IRC logs for #openrisc Tuesday, 2013-03-12

adam`Hi guys.  Just watched some of the OpenRISC PM 2012.  If you are in here Jeremy I want to thank you for being so awesome.  =)04:03
glowplugHello.  =)04:33
GentlemanEngineeHow fares the battle?04:57
glowplugI wish I knew.  I just joined the channel for the first time.  =(05:01
GentlemanEngineeI am in the same boat.05:02
GentlemanEngineeWhich of us is the captain?05:02
glowplugBased on the openrisc pm 2012 at least julius and jeremy I recognize.05:04
GentlemanEngineeYou apparently know more than I...05:05
GentlemanEngineeI was looking at contributing.05:05
GentlemanEngineeHowever, I have yet to have a reply to either emails or forum posts.05:05
glowplugSimon cook also is in the list.  I think thats everybody in the channel who attended.05:06
glowplugI would also like to contribute but it seems difficult for mortals.  O.o05:06
GentlemanEngineeIt was my understanding that the group was open to contributors...05:07
glowplug*difficult for mortals to comprehend*05:11
GentlemanEngineeI am no expert. However, I hope I can muddle on through...05:12
GentlemanEngineeI was looking at the Bugzilla for OR1200 RTL.05:13
GentlemanEngineeI saw bug #91.05:13
GentlemanEngineeIt is somewhat stale, and wished to confirm that it had not yet been dealt with prior to delving in.05:14
glowplugYou might want to hang onto your pants for the 2k version CPU because they are supposedly completely redesigning the core.05:18
glowplugWhich will bring a whole host of new bugs and needed work.05:20
GentlemanEngineeAre bugs no longer being dealt with on the previous core?05:20
GentlemanEngineeAlso, how does one become involved in the 2k core design?05:21
glowplugI assume that they will be.  But with no production ASIC based on 1k it's my opinion that those bugs can probably be ignored.05:21
glowplugThat I don't know yet.  And mostly why I'm hanging out in here.  =)05:21
GentlemanEngineeIt could be a springboard to bring myself up to speed...05:21
GentlemanEngineeIf you do mind my inquiry, what is your involvement with the project?05:22
glowplugI watched the entire 2012 openrisc meeting on youtube then joined this channel.  8)05:25
GentlemanEngineeAh! I was unaware of the youtube channel.05:26
GentlemanEngineeDo you have a URL?05:26
glowplugThis is the video I'm watching right now.  The description has the rest.05:26
GentlemanEngineeI thank you kindly.05:27
GentlemanEngineeWhat is your background on this?05:27
glowplugThat was my full disclosure of my background in the openrisc project specifically.  My personal background is amature software/hardware developer.05:29
GentlemanEngineeWhat manner of hardware, if you do not mind the inquiry?05:29
glowplugComputer machining, circuit design, hobby level work.  =)05:32
GentlemanEngineeBy computer machining, are you referring to CNC?05:32
glowplugRight.  CNC, 3d printing, robotics ect.05:34
GentlemanEngineeI have some experience with the third.05:34
GentlemanEngineeI have been eyeing the first two.05:34
glowplugIf you have experience with robotics then CNC and 3dprinting are trivial.05:35
GentlemanEngineeWhen I have some time (ha!), I have wished to build a RepRap.05:36
glowplugIf you don't have an immediate need for a repid prototype I would highly suggest waiting to build a printer.  There is going to be a massive influx of SCARA and DELTA based printers within the next year that are vastly superior to cartesian machines.05:40
GentlemanEngineeThat surprises me little...05:40
GentlemanEngineeI likely will not have the time to build it for some time, anyhow.05:41
GentlemanEngineeAre you based in North America, or Europe like much of group appears to be?05:43
glowplugI'm primarily interested in orisc for the implimentation a 6-axis articulated robot.  North America.  =)05:43
glowplugThat's unfortunate because this is the future of all technology.05:44
GentlemanEngineeThat *does* sound interesting.05:44
GentlemanEngineeAt the moment, I am attempting to stay one shovel ahead of the snow.05:45
glowplugHonestly it seems like most of us have spoons not shovels...05:46
GentlemanEngineeI do not believe a spoon would have great efficacy in battle against the snow I have...05:47
GentlemanEngineeHave you looked much at the internal architecture?05:49
glowplugBriefly.  I'm familiar with Harvard chips.  Still mostly interested in learning the details of the 2k architecture as information is released.  Mostly for the improved modularity and scalability (which the 1200 isn't so good at).05:50
GentlemanEngineeI am curious about this 2k as well.05:53
GentlemanEngineePerhaps inspecting the 1200 would provide you with some inkling as to the internals of the 2k...05:54
glowplugThe basic operation of a Harvard chips are all comparable.  What takes the most time to understand are the details of the implimentation.05:56
GentlemanEngineeHowever, their design methodology, and many of the modules would likely be reused.05:58
glowplugThis is true.  We could probably benefit from learning Verilog and C also.  Doesn't seem to be enough time for it all.  =)06:01
GentlemanEngineeI know C, and a little Verilog.06:02
glowplugYou are in a much better position to contribute immediately then me it seems.  8)06:03
GentlemanEngineeDid you not use C in your robotics work?06:04
glowplugLazy.  Python.  O.o06:04
glowplugThere is an OR2K discussion on Youtube.  The audio quality isn't very good but it's good information.06:06
GentlemanEngineeC is really worth learning.06:06
GentlemanEngineeIt is simple, elegant, and constructive.06:06
GentlemanEngineeWhat is that video?06:06
glowplugA discussion of the 2k architecture.06:08
GentlemanEngineeI shall view it later.06:08
GentlemanEngineeYou have obviously done some more research than I.06:08
glowplugProbably not much more.  Trust me.  8)06:09
GentlemanEngineeI can recommend "The C Programming Language"06:12
GentlemanEngineeIt is written by Kernighan & Ritchie, who created C.06:12
glowplugPython can compile down to C using PyPy and Cython.  I may eventually learn C but it06:13
glowplugIt's not a priority.06:13
GentlemanEngineeI don't know if I would attempt embedded programming in any other language--except for assembly, of course...06:14
glowplugNo reason to waste your time in assembly when python can run at near realtime.  Just get a superfast SoC for ~$5 and install linux on it.06:18
glowplugOur time is precious, cpu cycles are cheap.  =)06:19
GentlemanEngineeEmbedded is often accomplished without an OS, or something small like FreeRTOS.06:19
glowplugWeb applications are often accomplished without a framework.  But it's hardly an efficient use of anyones time.06:20
GentlemanEngineeI have written several simple web pages in that manner.06:21
GentlemanEngineeFor something simple, it is often not worth the effort in bending a framework to one's whim.06:22
glowplugThat is simply a case of using the wrong size framework.  When I design simple websites for friends and family I use html5 boilerplate and twitterbootstrap.  This is the propper sized framework for a simple informational site.06:23
glowplugStarting with raw html and CSS the same site would take me 20 times longer.  Extremely slow.06:23
GentlemanEngineePerhaps I have not spent the appropriate effort in acquainted with various frameworks.06:23
glowplugFrameworks is just a fancy way of saying "letting someone else do 80% of my work".  The 80/20 rule ect.06:24
GentlemanEngineeI am familiar with the concept.06:24
GentlemanEngineeI have been looking at the EPICS framework for control systems.06:25
glowplugIn many ways Python is a framework when applied to embedded development.  Because it does most of your work for you at the extremely small cost of extra hardware (32-bit SoC vs 8/16 bit uC).06:25
stekernglowplug and GentlemanEnginee: we are all mortal here (I can feel my mortality especially, with a flu coming down :()06:27
stekerni.e. contributors are more than welcome, no matter at what level06:27
stekernI was at the pm too btw ;)06:27
GentlemanEngineeI hope that I do not have to suffer the afflictions of mortality...06:27
glowplugThanks for trying to make us feel better stekern.  But I still know the truth.  =)06:28
GentlemanEngineeWhat initial steps are necessary to commence contributing?06:28
glowplugEPICS looks very interesting.  It's unfortunate that it's only soft-realtime it's unsuitable for robotics.06:29
GentlemanEngineeI suppose that would depend on your robotics requirements...06:29
glowplugstekern you are Stefan K?06:30
glowplugI need sub 4 microsecond latency unfortunately.  Soft-realtime doesn't qualify.  =(06:31
GentlemanEngineeIt might for your CNC and 3D Printing, though...06:31
stekernor2k is (still) in a very early stage of development, most of the development is still concentrated on or1k06:32
glowplugThe latency threshhold for that is sub 10 uS.  Really only LinuxCNC is suitable.06:33
stekernas for the or1200 development, julius (and I) are working on mor1kx, which is meant to be a "better" (cleaner/more modular/faster/smaller) implementation of or1k06:33
GentlemanEngineeI was looking at the Bugzilla for OR1200 RTL, and saw Bug #91.06:33
glowplugSo you are Stefan K.  I wanted to thank you for your incredible work.  =)06:34
stekernyes, I'm Stefan =)06:34
GentlemanEngineeIt appeared a bit stale, and was wondering if it was still worth looking at.06:34
stekernbut or1200 haven't been left to die, it's still been around a lot longer and is probably a lot more stable than mor1kx is yet06:35
GentlemanEngineeIs the bugzilla still being used?06:35
stekernfor or1200 development, yes06:36
glowplugWhat about for mor1kx?06:36
stekernmor1kx lives here:
glowplugI just puled it up.  =)06:37
glowplugGoogle is our friend.06:37
glowplugScaling sideways you guys have already figured out mostly (with regards to mor1kx).  What do you think the cores future is in terms of performance scaling per-core?06:39
stekernwe are a bit spread out, it's a long story about peoples preferences for tools and work flows, but we have at least agreed on keeping as the common place for all development06:39
glowplug^Already had that one bookmarked.  8)06:40
stekernglowplug: hmm, not sure what you are asking06:41
glowplugOn Altera I could probably expect ~60 mips.  Do you expect ~1000 mips on an asic at say 1ghz?06:42
stekernif you only execute out of 1-cycle memory, yes06:43
stekernpenalty for slow memory accesses grows with cpu freq (naturally)06:44
glowplugAny plans to tackle that in mor1kx?06:45
glowplug(by the way 1k mips is absolutely amazing if it is achievable with the existing architecture).06:46
stekernthe problem of slow memory is out of the scope of a cpu core (apart from having caches, which it already has)06:46
stekernI'd expect any RISC architecture to scale with the cpu-freq in regards of MIPS if you disregard memory accesses06:47
glowplugIt seems like there really is quite a lot more reading that I need to do.  This is what I meant about the mortal thing.  =)06:48
glowplugUnfortunately I'm in the States and it is getting pretty late.  I hope to catch you guys in here tomorrow.  =)06:49
stekernThe one that has learned most before he dies win, that's the game we mortals play ;)06:49
GentlemanEngineeUnfortunately, I must also arise in the morrow.06:50
glowplugThere is a lot of truth in that.  Talk to you later Stefan, and Gentleman.  :)06:50
GentlemanEngineeI thank you kindly, stekern.06:50
stekernsee you06:54
mboehnerthi there! is there anybody who has experiences with the advanced debug system (and ubuntu 12.10)12:01
_franck_mboehnert: what is your problem ?12:41
juliusbGentlemanEnginee, glowplug: Hi, thanks for your interest - as stekern said, you don't need to be an RTL wizard to help us out in this project13:28
juliusbbut it'd help to get up to speed with what we've been doing lately13:28
juliusbstekern: I was thinking of an idea which maybe Johan Rilegard or jeremybennett mentinoed at the meeting last year, which is a catch up skype conference call every now and again for people to talk about what they'ev been doing13:29
juliusbsince the meeting _franck_ has done a heap of awesome work on GDB and the sim13:30
juliusbit'd be cool to have people like Peter Gavin let us know what they've been up to etc13:30
stekernjuliusb: I think it was one of the polish guys that brought up that idea13:39
stekernbut regardless of whos idea it was, it could be a good idea13:39
-!- Netsplit *.net <-> *.split quits: juliusb, asm, forkG13:43
juliusb_woah netsplit14:05
juliusb_yeah I agree14:05
juliusb_i'll post to the maililng list suggesting it14:05
stekernproblem could be finding a time that suits everybody14:06
stekernI vote for 4:00 am EET ;)14:06
stekernat all other times I will have screaming kids around me (or be at work)14:07
juliusb_(says the single male)14:07
juliusb_the thing that I may have to occupy me is the pub14:08
amsjuliusb_: Whisky.14:08
* juliusb_ nods at ams14:24
juliusb_or that14:24
-!- juliusb_ is now known as juliusb14:24
jeremybennettjuliusb: That's a good idea. We also ought to be planning this year's opencores meeting. We'd be quite keen to host it in the UK (with Embecosm support). Possibly in London or Cambridge.15:31
mboehnert@_frank_ : it seems like ads does not compile right. i found an old binary which is running well, but a new one (svn/trunk) seems to detect a wrong jtag id and therefore does not connect to the openrisc-system15:32
juliusbjeremybennett: I like that idea.15:41
_franck_mboehnert: you may also want to give a try to openocd15:42
_franck_what is your openrisc setup (tap, debug unit, jtag adapter)15:43
mboehnerti use the vanilla openrisc1200v3 debug interface, ads (trunk) jtag tap and the OpenRISC USB-JTAG Degugger. i am aware of the breakpoint-bug in this configuration. other than that everything was fine until switching over to ubuntu 12.10 and recompiling the whole (gcc-)toolchain/simulator/ads15:46
mboehnertand i also use the jtag bridge supplied by ads...15:48
_franck_it's been a long time I haven't use advance debug gdb server. You should post on the openrisc forum15:56
mboehnertthank you, i'll post there and i will look into openOCD16:01
-!- asm_ is now known as asm17:56
glowplugGood morning.  I think Skype is a good idea.  Also a good quality conference microphone because I can't hear anything in the YouTube videos.  O_O18:23
glowplugAlso in regards to the conferences.  An overlay of the slides, and a microphone for the presenter instead of a cam-mic which tends to get what everyone else is saying very clearly and not much of what the presenter is saying.  8)18:26
glowplugNot a big deal for you guys but for outsiders looking to contribute it is hard to get information when both the YouTube videos and the website are not very friendly.18:27
juliusbglowplug: I agree with your suggestions - they would make for an improvement but it was an issue of lack of resources18:37
glowplugWe need to get some donated equipment located in that case.  =)18:38
glowplugI'm watching your mor1kx presentation right now.  Do you guys have any PDF's for it yet?18:39
juliusbyes I think they were put somewhere18:52
glowplugIf I get some spare time I will try and build a pair of microphones from this project and send them out to you guys before the next conference.18:59
juliusbglowplug: see this page for PDFs of the slides:
juliusb(it's linked to from the main page now, too)19:20
glowplugFantastic!  Also I finally noticed that you have a pdf build for the mor1kx docs.  8)19:22
juliusb(I think it works!)19:32
juliusbthat documentation isn't great at the moment19:32
juliusbI intend on adding a lot more detail before we do a release19:32
glowplugThat would be awesome.  The good documentation is so important for getting people onboard to contribute.  It can really seem impossible to help when you guys are all so talented.  O_O19:46
glowplugI'm interested in the ASIC production drive that you guys did.  Are there still plans to manufacture the ORPSoC in its current form?19:48

Generated by 2.15.2 by Marius Gedminas - find it at!