IRC logs for #openrisc Thursday, 2013-10-10

--- Log opened Thu Oct 10 00:00:33 2013
hansfbaier_franck_: stekern: First rather complete version of the tutorial. Due to lack of time, I was not able to test everything. Please let me know how it goes. https://github.com/hansfbaier/openrisc-sockit-tutorial/blob/master/openrisc-sockit-tutorial.pdf06:05
stekernhansfbaier: excellent, do you mind if I put a link to the in the README?06:08
hansfbaierstekern: to the _ ?06:10
hansfbaierPDF?06:10
stekernI should probably try to get the sockit system into openrisc/orpsoc-cores too06:10
stekerns/to the/to that06:10
hansfbaierstekern: I hesitated to include the vmlinux-x.bin blob in the git06:10
hansfbaierstekern: so I used the link from your chocolate factory.06:11
stekernah, that's fine. I don't expect it to be slashdotted ;)06:11
stekernit's hosted on a machine in my basement, so the up bandwidth is limited to 3mbit06:12
hansfbaierstekern: off-topic: I discovered arduino for android yesterday. Been able to compile and download to the nano via OTG from my Galaxy Nexus. Fun!06:13
stekerncool06:14
stekernI've never used arduino though06:14
hansfbaierstekern: I really like the mini-pros ($3.5/board) and the nanos ($7.5 / board).06:14
hansfbaierstekern: Because of the price and the size they are great as USB appliances for PC06:15
hansfbaierstekern: as GPIO/I2C/SPI-extenders for PC for example.06:16
hansfbaierstekern: Since the PC doesn't have those06:16
stekernI started out with AVR long before something like arduino even was considered, so it never really interested me06:16
stekernand I have a pile of custom avr boards with an ftdi chip and some pinheaders06:18
stekernand a custom bootloader that I wrote06:18
stekernso I kind of had my own "arduinos" ;)06:18
hansfbaierstekern: It's great for quick hacks. Due to the massive amount of code and libraries around, I was able to get a USB-serial to 1602 display appliance running in 1-2 hours.06:22
hansfbaierstekern: including patching the hardware06:22
hansfbaierstekern: coming back to the above, you probably mean the orpsoc-cores README, yes, of course.06:25
stekernit probably has it's place (the popularity shows that if nothing else), but I always felt that the AVR's are so simple on their own that abstractions aren't really necessary06:25
hansfbaierstekern: I really like their instruction set. I think it's beautiful.06:26
hansfbaierstekern: I always hated x86 assembly. That's why I never learned it. It's so ugly. I was spoiled learning assembly on the AMIGA / 68000. After that I would not touch x86 asm.06:27
hansfbaier(I actually did my first assembly on the Z80)06:28
hansfbaierI really like the RISC instruction sets, they are simple and beautiful.06:28
stekernI agree completely, I started with x86 assembler and later learned that there are actually more sane ways06:29
hansfbaierI read the OpenRISC instruction spec in 1-2 hours, and a day after that I had my first or1k-asm running06:29
hansfbaier(with some support on IRC)06:30
stekernand yes, avr has a pretty nice isa. I actually started out with PIC back in the day, but pretty quickly moved over to AVR06:31
hansfbaier stekern: Would be nice if someone tried out the tuturial (was it olof that's got is SocKit still in the garage?)06:33
hansfbaiers/git is/got his/06:33
stekernyes, let's push him into start making some use of it!06:33
hansfbaierstekern: He can mail it to me if he does not want to :]06:34
hansfbaierstekern: I'll be glad to have another one06:34
stekernI think _franck_ could be a potential guinea pig too06:34
hansfbaierstekern: Yes, he seemed quite enthusiastic about it.06:34
hansfbaierstekern: One of my medium-to-long term goal would be PCI-Express on the SocKit: I already have 2 HSMC-connectors here. I'd like to design a breakout to PCIe and plug the SocKit into a PC. That would be fun06:35
hansfbaiers/goal/goals/06:36
hansfbaierstekern: Could you recommend a PCB service that can do the hardened gold plating required for connectors?06:36
stekernhmm, I've used pcbcart for my hobby projects, but not sure if they can do that06:37
hansfbaierstekern: Do you know whether 2 layers are enough for PCIe, or would I need 4?06:39
hansfbaierstekern: impedance requirements etc.06:39
hansfbaierstekern: thanks for the tip06:40
stekernI have no idea, that's way above my PCB design capabilities ;)06:40
hansfbaierolofk: https://github.com/hansfbaier/openrisc-sockit-tutorial/06:42
olofkGood $localtime everyone06:42
hansfbaierolofk: Ready to get the SocKit out of your garage?06:42
olofkno :(06:42
olofkBut it's great with a tutorial. I will take a look at it even if my hardware is out of reach06:42
hansfbaierolofk: I could not test anything in my tutorial, so I really would appreciate some testing...06:43
hansfbaiers/anything/everything/06:43
olofkI can understand that. I'll tell you how it goes when I have time to power up my board06:44
hansfbaierI tested it up to building of uboot+spl06:44
hansfbaierolofk: thanks06:44
olofkWe SoCkittens try to help each other out :)06:44
_franck_hansfbaier: you need a 4 layer board for PCIe07:28
_franck_plus if you choose a cheap PCB compagny, make sure they give you the board stackup07:29
_franck_hansfbaier: and make sure they have always the same stackup, because they often do not garantee it07:30
olofkhansfbaier: I got stuck at Section IV, step 9 in you tutorial07:39
hansfbaierolofk: symptoms?07:45
olofkNo orange juice. Couldn't do pushups07:46
hansfbaierolofk: haha07:46
hansfbaierI forgot to mention that was optional07:47
olofkIt's a great tutorial. We should get a link from somewher on the OpenCores wiki when you feel somewhat confident about it07:47
hansfbaierolofk: Well, if it works for you and franck I would be confident enough07:48
hansfbaierolofk: The tutorial also helped to serve me remember the steps I had taken myself. While writing I just realized I forgot quite a bit. Would have been hard to reproduce, If I needed that later07:49
hansfbaier_franck_: Thanks, good to know. On page 4 of http://agata.pd.infn.it/LLP_Carrier/New_ATCA_Carrier_web/Appnotes_And_Reference_Designs/PCB_SI/pcie_pcb.pdf it looked like 2 layers are enough for Microstrip. What source do you recommend to learn more about PCIe PCB design?07:57
_franck_2 layer are enough but wont match the impedance on a 2 layer PCB, the height is too big between layers08:01
hansfbaier_franck_: I see, thanks.08:02
_franck_you have to have 2 layers "close to top" and 2 layers "close to bottom"08:02
hansfbaier_franck_: ?08:02
_franck_I mean on a 4 layer board you have: copper -- prepeg -- copper ------------CORE --------------copper -- prepeg -- copper08:05
_franck_http://www.google.fr/imgres?sa=X&biw=2133&bih=1146&tbm=isch&tbnid=XD-oykfGsggAOM:&imgrefurl=http://electronicdesign.com/embedded/resolve-stress-caused-high-speed-interfaces&docid=SGhj8QwlhS-jkM&imgurl=http://electronicdesign.com/site-files/electronicdesign.com/files/uploads/2013/06/0808DSpericom_Fig1.gif&w=595&h=335&ei=4V9WUqG6HJKQhQe1jIGIAw&zoom=1&iact=rc&dur=1&page=1&tbnh=140&tbnw=261&start=0&ndsp=20&ved=1t:429,r:0,s:0,i:84&tx=131&ty08:06
_franck_hell of a link !08:06
_franck_http://electronicdesign.com/embedded/resolve-stress-caused-high-speed-interfaces08:07
hansfbaier_franck_: Wow that's a great article thanks!08:18
olofkI really want to find the time to get started with doing PCB. Seems like fun, and I have some ideas I want to try out08:18
_franck_hansfbaier: https://docs.google.com/file/d/1ojWM85LQ4Ho6FHA2PUDassS7i_ieM__LgB14zqg89CbBYD1ZoPst41b3v7ZK/edit?usp=sharing08:33
hansfbaier_franck_: great, thanks!08:35
hansfbaier_franck_: That looks pretty comprehensive08:37
_franck_the best doc I've ever found08:39
hansfbaier_franck_: Quite ambitious for one who never sent something to the fab, but I think a PCIe-HSMC breakout should be doable given enough time and care08:40
_franck_pcie is not that hard to get working. I even made it works with hand solder wires to switch TX-RX lines !08:42
_franck_(short wires :))08:43
hansfbaier_franck_: good to hear. Nice app would be to implement digital audio effects on the FPGA and use that as a DSP plugin for ardour/jack/LV208:43
olofkhansfbaier: That sounds a lot like what stekern was doing before he got sucked into the OpenRISC vortex08:47
hansfbaier:)08:48
olofkIIRC, he wanted to add a CPU to his FPGA synth design.. he just needed to fix some things with OpenRISC first. That was a few years ago08:48
_franck_did you see that post here: http://opencores.org/forum,OpenRISC,0,3509 ?08:55
olofk_franck_: Yes, I did. I found the old posts very interesting and was planning to respond to it08:56
olofkA lot of the things there have been taken care of08:56
_franck_ok08:56
olofkAnd other parts are on my todo list :)08:56
stekernheh, yes, looks like we actually live in the predicted future on many of the points made back in 200809:05
stekern#5 I discussed with juliusb during the conference, because that's something we should have09:08
stekernif nothing else, olofk beating my SDRAM controller about is a good example why09:09
stekernthe problem he identified is that you have to have very well defined interfaces for that to make sense and even if you do, it will be harder to change them if you have testbenches depending on them09:10
stekernbut I don't think that's an excuse good enough09:11
stekernso, I've put on my list of todo's to at least write some unit testbench for the MMUs09:13
stekernthose have a fairly well defined interface that will not change a lot09:13
stekern(openrisc vortex) yup... but I've actually started to bring the synth back in to the picture, already translated some small module into verilog09:20
stekernI've also got an idea to my head to do a OPL2 midi synth clone, to get the right scummvm experience from mid-90's PC09:22
knznice :)09:33
olofkstekern: That would be so cool. :)09:39
stekernI know, there's such a buzz around the old SID chips, no-one seems to really appreciate the crappy sound those old chips emit09:43
_franck_I won't be able to generate the bitstream for the sockit on my virtual machine. What was the memory requirement ? 4GB ?09:45
stekern> 4GB09:46
stekerndoing a SID clone would be kind of fun too though, not very original though. and a lot of it's appreciation comes from the analog filters, which are hard to emulate (properly) with digital circuits09:48
stekernI like the "analog" data format in gtkwave11:49
amsstekern: speak for yourself.11:50
olofkstekern: Yeah. I think it sucks11:50
stekern?11:51
olofkThat was a bit harsh, but I find it annyoing that it rescales all the time. I've been trying to use it to track the PC, but it has been frustrating11:51
amsstekern: not appreciating SID tunes11:52
stekernams: I think you misunderstood, I meant that no-one seems to appreciate the nostalgia of the crappy sound the OPL2 emits11:54
stekernolofk: ah, yeah, it's perhaps not perfect. But it's pretty useful for synth development ;)11:55
olofkstekern: Nostalgia is extremely selective11:55
olofkstekern: But then you are using it for it's intended purpose which might be a better idea than my PC hunting :)11:56
amsstekern: i appreciate it, i love sid11:57
stekernams: yeah, me too... but SID != OPL211:57
amsstekern: shurg, OPL2 is also nice :-)11:58
stekernams I even did this: http://www.youtube.com/watch?v=qYdOhU_aUH011:58
amsnice :-)11:58
amsnot 8-bit, but i still listen to a lot of xm11:59
stekernme too11:59
amswas happy to find a clone of ft2 that one can use in gnu ...11:59
ams(never was a scream tracker person)12:00
amsstekern: beat this, hand cranked 33 player is what my girlfriend and i listen to when chilling to music ... 1920's jazz, foxtrot, stuff like that :-)12:01
olofkams: Ha! 33 1/3? Is that what you youngsters use to play your stressing noises? Everyone knows that the good music is on the 78's12:03
olofk(I actually do have a couple of 78's, but nothing that plays them)12:06
amsolofk: erm, sorry, i meant 78! you didn't have 33's in the 20ths12:06
amsi think they came about in the 30ths ...12:07
amshttp://commons.wikimedia.org/wiki/File:Portable_78_rpm_record_player.jpg12:08
olofkI also do some hand cranking when I want to relax12:08
amssomething akin to that ... gf bought it at some scrapyard12:08
amsolofk: haha12:09
olofkThat's a nice machine.12:09
amsonly thing that sucks is that each side is 2 minutes and you need to crank often :-)12:09
olofkI have some fetish for portable record players, but all mine are more modern 33/45 and battery powered12:10
amsnod12:10
amsget a 78 .. seriously, they are quite fun, and cheap12:10
amsor gimme your vinyls!12:11
olofkI have two of these http://farm8.staticflickr.com/7166/6431129629_10016ff1c9_z.jpg . They should play 78, but I'm not sure that the needle will handle them12:11
amsphonographs are i find quite relaxing...12:11
olofkpornographs, yeah. I agree12:12
amsyou need a gf ... both for handcranking, and other things.12:12
olofkOh, I have plenty of gif already12:12
amshaha12:13
amsas for needles, your needles would get crappy really quickly ..12:13
amsaccording to normal usage, one needle is used for 20 playings or there abouts ..12:13
amsnormally small gold needles as well that you would give to your grocery shop to sharpen12:14
amsand the difference in needles is huge ..12:14
olofkOh.. and I guess using a vinyl needle would wear it down even quicker12:14
olofkA needle for vinyl records I mean12:14
amsdefinitly12:14
olofkNeed to get a dedicated player12:15
amsi need to get a 3312:15
olofk:)12:15
olofkMy biggest problem is that I don't have a CD player any more. Still like them, but can't play them anymore12:16
amsi wonder actually if one could get a needle sharpener or something12:16
amshttp://www.windupgram.co.uk/hints.htm12:17
amscool ... i'm getting such needles12:17
olofkLast christmas I hooked up a USB DVD writer to my raspberry pi, wrote a udev rule to launch mplayer when a disc was inserted. Set mplayer to listen to a pipe and connected a TAC-2 via GPIO with a small daemon that read the gpio and wrote to a pipe12:18
olofkWhat actually prevented it all from working was that apparently mplayer refuses to read from the pipe when it's not associated with a terminal, which is the case when it's started by the udev rule. Stupid OS12:19
amslame12:20
amsstupid mplayer i say12:20
olofkyeah. Maybe that's the main problem12:20
juliusbYou're a busy man during the Christmas period, olofk. OR1200 patches and raspberry pi hacking12:37
olofkjuliusb: Yeah. That was embarrasing. Stupid bugzilla timestamps :)12:44
ssvbdoes anybody know if any possible Intel MMX patents have already expired by now?14:03
juliusbit seems we weren't alone in holding a conference last weekend14:52
juliusbhttp://www.ohwr.org/projects/ohr-meta/wiki/OHW201314:52
juliusbJavier Serrano has a good intro14:52
juliusbIm thinking a state-of-the-project presentation might have been good, too14:53
juliusbHis slide on ohwr.org mentions it doesn't require registration ;)14:54
frubblThis could be interesting: http://www.ohwr.org/projects/ohr-meta/wiki/OHW2013#1430-1500-Migen-Sébastien-Bourdeauducq-Slides22:04
frubblHas a wishbone interconnect generator.22:05
juliusbthere's some really interesting presentations on there22:06
juliusbthe one given by Matthieu Cattin looks very interesting22:06
juliusbtalks about the different components they have22:06
juliusband mentions the SDB - self-describing bus, that's certaliny something which should go into ORPSoCv322:07
juliusbit looks like SystemVerilog support is picking up, too. Maybe we can start using typedefs for busses and the like22:11
juliusboh cool!: http://iverilog.wikia.com/wiki/SIMBUS22:13
juliusbnot sure we'd need it, but it's a cool idea22:14
--- Log closed Fri Oct 11 00:00:34 2013

Generated by irclog2html.py 2.15.2 by Marius Gedminas - find it at mg.pov.lt!