#milkymist IRC log for Thursday, 2012-03-15

wpwrakcladamw: what operating temperature do we assume ?01:23
wpwrak(ambiental)01:23
wpwrak(well, ambiental ... inside the box :)01:23
wpwrakcladamwa: what operating temperature do we assume ?01:24
cladamwawpwrak, hi suppose it still works for 70 degree, but we've never done more analysis on M1.01:25
wolfspraulcladamwa: hi good morning :-)01:26
wolfspraulI see you online since 6 AM or so - you start that early?01:26
cladamwano baker test, no design analysis first01:26
wolfspraulI was a little worried that R4 proves so stressful that you cannot sleep :-)01:26
cladamwawolfspraul, ah ? no, maybe only am 8:4001:27
wolfspraulgood01:27
wolfspraul:-)01:27
wpwrakwolfspraul: or so invigorating that he has no need for that much sleep ;-)01:27
cladamwawolfspraul, ha..no, i'm happy that now joerg & werner jumped into design stage/verification which i was weak. ;-)01:28
wolfspraulwpwrak: can you do me a favor and check what type of riitek/remote keyboard you have?01:29
wolfspraulturn it around - do you see any label or marking?01:29
cladamwawolfspraul, wpwrak, i git cloned wernermisc, maybe include me into member ?01:29
wolfspraulcladamwa: done01:30
cladamwaso i can update _r4.csv just in case when i update it.01:30
wolfspraulyou need to clone again with write permission01:31
cladamwawolfspraul, thanks.01:31
cladamwayeah...long time no use, i'll do it again later.01:31
wpwrakwolfspraul:model RT-MWK0101:34
wolfspraulah perfect, the exact same we will inlude01:34
wolfspraulinclude01:34
wpwrakwolfspraul: (non-BT 2.4 GHz, laser pointer)01:34
wpwrakhah :)01:35
wolfspraulwe compared easily 10-15 different ones, also different form factors etc.01:35
wpwraki think i would have tried to lose the laser pointer :)01:35
wolfsprauland in the end come back to the first one we started with, urgh :-)01:35
wpwrakwow, quite a lot01:35
wolfspraulbut that's ok, if you knew before what you know later something would be wrong... :-)01:36
wolfspraulriitek is nice, other vendors are quro, zoweetek, unisen, anterton01:36
wolfspraulbut they all have quality issues imho01:36
wpwrakmy riiteks are still going well. they must be something like 2 years old now01:38
wpwrakvery light use, though01:38
wpwrakcladamwa: regarding the TPS76301T (U12, etc.) which dissipation rating would we apply ? Low K or High K ? or something else ?01:52
cladamwaHigh K suppose.01:56
cladamwachip vendor produced reference design(multilayers and 2 layers) about measurement to develop dissipation rating table i believe.02:00
wpwrakhah. the power supply is more fun than the previous sheets :)02:01
wpwrakyes, page 3 describes the high/low K boards02:01
rohwpwrak: hey... i stumbled onto a ar6k fw lately02:06
rohone which does ap mode02:06
rohany interrest or know somebody who does?02:07
wolfspra1lpainful memories :-)02:07
rohi know.02:08
wolfspra1land this stuff is deader than dead too :-)02:08
wolfspra1ldon't you think?02:08
rohwell.. which chipsets for wifi you know of for < usb/PCIe ?02:09
wolfspra1loh sure02:09
rohmeans sdio or so...02:09
rohnot that much to choose from02:09
wolfspra1lbut the fate of this path of development was decided what. 2-3 years ago?02:09
wolfspra1lsomething like that02:09
wolfspra1lnow they go full power to 802.11ac and more02:10
rohdunno. i just know that i find it in todays devices here and there02:10
rohphones and such. android devices. tablets02:10
rohsuch stuff02:10
wolfspra1lI certainly don't know anyone still actively on this, but I also moved my attention elsewhere years ago.02:12
wpwrakwhat is an "ar6k" ? doesn't sound edible02:15
wolfspra1lfrom the perspective of a wifi chip maker, I think it's safe to say that the experience of the last 5-10 years tells them that the additional momentum from open source drivers does not match the capital needs to compete in the ongoing speed, cost and low-power race02:19
wolfspra1lin other words - open source drivers are a distraction02:20
wolfspra1l(for them)02:20
wolfspra1lI could not go with a straight face to Broadcom, Marvell, MediaTek, Qualcomm, Intel or the other Asian corps and tell them that they should spend brain cells on open source drivers :-)02:22
wolfspra1l(for wifi)02:22
wpwrakintel may be a good candidate. with things moving into the core, closed wifi taints the core02:36
wolfspra1ldon't understand02:36
wolfspra1lwhat do you want to propose to Intel?02:36
wolfspra1lI think their integrated wifi is one of the best supported in Linux now, indeed02:37
wolfspra1lwe have some contacts 'into' Intel, but it's a large corporation and my expectations for getting anything tangible out of 'Intel' are zero :-)02:37
wolfspra1lI have no idea now Intel sees the wifi thing, and I am sure even at Intel there are at least 10 different strategies about it, each one thought through in sophisticated ways, of course.02:38
wolfspra1lthey had their best year ever last year :-)02:39
wolfspra1lI just want to be hired by Intel and become a futurologist.02:43
wolfspra1lat least my monthly paycheck is safe until (early) retirement02:43
wpwrakdunno the wifi situation. just commented that intel should have an interest in being at good terms with open source also on the wifi front. if they already are, even better.02:44
wolfspra1lI think so, yes02:44
wolfspra1lbut of course Intel wifi tech is integrated with pretty much the largest chips you can imagine02:44
wolfspra1lgood luck trying to get that tech out there into an embedded product02:45
wpwrakand there i was, thinking you were pining for a job at sisvel :)02:45
wolfspra1lfuturologist at Intel02:45
wolfspra1lthat'd be it02:45
wpwrak(chips) yeah. if course, they'd be really happy if you manage to pull that off ;-)02:45
wolfspra1lmore money than brain, just think and think and think more about the future.02:45
wolfspra1lthe only risk is depression02:46
wolfspra1lif you have a notebook with integrated intel wifi, be happy02:46
wolfspra1lthat's my bottom line02:46
wolfspra1lI'm not going to over-strategize this stuff...02:47
ThihiI replace my notebooks wifi with intel models usually. Gotta love mini-pci-e.02:53
ThihiUsually == when it's doable.02:56
ThihiLike in my netbook, and in my other laptop, and two of my previous ones.02:56
Thihi:)02:56
wolfspra1lmaybe the enormous investments that are needed in that competition lead to chips that are standardized at a much lower level that is a better starting point for open hardware02:59
wolfspra1lI will keep my eyes open for that03:00
wolfspra1l(and needless to say we will need the Milkymist SoC for that stuff :-))03:01
frankD2hey guys, where do i start if I want to port effecTV as a feature ?06:32
wolfspra1lnever heard of that but looks nice (reading homepage now)06:34
wolfspra1llast news update was over 5 years ago, that makes me feel a little better about the Qi news :-)06:35
wolfspra1lwithout looking at the codebase now, my guess is if you are trying to bring this to Milkymist One, you best start from scratch first, then over time as you understand Milkymist/Flickernoise better and better, maybe there are bits and pieces you can reuse06:35
wolfspra1lcertainly knowledge06:35
wolfspra1lif your goal is to 'port' the entire codebase, that will be a long and tedious path since it sounds like you first need to help making the switch from rtems to Linux, then probably help with the currently broken C++ compiler, and then most likely many other things...06:36
wolfspra1lI suggest taking M1 as-is and then trying to apply things learnt from EffecTV06:36
frankD2got that idea while watching RHCP- The zephyr song.06:38
wolfspra1ldo you have a Milkymist One?06:57
frankD2oh yeah, it's two days old. lol. Thanks dude, it's been great so far. Sorry for not keeping you informed that i've received the package.07:10
wolfspra1loh, nice :-)07:11
qi-botThe firmware build was successful, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120315-0726/08:06
qi-botThe firmware (using branch) build was successful, checkout the VERSIONS for detail, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120315-0906/09:46
cladamwwpwrak, i got Error 1 after I updated milkymist_one_bom_r4.csv under your tree/master/m1r4/dsv  , i think that i was wrong that i shouldn't have updated this file in this way, right ?10:43
wpwrakcl... grr11:21
qi-botThe firmware build was successful, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120315-1046/11:25
wpwrakah, here he is :)11:32
wpwrak1) you should test before committing :)11:33
cladamw1) yup..stupid :(11:33
wpwrak2) the problem are line breaks embedded in some fields11:33
wpwrakmaybe we can fix that at the source ... one is11:33
wpwrak"ACâ~@~Y97 Revision 2.1 Audio CODEC with SPDIF Output11:34
wpwrakthen there is a newline11:34
wpwrak","48-lead TQFP11:34
wpwrakanother newline11:34
wpwrak",,5.030,5.030,,,"NOT START",11:34
cladamwDied at ./bomshelf line 10, <> line 2.11:35
wpwrakyes, that's because the .csv is incorrectly formatted - due to the newlines inside fields11:35
wpwrakif you delete the last characters in fields G2 and H2, that should get rid of the newlines11:38
wpwrakmaybe also change the ' in G2 (AC'97) to ASCII and not the non-ASCII lookalike11:39
cladamwokay... i do it11:39
wpwraklet's see if there are more ...11:40
wpwrakD29 and G29. D29 at the end, G29 between ...49SMD and , 20PF...11:43
wpwrakthen bomshelf should be happy11:43
wpwrakit will warn about three cases of "-", but that's okay11:44
cladamwphew~ lots of newlines, now it searching pdf... :-)12:05
cladamwit means i have to fill *.ods up very carefully. :(12:08
wpwrakjust don't insert newlines ;-))12:18
wpwrakpity that *office doesn't keep them out on its own12:18
wpwrakwell, i suppose i could also teach bomshelf to handle them12:19
cladamwwpwrak, you set a .gitignore for BOOKSHELF on your own directory, right ?12:23
wpwrakno. i just don't  git add  BOOKSHELF :)12:30
wpwrakand having a bit of noise at the bottom of the commit "form" doesn't bother me12:31
cladamwokay :-)12:31
wpwrakwhee, works ! :)12:34
cladamwwpwrak, :) thanks for leading me to remove the err of embedded newlines.12:37
wpwrakthanks for fixing them ! ;-)12:38
ArtyomHello to everyone. I want to give a link to the fpga-board that I find interesting. May be it can be interesting to someone else too: http://www.enclustra.com/en/products/17:55
Fallenou[6~[6~[6~/win 5717:58
lekernelArtyom: what's particularly interesting about it?17:59
lekerneland which board are you talking about :)18:00
ArtyomFor me there are several points (Among these boards I mainly looked at "Mars PM3" + "Mars MX2") 1. I don't know any other board with Gigabit Ethernet + USB-FX3 interfaces.18:03
Artyom2. I liked their idea of spliting FPGA-system on 2 parts: mother-board + FPGA-board.18:04
Artyom3. Standard FMC connector support.18:04
ArtyomI understand that most of these points stand in the side of milkymist. But these boards attracted my attention and I hope they can be interesting to someone else.18:06
lekernelwhat are you planning to do with them?18:07
ArtyomI need something to transfer data from high-speed ADC to PC for post-processing. I think I described my task earlier. These boards seems to be the best alternative for my task. (Until I wouldn't try to develop something similar by myself ;) )18:10
ArtyomI also keep in mind building a gps/glonass/any_other_gnss-receiver on a fpga-board. This board can be used for that too as my digilent-board (based on s3e500) is too small. But anyway I will try to use MM SoC on it (of course if I will get them...)18:15
larsc<shameless plug> Artyom: Have you considered ADIs high speed data converter FPGA solutions? </shameless plug>18:18
Artyomlarsc: Do you mean some ADIs ADC or some demo-board?18:20
larscArtyom: http://wiki.analog.com/resources/alliances/xilinx and http://wiki.analog.com/resources/alliances/altera18:23
kristianpaulhi Artyom18:26
Artyomkristianpaul: hi :)18:27
kristianpaulah USB-FX3 i tought you liked that a lot :)18:27
kristianpaulmost of these points,hmm milkymsit dont have a fast tranfer bus for comunicating with host computer18:28
kristianpaulat least can beat a usb-fx* thing18:28
kristianpaulof coruse in the sense of open hardware ;-)18:29
kristianpauls/course/course18:29
kristianpaulArtyom: how afortable are those enclustra boards?18:30
kristianpaullets see if there is one based on spartan-6..18:30
Artyomlarsc: I didn't search for FMC solutions yet. Because I still don't have exact plan of solving my task. I slowly worked on developing my custom solution. I thought to split design on two parts: ADC-board and FPGA-board connected with each other through SerDes. So I was mainly focused on ADC-boards design and FPGA-boards design. I roughly chose basic components including ADC, FPGA, memories...18:31
Artyom...and so on...18:31
kristianpaulha cool the fpga is plugable ;-)18:31
kristianpauloh wow artix 7 !18:31
kristianpaulnow where are the specs and the prices...18:32
Artyomkristianpaul: I know very little about these boards. I found them couple of hours ago...18:32
kristianpaulok ok :)18:32
kristianpaulbtw had you looked at gpl-gps namuru port ?18:33
kristianpaullooks very promising for porting18:33
kristianpaulit uses ecos threads.. still need see how adapt to rtems18:33
kristianpauloh, this is milkymist ;-)18:34
kristianpaulnice read you here :)18:34
Artyomyes, I looked a little at gpl-gps namuru port. But I didn't have time to study it deeply. Anyway it's a working port. So it's the best starting point.18:36
kristianpaulindeed18:37
kristianpaulah18:39
kristianpaulArtyom: had you looked at Icarus board?18:39
kristianpaulhave lots of pins18:39
kristianpaulperhaps you can wire a USB-FX3 board to it18:39
kristianpaulalso is dual core LX150 spartan6 !18:40
kristianpaulArtyom: http://en.qi-hardware.com/wiki/Icarus18:40
ArtyomNo, I didn't. But building custom USB-FX3 board is too expensive (if we compare it to the products that I found). FX3 requires expensive pcb-design.18:43
kristianpaulok, what about buy it separatellit?18:44
kristianpaulseparatelly*18:45
Artyomspartan6 lx150. Real beast ;)18:45
Artyomtwo FPGA cost as 2 MM one ;)18:46
kristianpaulbtw from your university can you get this http://www.ion.org/search/view_abstract.cfm?jp=p&idno=8133 ?18:46
kristianpaulI was introduced to Michele Bavaro, and he mention some guy working on another gp2021 correlator implementaation18:47
kristianpaulbased on that paper about18:47
kristianpaulYacine Adane, University of Westminster,18:48
kristianpaulbut they have the old idea of implementing  tracking loops and PVT on cpu,18:49
kristianpaulhe18:49
ArtyomI'll try to find this paper.18:50
kristianpauli wonder if you are aware of that, just telling, may be may be not they found similar problems you had with the arm fpga combo..18:50
kristianpaulwho knows..18:50
kristianpaulpaper, good, perhaps you look it interesting18:50
ArtyomI didn't know about Yacine Adane ;)18:52
kristianpaulwe should :)18:52
kristianpaulbtw this board comparisong is part of a short term sourcing plan?18:53
ArtyomThey go traditional way. I know working receivers based on spartan3-correlator + TMS DSP18:53
kristianpauli mean buy18:53
ArtyomI don't know. I will offer this solution to my employer ;)18:54
kristianpaulI see :)18:54
kristianpaulonce i considered buying this http://shop.ztex.de/product_info.php?cPath=22&products_id=73&osCsid=a906b324dc0113bd7b1fa6825fdea31418:54
ArtyomSorry, kristianpaul I need to leave now...18:55
kristianpauloh sure18:55
kristianpaulleave !18:55
kristianpaul:-)18:55
Artyomlarsc: that for the links.18:55
kristianpaulread you later18:55
ArtyomPS I know these boards from ztex. I found them rather interesting too.18:56
kristianpaulbtw mwalle you there?18:58
kristianpauldid you finally be able to ran uart core faster than current design?18:59
kristianpauland take some advantage of the high  speed usb ftdi chip from the jtag-serial pod :-)18:59
GitHub62[migen] sbourdeauducq pushed 1 new commit to master: https://github.com/milkymist/migen/commit/e969b9afc3c31d3a5287c2a126af3e6cbc9617fd19:32
GitHub62[migen/master] corelogic: convert timeline to function and move to misc - Sebastien Bourdeauducq19:32
GitHub160[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/ZsAy2w19:32
GitHub160[milkymist-ng/master] norflash: use new timeline API - Sebastien Bourdeauducq19:32
GitHub137[milkymist-ng] sbourdeauducq pushed 1 new commit to master: http://git.io/V-gEGg19:36
GitHub137[milkymist-ng/master] asmicon: refresher (untested) - Sebastien Bourdeauducq19:36
Action: wpwrak wonders why data sheets specify thermal performance as "degrees C/W" and not simply as K/W23:25
--- Fri Mar 16 201200:00

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