpoke53281: I'm very greatful for any help :)07:25
poke53281You did already a great job.07:25
blueCmdthanks, don't forget to thank wkoszek - he wrote the big projects07:26
olofk_Hey, no one has mentioned porting Windows CE to OpenRISC. That should be our number one priority07:26
blueCmdolofk_: will you mentor it? and pay the psycriatric bill?07:26
blueCmdolofk_: I heard that plan9 was GPL now, maybe port it?07:27
poke53281Build a QEMU image to provide a native building environment. Or an image which is using the chroot qemu-user-mode. This was my plan last year.07:27
olofk_poke53281: Yes. That's a good idea. Anything that lowers the barrier of entry to the project is needed07:28
blueCmdpoke53281: that will be trivial when or1k is a supported Debian architecture07:28
blueCmdbut in the nearest 10 years, I agree that we need something until then07:28
poke53281Then we should try to compile systemd. It failed last time I tried it. But mybe with glibc.07:33
blueCmdI have a deb with systemd for openrisc07:36
blueCmdI haven't tried it though, but it built after I poked it a couple of times07:37
blueCmdoh well, time to sleep - busy day tomorrow. I'll be flying back to europe so don't expect too much communication from me07:39
poke53281As far as I remember the problem was uClibc07:40
poke53281I have to sleep as well now.07:40
stekernnetbsd/freebsd for openrisc sounds like an interesting project08:57
jeremybennettwkoszek: I wasn't planning on being a mentor, but if it helps, tell me what I need to do.08:59
stekernysionneau: I bet you would be an asset in that if someone wants to pick that up09:00
ysionneauI could certainly help09:00
stekernI've never got my head around the different *bsds, how does freebsd differ from netbsd (in complexity)09:01
ysionneauI honestly only read some low level parts of netbsd kernel so far, no other bsd09:02
ysionneauI just read what I'm interfacing with: the low level MD part09:02
ysionneauand the low level MI part which talks to my MD part09:02
stekernMD, MI?09:20
ysionneauMachine Dependant09:21
ysionneauMachine Independant09:21
ysionneauarch specific/not arch specific09:21
stekernah, ok09:24
LoneTechblueCmd: while I do love teaching and joint learning, at this point I don't think I can take that on. sorry.09:37
olofk_I should probably write up some proposal for versioned dependencies for orpsocv3 core files.09:56
-!- FreezingAlt is now known as FreezingCold10:28
olofk_Seems like there was some OpenRISC coverage at FOSDEM this year after all.11:54
jeremybennettolofk_: There was a lot of hardware around this year. LoneTech was there.11:55
olofk_Didn't know that there was an OptimSoC presentation11:55
jeremybennettI saw that was there, but because we were busy with our own dev room, didn't have time to see it (I only saw one other presentation all weekend)11:56
LoneTechjeremybennett: hoping to get some experiments done with the power stuff next week. sorry for the delay13:06
LoneTechbeen rather a lot going on13:07
jeremybennettnp - look forward to hearing more on the mageec mailing list13:34
joaocfernandesi am a little lost, i have built both the stable and dev toolchains, if i try to load a simple hello world application to or1ksim i get this error (elf is not an executable). what can be wrong?16:32
stekernjoaocfernandes: you are not compiling it as an executable, but as an object file (remove the -c)17:20
joaocfernandesstekern, thanks :)17:27
olofkI just released orpsoc 3.222:12
olofkTar ball here if you don't want to clone it from github22:12
olofkIt would be cool if someone wanted to build a .deb for it.22:14
olofkI'll do an ebuild for it22:14
wkoszekYou guys should add more ideas and we'd mentor people here.22:14
olofkwkoszek: Yes, I agree. I just haven't come up with any more good ideas22:15
wkoszekorpsocv3 - how it works?22:16
olofkwkoszek: It works quite well now22:16
wkoszekTypically people from OPenCores solve it by: copy&paste IP core from somewhere else and commit to New Repository22:17
olofkOr do you mean in more detail? :)22:17
olofkwkoszek: orpsocv3 is more like a package manager/distribution22:17
wkoszekSo if you fetch 3 SOC projects, you have wb_muxbus copied 3 tiles22:17
wkoszekSo how do you add new WB IP core there?22:18
olofkYou create a new .core file (a core description)22:18
olofkThe .core file tells where to get the code, which files it contains, which dependencies it has on other cores and stuff like that22:19
olofkTop-Ãlevel generation is not in the scope right now, butthere are some ideas for that as well22:20
olofkMany cores can be simulated individually if the have a test bench22:21
wkoszekPerfect. So I suggest you co-mentor this idea.22:21
olofkwkoszek: Yes, definitely22:21
wkoszekI'd like tool like that to dump the schematic to .dot or some better program. and json22:21
wkoszekSo that we could have customized editor/GUI which would give you working system22:22
olofkwkoszek: Have you considered IPXact for that? It's a bit clumsy, but it is a standard after all22:22
olofkI'm working on an IPXactwriter for some of the cores, so that I can pull in the SoCs in Kactus2 and generate the top-level stuff there22:23
olofkGot to sleep now, but we should talk more about this. It sounds like you are looking for the same things that I am22:24
ysionneauMigen allows to connect wishbone masters/slaves and generates the arbiter and the interconnect for you fyi wkoszek22:25
wkoszekolofk: Makes sense.22:30
wkoszekAll I want is 1 tool which I install with 1, at most 2 commands.22:30
wkoszekAnd which would configure Verilog stuff for me22:30
wkoszekSo that I don't spend 2 weeks debugging RTL to get Hello world to work22:31
wkoszekSo if tool like that exists, I'd like to know about it.22:31
wkoszekjeremybennett: Do you know anything about LLVM?22:55
wkoszekWhat are other un-ported RTOSes ?22:55
wkoszekAnother idea would be to write Tcl stuff, so that Wishbone cores would be drag & dropped into Vivado.22:56
stekernwkoszek: what do you want to know about LLVM?23:02
wkoszekstekern: Whether OpenRISC port exists, what's its state and whether we could have this as a part of GSOC project.23:14
wkoszekstekern: e.g.: if 64-bit compiler doesn't exists, could we have 64-bit port of LLVM for OpenRISC proposed.23:14
stekernyes, there is an existing OpenRISC port:
stekernthe state is that it is about as "complete" as gcc, sans TLS support that was added after I last worked on it23:16
stekernand it hasn't been synced with upstream for a while (but simoncook have done that a bit more recent than what's in the main repo)23:17
wkoszek64-bit is suppported?23:18
wkoszekI think it was called OR64 or something like that.23:18
stekern64-bit would be interesting, but I don't know i the LLVM angle of it would be the first step, we are missing implementations23:18
wkoszekYou mean we miss RTL to verify 64-bit OpenRISC?23:18
stekernyes, and simulators23:19
wkoszekDoes Or1ksim support 64-bit OpenRISC?23:19
wkoszekMaybe this is the project for GSOC>23:19
wkoszekDoes QEMU support OpenRISC?.23:19
stekernand binutils (although there are some work in binutils towards 64-bit support)23:19
wkoszekMaybe this is the next project?23:19
stekernno 64-bit in or1ksim, and yes, there is an upstream openrisc qemu port (but 32-bit only)23:20
wkoszekI'd target 100% energy to LLVM, given it's stable enough to compile and run FreeBSD.23:20
wkoszekOk, so 64-bit support for QEMU could be the project. Do you agree?23:20
stekernso any of that would be the first step towards any 64-bit work23:20
stekernI think starting with or1ksim would be better23:21
stekernit's the golden reference23:21
wkoszekYou guys didn't switch to QEMU?23:21
stekernor1ksim is a simulator, qemu is an emulator23:21
stekernboth are used23:21
wkoszekOr1ksim is cycle accurate?23:22
wkoszekDoes it model pipeline stages?23:22
stekernlast I worked LLVM, I could compile the Linux kernel, uclibc and busybox, so it should be about stable enough to get a small freebsd system going23:23
stekernno, it's not that accurate simulator, it's a functional simulator, but it ends at that23:23
wkoszekstekern: So it's basically same as QEMU23:27
stekernyes, but they serve different purposes, or1ksim should be correct, qemu fast23:28
poke53281Bu at the moment, QEMU is almost as slow as or1ksim.23:44
poke53281I have repository with patches which speeds up QEMU by an order of magnitude or so.23:45
stekernpoke53281: have any set of your patches been applied to upstream?23:49
stekernI mean, of the ones you've sent23:50
poke53281yes, some of them, but not all.23:50
poke53281I wanted to submit V2 of the patches but coundln't find a problem. Then I was too busy with other stuff the last two months.23:51
poke53281But unfortunately the main patch to speed up everything is not upstream.23:51
stekerndo you agree with me that it would make more sense to start playing with 64-bit support in or1ksim than in qemu? (or should it be jor1k? =))23:51
poke5328164-bit support would made more sense or1ksim.23:52
poke53281No, forget jor1k. 64 Bit in Javascript is ugly because it is not supported.23:52
stekernhehe, ok, it was not a serious suggestion, but then we have a serious reason to not do it ;)23:53
poke53281But in the end I think it does not really matter if you put 64 Bit support in QEMU or or1ksim.23:53
poke53281Next step is libdiff, which should be very easy to support. Then again QEMU, and then I hope that we can build with bluecmd something like a tiny debian repository.23:55
poke53281at least this is my plan for the year 201423:57
stekernsounds like a great plan =)23:58
--- Log closed Sat Feb 15 00:00:36 2014

Generated by 2.15.2 by Marius Gedminas - find it at!