#milkymist IRC log for Sunday, 2012-02-05

Fallenoujust to let you know Milkymist One RC3 is published on Solderpad : http://solderpad.com/solderpad/milkymist-one/00:35
FallenouPaul and Andrew, two guys I met in London are the founders of Solderpad, they are open source hardware activists00:36
Fallenouthey contacted me about it in october 2011 but they didn't want me to talk about it, it was still not released and not accounced publicly00:37
Fallenoubut it seems they have gone public now so I can release it here :)00:38
Fallenoufeel free to contact them after RC4 release, I could contact them to ask them to update their data00:38
Fallenoulekernel_: I paste again because you got disconnected :00:39
Fallenou01:40 < Fallenou> just to let you know Milkymist One RC3 is published on Solderpad : http://solderpad.com/solderpad/milkymist-one/00:39
Fallenou01:40 < Fallenou> Paul and Andrew, two guys I met in London are the founders of Solderpad, they are open source hardware activists00:39
wpwrakthe schematics as one massive image. the PNG from hell for sure ;-)00:46
Fallenoutheir viewer is nice00:49
Fallenouthe zooming is easy00:50
Fallenouit's funny the M1 is lost among tons of Arduino*** projects ahah :)00:51
Fallenouon the http://solderpad.com/solderpad/ pae00:51
Fallenoubut soon Milkymist will be able to fight back for the "blinking LED war"00:52
Fallenouit will be the shiniest, for sure00:52
wpwrakyeah. victory in the LED arms race shall be ours ;-)00:58
wpwrakwhat i don't quite understand is what value solderpad adds. are they just rebroadcasting projects that are already properly hosted somewhere else ?00:58
Fallenougood question00:59
FallenouI think they would prefer project maintainers to update their project page themself00:59
FallenouI emailed them about that00:59
Fallenouwarning them about one of OpenCores problem00:59
Fallenouyou don't want to maintain two repositories01:00
Fallenouso usually one ends up not maintained01:00
Fallenouand it's usually opencore's / solderpad01:00
wpwrakwhy should project maintainers maintain solderpad's project page ?01:00
Fallenoubut they rely on git, and they answered me that it's easy to automate publishing to two remotes (using hooks)01:00
wpwrakif they already have their own infrastructure01:00
wpwrakit doesn't make sense01:00
wpwrakyes, you can easily duplicate a git. but why ?01:01
kristianpaulThe number of logical LUT blocks exceeds the capacity for the target device. :-|01:01
Fallenouthey want to centralize stuff I guess01:01
Fallenouto be like the github of Open Source Hardware01:01
Fallenouit makes sense01:01
Fallenouthere is one website for social+source for open source *software*01:02
Fallenouwhy not making one for open source *hardware*01:02
Fallenoukristianpaul: woow :x01:02
Fallenoukristianpaul: your GPS project ?01:02
kristianpaulsome sha256 stuff..01:02
wpwrakFallenou: well, but github offers hosting. that's a useful service. much like qi-hw, sourceforce, launchpad, etc., offer repository, wiki, mailing lists, etc.01:03
Fallenouwpwrak: they will be adding functionnalities, like diffing between schematics01:03
wpwrakFallenou: but mirroring these places seems kinda pointless01:03
Fallenouwell I think they offer/will offer hosting too01:04
Fallenoubut for already existing projects indeed it's mirroring01:04
Fallenoutheir goal, as they told me is : "we just01:05
Fallenouwant to build something that is useful to them and try to improve01:05
Fallenoucollaboration in electronics."01:05
FallenouI think it's early to say, let's see what kind of service/value they will add :)01:05
Fallenoubut knowing these people, i think it could get interesting01:05
kristianpaulFallenou: https://github.com/ngzhang/Icarus01:06
Fallenouthey really are interesting and experienced men :)01:06
Fallenouso they must have a real value that I don't get yet01:06
kristianpaulsome boring waste of time bitcoin miner ;-D01:06
Fallenouoh ok01:06
FallenouI've hread of it01:06
wpwrakwell, there's no harm in keeping in touch :)01:07
FallenouI just wanted to let you know about the page01:07
Fallenounot more actually01:07
Fallenoukristianpaul: 2 Spartan 6 wow, is the PCB built already ? is it working ?01:10
Last message repeated 1 time(s).01:16
Fallenougn8, tomorrow debugging why lm32_dcache does not work in ISIM !01:27
azonenbergHmm, I think i'm going about FPGA development more like ASIC development06:31
azonenbergI have printouts of spartan6 slices at various magnifications (ranging from 1x2 slices to the whole XC6SLX45)06:32
azonenbergthat i mark up with a pencil06:32
azonenbergdoing physical layout :p06:32
azonenbergwhoever sent me the link to Lava, i cant thank you enough06:32
wolfspraulazonenberg: cool! :-)06:41
wolfsprauldo you care to upload some high-res photos about that?06:41
wolfspraulthe physical realities on the fpga die seem to be overlooked too often06:41
azonenbergwolfspraul: this isnt physical, its from planahead06:45
azonenbergalthough i am studying physical06:45
azonenbergas well06:45
azonenbergi have some XC9536 CPLDs that are decapped already and being imaged shortly06:45
azonenbergi hav esome pics already but they're not full die06:45
wolfspraulyes but if you have nice looking graphics or photos, I am always interested in promoting them a little06:45
azonenbergspartan3a will be next06:45
azonenbergs6 is further out as they're pricier :p06:46
azonenbergi might do like an LX4 at some point though06:46
wolfspraulvisual understanding can help a lot06:46
azonenbergactually, let me upload a pic of this design now06:46
wolfspraulnice :-)06:46
azonenbergpencil on paper printout from planahead lol06:46
azonenbergi'm not one of those "write the HDL and let the synthesis tool do the rest" types lol06:48
azonenbergin fact, as soon as i fully understand one level of abstraction i find myself wanting to see what's underneath06:48
azonenbergdid you see my BGA board btw? http://colossus.cs.rpi.edu/pictures/2012/January/1-31-2012%20-%20BGA/S7302457.JPG06:48
azonenbergthis is spartan3a but i have some XC6SLX16 in the same package that are going on the next board06:48
azonenberghome soldered06:48
azonenbergthat's FTG25606:49
azonenbergsorry for bad lighting, its 2 AM here and my room isnt the brightest place06:49
wolfspraulwhat is planahead - is that some software?06:50
azonenbergyou're working with xilinx FPGAs and you've never used it???06:51
azonenbergyou have no idea what you're missing06:51
azonenbergit's part of ISE06:51
wolfspraulme 'working' with xilinx fpgas may overstate what I do every day06:51
wolfspraulah ok, googling didn't find anything but that makes sense :-)06:52
azonenbergit lets you do graphical IO pin planning, view a placed design on a physical (though somewhat stylized) layout of the chip06:52
wolfspraulazonenberg: I'm in the learning department, forgive me ;-)06:52
azonenbergcolor-code by hierarchal blocks06:52
azonenbergview critical paths on the physical layout so you can get a feel for what's trying to talk to what06:52
azonenbergthat combined with color-coding by hierarchy is a very powerful tool for understanding when you're constrained by routing and placement rather than too many gate delays06:53
azonenbergrandom example of how i'd use it for timing analysis06:54
azonenbergthis is a rather extreme design (note the timing constraint in the lower panel)06:54
azonenbergso i was hand-placing every FF and LUT06:54
wolfsprauldoes this relate to your homecmos project in any way?06:54
wolfspraulor separate line of activity...06:54
wolfspraulhow is homecmos doing btw? I admit I am not regularly reading the backlog of the channel...06:55
azonenbergIt's been slow due to school etc06:55
azonenbergcurrent bottleneck is still deep Si etching06:55
wolfspraulit's difficult to extract the bottom line after some months of intricate technical details06:55
azonenbergi determined that i need to be able to etch with an evaporated Ni hardmask06:55
azonenbergmy last experiment failed because i couldnt evaporate it using the default configuration of the tool on campus06:56
wolfspraulso homecmos is still ongoing?06:56
azonenbergbecause nickel sublimes rather than melting and then evaporating06:56
azonenbergSo the current TODO is designing an adapter to hold my wafer *above* the evaporatoin source instead of below it like the tool is normally used06:56
azonenbergthere's room in the bell jar if i do it right06:56
azonenbergand i have some aluminum bar stock that should work fine at those vacuum levels06:57
azonenbergi just have to actually find time to go down to the machine shop06:57
wolfspraullast I remember something about comb drive...06:57
azonenbergand yes, there's a lot of process development going on so i havent had much to show06:57
azonenbergright now its tooling06:57
azonenbergfor example, i built myself a fume hood over break06:57
wolfspraulmaybe you write a blog post one day summing up the bottom line for the dummies :-)06:57
azonenbergnow i can work with more aggressive etchants safely06:57
azonenbergor just larger volumes of the existing ones06:58
azonenbergthe next step is to make a nicer spin coater06:58
azonenbergi'll be blogging on those shortly06:58
azonenbergi also did some work on high aspect ratio etching of copper06:58
azonenbergwill be posting on that as soon as i get SEM cross section images of the sample06:58
azonenbergwhen *that* happens is TBD06:58
wolfspraulnice. when you blog, also talk about what use cases this could potentially lead to. because I'm sure a lot of readers would ask that...06:59
azonenbergof course06:59
azonenberganyway so did you see the pencil image i linked a minute or two ago?06:59
azonenbergthat's part of a device i am hoping to have done by weds06:59
azonenbergit's a clock glitch generator for an in-class demo07:00
azonenbergi want to be able to generate a ~5 MHz clock signal07:00
azonenbergwith one pulse (at a precise position in the train) a lot shorter07:00
azonenbergbut always the same length, and at the same offset07:00
azonenbergso the thinking is to make a tapped ring oscillator much like the delay-locked loop in a DCM07:00
azonenbergexcept the tap position is user controlled rather than being based on detecting phase shifts07:01
azonenbergso i can hold it at tap 15 for a while, then suddenly jump to tap 307:01
azonenbergand the period of the signal suddenly gets longer07:01
azonenbergthe hard part will be glitch-free shifting of frequencies07:02
azonenbergso i can jump straight from Fa to Fb with no unwanted pulses of different lengths07:02
wolfspraulunderstand (and not)07:05
wolfspraulbut sounds good07:05
azonenbergi'm the TA for a class on cryptography at my school07:05
azonenbergwe're talking about hardware vulnerabilities and fault attacks07:06
azonenberghow flaws in the physical implementation can make a strong algorithm like AES or RSA fail07:06
azonenbergso i'm gonna try and put together a nice demo07:06
azonenbergrun a stripped down RSA on a little 8-bit microcontroller or something07:06
azonenbergclocked by my gadget07:06
azonenbergand demonstrate how we can cause single-bit errors by giving it a glitchy clock07:06
wolfspraulazonenberg: colossus.cs.rpi.edu/pictures are all your pictures?07:30
wolfspraulI'm browsing through some of that...07:30
azonenbergThats a raw dump of every photo i've taken since 200407:30
azonenbergexcept for a tiny fraction that i didnt want public for one reason or other07:31
wolfspraulthey are all freely licensed like your homecmos pics?07:31
wolfspraul(just asking in case I want to include one somewhere, with attribution of course...)07:31
azonenbergI've never officially set licenses on them since the server is mostly there for me to show people specific pics without the hassle of uploading it on demand07:32
azonenbergi'd probably be willing to share almost all under cc-by-nc or similar07:32
wolfspraulcan you remove the -nc still?07:32
azonenbergI'd have to decide on an exact license depending on the pic07:32
azonenbergif its stuff of my lab work, then cc-by or cc-by-sa would be fine07:33
wolfspraulok, too much hassle :-)07:33
wolfspraulcc screwed up royally with -nc07:33
azonenbergbut if you're talking, say, my nature photos i might want -nc on those07:33
azonenbergOk, how about this07:34
wolfspraultotally inefficient to set or even enforce07:34
wolfspraulbut I like how the chinese shanzhai now steal -nc licensed ifixit photos07:34
wolfspraulthat's so twisted my head hurts07:34
azonenbergany pics of gadgets i've built or lab stuff are cc-by07:34
azonenberganything else, ask07:34
azonenbergi'm not opposed to sharing, understand07:34
wolfspraulok, I think that's clear for me07:34
wolfspraulI'm more and more leaning to just public domain07:35
azonenbergbut the archive is something like 30,000 pics and its hard to make a blanket statmeent :p07:35
wolfspraulcopyright is broken, and cannot be fixed with more copyright07:35
azonenbergIt really boils down to two things07:35
wolfspraulthat's just my personal opinion though, hope this doesn't set off a flame war here...07:35
azonenbergI want to be credited when my work is used07:35
wolfsprauloh you bet07:35
wolfspraulbut that's a matter of courtesy, social standards and values07:35
wolfspraullike I don't spit in my hand before shaking yours :-)07:35
wolfspraulto be very blunt the Chinese way :-)07:36
azonenbergAnd i would like, but do not mandate, for people to drop me a line and say what it's being used for07:36
wolfspraulI am 100% with you07:36
wolfsprauland cc failed on all of this07:36
azonenbergi understand thats far too complex to put in a license, nearly impossible to enforce, and sometimes too restrictive07:36
wolfspraulinstead they have poeple argue about -nc -nd -sampling or whatever clauses07:36
azonenbergwhich is why i dont mandate it07:36
wolfspraulyes but that's what matters, anyway, I got it and THANK YOU!07:37
azonenbergThe only time i would do -nd is if i had a very specific reason not to have the image modified07:37
azonenberglike, say, an identifiable person or event07:37
wolfspraulif I ever want any of those pics, and that will be the technical ones (gadgets or lab stuff), I know you are happy about that07:37
wolfsprauland the attribution and all come included when I do anything07:37
wolfspraulyes but -nd doesn't work either :-)07:37
azonenbergYeah, again thats just when i'd consider it07:38
wolfspraulit's unpractical, unenforceable, I would argue even unsocial07:38
wolfsprauljust rubbish07:38
azonenbergrealistically, i dont expect to profit off any of my photography07:38
azonenbergand i dont want to limit what other people can do with it, within reason07:38
wolfspraulmy favorite now are Chinese Shanzhai repair manuals (sold for a profit of course) using -nc licensed ifixit photos07:38
wolfspraulI love that07:38
wolfspraulfirst ifixit has a very twisted busines model07:38
wolfsprauland then the strange shanzhai add their 3 twists on top of that07:38
azonenbergjust out of curiosity, which pics in particular were you thinking of using and for what?07:39
wolfspraulthey even leave the ifixit watermark in there, so what07:39
azonenbergI like to have some idea of what my work is being used for :)07:39
azonenbergif nothing else it gives me a hint as to what would be most useful to others if i did more of it07:39
wolfspraulnone right now, I'm just trying to not trample over your work and how you feel it should be used07:39
wolfspraulprobably technical pictures that are kind of eye-opening to newbies07:39
wolfspraulsomething that opens a window to a new thought or imagination07:40
wolfspraulcan't pin it down more than that now07:40
azonenbergwell i probably have a lot of that07:40
azonenbergall of Silicon Exposed is intended to be breaking down barriers as to what's assumed possible for hobbyists07:40
azonenbergincluding but certainly not limited to ic fab07:40
wolfsprauln8 (calling it a day :-))14:08
Fallenoun8 !14:11
mwallelekernel: do you know if there is a difference between an fpga bitstream and a cpld 'bitstream' apart from the fact that the actual plds have another architecture16:25
lekernelwell, because of that different architecture, there are probably large differences in the bitstream format as well?17:20
mwallemh there doesnt seem to be a 'bit' file for cplds.. seems to be called jed (jedec bit map or sth like that)17:28
lekernelhaha, still the old PAL/GAL stuff?17:36
GitHub150[milkymist-ng] sbourdeauducq pushed 2 new commits to master: http://git.io/Ouvlaw18:19
GitHub150[milkymist-ng/master] flash: remove splash screens - Sebastien Bourdeauducq18:19
GitHub150[milkymist-ng/master] Add tools - Sebastien Bourdeauducq18:19
GitHub14[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/629e771fc00911f1d1477adc40f61278cf410a1e18:37
GitHub14[migen/master] fhdl/structure: binary constant builder - Sebastien Bourdeauducq18:37
GitHub27[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/yHqzoA18:59
GitHub27[milkymist-ng/master] Memory map - Sebastien Bourdeauducq18:59
GitHub51[milkymist-ng] sbourdeauducq pushed 2 new commits to master: http://git.io/GakbGw19:06
GitHub51[milkymist-ng/master] Update gitignore - Sebastien Bourdeauducq19:06
GitHub51[milkymist-ng/master] BIOS: hello world - Sebastien Bourdeauducq19:06
lekernelazonenberg: you should combine lava with the bitstream reverse engineering work at recobus.de / debit19:42
Fallenoulekernel: my lm32 D_DAT_O stays at 0x00000000 even with a few lb r1, (r0+something)21:02
lekernelD_DAT_O is used to transmit the data to *write* to memory - lb *reads* memory21:44
Fallenouerr my D_ADR_O sorry21:45
FallenouI had to remove a few bottom lines from lm32_cpu.v about registers initialization in case of simulation21:51
Fallenoubecause it didn't compile21:51
Fallenoumaybe it's the source of my problems21:51
lekernelI can't see why it shouldn't compile22:00
Fallenouok it works now22:05
FallenouI removed this code which does not compile and I put an initial begin ram[0] = {data_width{1'd0}}; end in lm32_dp_ram.v22:05
Fallenouand now data works :)22:05
FallenouD_ADR_O changes from 0 and it seems way better now22:06
Fallenougoing to bed, gn8 !22:07
mwallelekernel: old pal/gal stuff?22:07
mwalleFallenou: gn822:08
kristianpaulwho sell PALs right now? and what integration scale?22:31
wpwrakkristianpaul: the museum of prehistoric arts ;-)22:31
kristianpaulwpwrak: ;)22:32
kristianpaullattice seems to call it GAL22:32
wpwraklekernel: what would you think of giving xiangfu commit access on github, with the stipulation that he should do things that need review (like his current softusb work) in topic branches ?22:34
kristianpaulbah discuntinued..22:35
kristianpaullekernel: what was the point of showing us this wikipedia article?22:35
mwallekristianpaul: i think because i asked about pals22:36
kristianpaulahh ;)22:37
mwallelekernel: what was the other project about fpga bitstream revers engineering besides debia/ulogic.org ?22:38
lekernelmwalle: recobus22:44
lekernelwpwrak: so far he hasn't sent so many soc patches...22:45
lekernelhe already has commit access to FN and RTEMS22:45
wpwraknow he's working on softusb, those hid(eous) reports22:46
wpwrak(fn) ah, didn't know that22:46
kristianpauloh dear, http://paste.debian.net/155141/23:49
kristianpaullekernel: how do i know wich parts of  a HDL code is requiring more fpga resources?23:50
dvdkkristianpaul: what did you attempt to do?23:52
kristianpauldvdk: fit a bitcoin miner core in the M1's lx45 fpga23:52
dvdkkristianpaul: good luck.  for a sha256 hash algo you need > 256 bits of registers. that's a tough number of slices (at least 64?).  most miner designs might use >1 hashing circuits,23:55
dvdkbut the error message is something more specific.23:55
dvdkno, it's 8 register bits per slice.23:56
kristianpaulbtw di you get a M1 right?23:57
dvdkok, but from the error message, it looks quite clearly like you defined waay to many registers.23:57
kristianpaulyes, but i need find the "leak" of registers first :)23:58
dvdkso I'd guess you instantiated some 100-way parallel miner.23:58
kristianpauland all the other warnigns are nothing but friendly to suguest part of code to check23:58
dvdkthe hash algo needs lots of regisers, go look how many of these algos are instantiated23:59
kristianpaulwhat is 100-way?23:59
kristianpauli still getting in to this miner world :)23:59
--- Mon Feb 6 201200:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!