#milkymist IRC log for Thursday, 2011-12-01

wolfspraulwpwrak: wow02:13
wolfspraulso you got usb-midi to work?02:14
wpwrakyup :)02:19
wolfspraullet's get a bunch of controllers!02:20
wolfspraulwhich ones?02:20
wpwraknow can look for places to add more host ports ;-)02:20
wpwrakgood question :)02:20
wpwraki found one that looks quite versatile while being inexpensive. the iCON i-Creativ, MSRP USD 125. http://www.icon-global.com/ShowPro.aspx?ID=13602:22
wpwrakbut it has a very weird limitation on the pad resolution that noticeably restricts what you can do with it02:22
wpwrakit's borderline with my Tornado MIDI RMX, particularly if you don't know any other controller, but you'd immediately notice with the pac man02:23
wpwrakthen i have the korg nanoKONTROL2, but that one has such a cheap feel to it that i'd be reluctant to recommend it. there's an even cheaper variant, the nanoKONTROL (without "2"), but i don't know that one02:29
wpwrakiCON have a direct competitor to the nanoKONTROL, the iControls: http://www.icon-global.com/ShowPro.aspx?ID=1902:29
wolfspraulcatching up :-02:31
wpwrakif the build quality is anything like the i-Creativ, that would be a better entry-level choice than the korg ones. USD 73 MSRP. even has a joystick, but i think that one pretends to be a mouse, so may not be immediately useful for patch control (of course, we can implement whatever we want :)02:31
wpwrakakai also have some devices in the low-cost segment, although a bit more expensive than Korg or iCON. e.g., this one: http://www.akaipro.com/lpd802:34
wpwrakdunno about build quality - or whether a controller with many pads would actually be very useful with the m102:35
wpwrakthe we have the lv3. this is a more complex device. for faders, it competes with the nanokontrol and icontrols. quality is much better than in the nanokontrol. dunno about icontrols.02:40
wpwrakmy tornado remix is better suited for an x/y pad than for a joystick. i'm still trying to see how to make it work pleasantly with faders. (it takes a while to unlearn some style of controls and to learn another style. for now, i find it awkward to control that patch with just sliders, but that may change.)02:43
wpwraki still have to use the LV3 with camera input. that should be the ideal scenario for the joysticks.02:45
wolfspraulwpwrak: ok I added your feedback here http://en.qi-hardware.com/wiki/Milkymist_One_accessories#Products02:53
wolfsprauljust copy/paste for now, need to go over the links etc. and cleanup02:53
wolfspraulthen we can zoom in somewhere02:53
wolfspraulI think a lot of people want to have some small portable well integrated controller to show/demo m102:53
wpwrakyes, and we should make it a clear message that M1 needs that :)02:56
wolfsprauloh definitely02:57
wpwraki mailed iCON to ask whether they plan to improve the resolution on their i-Creativ. if you don't mind packing an extra pound, that one looks like a very suitable moderately low-cost choice that covers a lot of uses.03:00
wolfspraulseems they try to have as many products and names start with 'i' as possible03:01
wolfspraulmaybe one day can sell one to the overlord...03:01
wpwrakyeah. they shouldn't do that. makes them look cheap :)03:02
wolfsprauldo you really think we should go to 4 USB connectors?03:02
wpwrakthe i-Creativ is remarkably heavy and solid. you could use it as a blunt weapon :)03:03
wpwrakif we can ...03:03
wolfspraulwhich ones do you have right now?03:03
wpwrakwith usb, the more the merrier :)03:03
wolfspraulseems you always pull out a few more names :-)03:03
wpwrak(usb connectors) may be a bit tricky to find room for them, though :)03:04
wpwraki have faderfox lv3, icon i-creativ, korg kaossilator pro, and korg nanokontrol203:04
wpwrakmaybe i should get an i-controls, too. that one's cheap even here :)03:06
wpwrakakai also have some small and inexpensive midi controllers, but then i think we're already at the end of that product class03:08
wpwrakthen there are controllers that are more aimed at studios. things like the zero or this http://www.cme-pro.com/products-list/product-Bitstream3X.html (to toss you a new name :)03:09
wpwrak(zero) this one; http://www.novationmusic.com/products/midi_controllers/zero_sl_mk_ii/03:10
wpwrakthese are generally in the USD 500+ segment. and you probably need to make some room :)03:11
wpwrakyet another segment are the ones with a piano keyboard. midi keyboards often have a number of sliders, knobs, etc., too. some even have a lot. e.g., the zero it just the upper half of such a keyboard:03:15
wpwrakthinking of it, the 25 SL Mk II may not be such a bad device. lots of controls, x/y pad (it seems), and also a joystick03:18
wpwrakprobably costs and arm and both legs, though ... let's see ...03:19
wpwrakusd 719 in argentina ... quite reasonable actually03:20
aw_(pre-rc4 reset sch patch): http://en.qi-hardware.com/wiki/File:M1pre-rc4-r-A007-1.JPG03:39
aw_(pre-rc4 sub switch sch patch): http://en.qi-hardware.com/wiki/File:M1pre-rc4-u-A022.JPG03:40
wpwraknice and clean !03:41
wpwrak(the reset)03:42
wpwrakthe usb power switch is messy ;-)03:42
wpwrakbut amazing that you could rework an existing rc3 with it. that's quite a lot of changes.03:43
aw_(7181C video decoder) replaced a new 7181C, the oscillate freg voltage(AC) on C225 is small @ 100mV, C226 is @ 64mV. A good voltage of 7181B: C225 @ 284mV, C226 @ 240mV.08:35
aw_needs to check.08:35
GitHub180[milkymist] none pushed 2 new commits to master: http://git.io/9hltUA09:52
GitHub180[milkymist/master] softusb: don't forget notifying the host when there are news from MIDI - Werner Almesberger09:52
GitHub180[milkymist/master] softusb: accept everything but non-MIDI and system messages - Werner Almesberger09:52
lekernel_elldekaa: soo.... under debian (haven't tried ubuntu yet) it crashes with the same exception when I try to open files. at least, java is successful with porting errors.10:34
lekernel_for some reason the ubuntu installer fails to find my hard disk... grmbl10:37
qi-botThe Firmware build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-12012011-1048/11:38
aw_lekernel, will m1 still work well with i2c without changing s/w after replaced 7181C?13:18
lekernelI think so13:19
lekernelat least I didn't spot obvious differences in the datasheet13:19
lekerneldoes this question mean you have problems?13:19
aw_i saw those signals(outputs) of VS, FIELD and HS are shown well on 7181C which same as other 7181B .13:20
aw_i compared 7181B and 7181C recommended circuit, only pin34 (FB), pin52(ALSB) not the same, also surely many pins on 7181B are NC( those NCs are now available on 7181C, but i don't think they are problems).13:23
aw_so my question is will s/w need to detect an unique id in 7181C? seems it's failed while using current test program.13:25
wolfspraullekernel: do you think Adam should switch to the latest release when flashing units to be shipped out?13:25
wolfspraulwe have a lot of goodies in there :-)13:26
lekernelphew, no13:26
lekernelpeople who can't run the web update probably won't do anything interesting with the M1 anyway13:26
aw_he~ the people like me.13:29
wolfspraulaw_: so you say you put an adv7181c on a board, but now video doesn't work?13:37
wolfspraulhow can we trace it down?13:37
aw_i think using test program to check if I2C is pass. i got "Unexpected register value: 0x19"13:45
aw_"When TIM_OE is 1, HS, VS, and FIELD are forced active all the time." and "When TIM_OE is 0 (default), HS, VS, and FIELD are threestated according to the TOD bit." so a TIM_OE is 1 for surely now. But this set by s/w or decode itself?13:47
lekernelah, moment13:48
aw_another truth is  the oscillate freg voltage(AC) on C225 is small @ 100mV, C226 is @ 64mV. A good voltage of 7181B: C225 @ 284mV, C226 @ 240mV.13:48
aw_so maybe just oscillate voltage problem? or s/w some where? now I am confused at all. ;-)13:49
lekernelI think 0x19 is probably ok, you can ignore this problem13:49
lekerneldoes the capture work afterwards?13:49
aw_really?? yes I got 0x1913:49
aw_when i entered gui mode13:50
lekernelit's just the chip ID13:50
aw_there's no video shown13:50
lekerneland what about the test program?13:50
lekerneldoes capture work there? what messages does it print?13:50
aw_Waiting for a captured frame. f to fail test, s to skip.13:51
aw_LOCKS: 213:51
aw_AD_RESULT: 013:51
lekerneldoes it display something?13:51
aw_none shows on display13:51
lekernelnone? what is "none"?13:52
lekernelblack? blue? white? garbage?13:52
lekernelcan you try without video signal?13:52
lekerneldisconnect video signal, reboot, wait a few seconds, run video in test13:53
aw_run gui or run test program?13:53
lekerneltest program13:54
lekerneldon't touch the gui yet13:54
aw_same result after disconnect video signal13:56
aw_i saw SDA signal there but not sure if it's okay though.13:57
aw_so 0x19 now is 7181C's ID correctly?13:59
aw_if yes, at least explains my soldering is now problem.13:59
lekernelwell, check the datasheet...14:00
lekernelthat's the value from register 0x1114:00
lekerneland yes, 0x19 is the correct value, see p. 22414:06
aw_downlaoding.... great 0x19. so at least  soldering is okay...and can we say at leasi also that TIM_OE is "1" now to force active all the time(I think).14:08
GitHub152[autotest-m1] sbourdeauducq pushed 2 new commits to master: http://git.io/4sIlgQ14:09
GitHub152[autotest-m1/master] videoin: also accept chip ID 0x19 (7181C) - Sebastien Bourdeauducq14:09
GitHub152[autotest-m1/master] Merge branch 'master' of github.com:milkymist/autotest-m1 - Sebastien Bourdeauducq14:09
lekernelwell, digital power supply and I2C soldering is OK14:09
lekernelmaybe the 7181C needs a slightly different I2C programming... can you check using the datasheet?14:11
aw_yes, yeah..saw it. wow...great now the source are also accepts (7181C) thanks.14:11
lekernelthat's the programming used by the test program: https://github.com/milkymist/milkymist/blob/master/software/libhal/vin.c#L12814:12
aw_yes, i'm checking this: https://github.com/milkymist/autotest-m1/blob/master/src/tests_videoin.c14:12
lekernelwrite 0x00 to 0x15, 0x41 to 0x17, etc.14:12
aw_aha...i see now.14:13
lekernelthere's a lot of useless or semi-useless writes in this, it can be simplified to14:13
lekernelstatic const char vreg_addr[] = {14:13
lekernel  0x1d, 0xc3, 0xc414:13
lekernelstatic const char vreg_dat[] = {14:13
lekernel  0x40, 0x05, 0x8014:13
aw_I'll check each address of reg. surely needs to be compared 7181C14:13
lekernelprobably the 3-register sequence is easier :)14:14
lekernelthe 7181b works with only these 3 registers14:14
aw_although i have long long time ago no write c codes but see / check this ...i think i can do. but for make i need to learn again. phew....14:15
lekernelit's completely trivial, all it means is that 0x40 is written to register 0x1d, 0x05 to reg 0xc3, etc.14:16
aw_btw, what difference between your link and /autotest-m1/..../tests_videoin.c ?14:18
lekerneltests_videoin.c uses the library which contains the code I linked14:18
aw_alright, I'll check them tomorrow, maybe can find out or maybe not from me. well...I'll try. ;-)14:21
aw_thanks for helping me clarify the 0x19 result.14:21
lekernelI'm setting up http://milkymist.org/blog/ - who wants an account?15:04
qi-botThe Firmware build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-12012011-1513/16:03
lekernelmake that http://milkymist.org/wp/17:00
kristianpauloh xilinx donate milkymsit for show its logo here?17:08
kristianpaulvery nice banner17:08
wpwraknice design ! navigation may be a it confusing without the customary sidebar, but let's see17:25
kristianpaulyeah design very fresh17:31
sh4rm4(http://milkymist.org/wp/) when i hover the outermost right item on the page the text is cut17:47
sh4rm4that's the one right of the flickernoise screenshot17:48
sh4rm4browser is chrome17:48
lekernelcan you send a screenshot?17:48
lekernelit works for me, of course - also with chrome ...17:48
lekernelah, on the last page17:48
lekernelseems to be a theme bug when there are fewer than 4 pictures :(17:49
lekernelprobably not worth spending my time fixing, hopefully we're going to have a lot of posts and no one will scroll this far17:49
lekernelyes, indeed... thanks for the screenshot17:49
sh4rm4also on tutorials the black bar with the name of the author is somewhat hard to spot on a blackish bg image17:51
sh4rm4maybe a slight gradient or a 1pixel white border could make it more distinct17:55
rohhm.. maybe ease up on the js animations and show one of the pics expanded by default18:11
rohits very slow on my broser and by default it doesnt show anything nice18:11
lekernelthis is a sign of obsolete hardware18:12
rohno. bad programming.18:13
rohi wouldnt call c2d cpus obsolete btw.18:13
lekernelweb designers are bad programmers. nothing new under the sun :)18:13
rohlekernel: well. on js there are 2 kinds.. the one thats useful and serves a purpose/makes things nicer, and the other one which makes stuff slow for no gain whatsoever.18:14
rohwhat i means is that its not the functionality which makes it slow, its how its done. but i am not that good in js to know why some constructs are slow. just noticed the sluggishness and the spiking cpuload18:15
kristianpaulsigh, and i bought finally a dual core netbook because that js slowdonws :-| (also to run ise ;-))18:27
wpwraklekernel: seems that your blog already inspired some technological progress :)18:29
wpwrakkristianpaul buying a more powerful netbook ;-)18:37
wpwrakbtw, the plan for cache-with-MMU was something along L1: virtual with physical tags (also keys ?), and L2: physical ?18:37
wpwraki.e., L1 latency would be max(latency(cache), latency(MMU)) on a cache hit18:40
kristianpaulwpwrak: yeah :)18:41
wpwrakas opposed to just latency(cache) on a hit for virtual without physical tags (but frequent flushing)18:41
wpwrakmhm. efficient but also demanding.18:45
lekerneldemanding? why?18:46
wpwrakyou still need a fast mmu. without phys tags, the cache acts as your tlb. but of course, software has to work harder (aliasing, flushing)18:48
lekernelof course I need a fast MMU :) lower FPS counts a "regression" :)19:01
wpwrakah, but without physical tags, you wouldn't care as long as your working set stays in the cache :)19:03
rohsometimes i wish we wouldnt need caches as workarounds to the 'ram too slow' issue.19:06
lekernelvirtual index/physical tag isn't much harder to implement ...19:06
rohwould make quite some things much easier. imagine all your ram at full busclock19:06
wpwrakroh: well, the M1's lm32 core is kinda slowish ... maybe with a bit of tweaking, your dream can come true ;-)19:09
rohwpwrak: hrhrhr19:12
lars_sometimes i wish we wouldn't need ram as workarounds to the 'disk to slow' issue19:12
rohlars_: they are workin on that one.19:13
rohonly problem: sw must be much higher quality then. the 'switch it of an on again' will not work with persistant memory.19:13
lars_and probably the only reason why i need hard disks is because of the slow internet issue19:15
wpwrakroh: naw, you'd just wipe out the transient area before using it, just like today19:18
wpwraklars_: and of course, the internet is only so slow because people don't type in things faster :)19:19
lars_wpwrak: given enough monkeys with trigger buttons generating ones and zeros in your basement the internet becomes obsolete?19:20
wpwrakwe should invest in banana plantation stock19:22
wpwrakand monkey feces management services. hmm .. that still means we have to watch that facebook IPO19:23
wpwraklekernel: by the way, in my latest build, the borders in the gradients have disappeared. so it seems to come and go somewhat arbitrarily :-(19:52
lekernelyes, it does19:55
wpwrakat least there's hope that it only changes from build to build, not from run to run20:01
wolfspraulhey nice /wp/ page - great!21:41
lekernelwolfspraul: want an account so you can post your quarterly news? :-)21:42
wpwraki think the quarterly news (soon: annual address) need a non-wolfgang solution. the mismatch between effort needed and available motivation has been steadily increasing, until it seems to have reached the breaking point21:45
wolfspraulwpwrak: see it more positively. it's slow for 2 reasons:22:03
wolfspraul#1 - I am constantly hoping real-soon-now I get some great videos like the masala video or cheap dinosaur high-res22:03
wolfspraul#2 - I keep thinking about the marketing message, which is real work and needs time and talk/meetings with people to gather feedback and so on22:03
wpwrakhehe ;-)22:04
wolfsprauland unless there are emergencies, I am back at the page in a bit22:04
wolfspraul6 AM here, no complaints I don't start early! :-)22:04
wpwrakthat's your guilty consciousness not letting you sleep before the news are out ;-)22:05
wpwrakhmm, "masala" must mean something naughty in india :) first google hit for "masala video": "[...] the hottest real hot video of a south indian women doing with a worker [...]"22:08
wpwrakand the following hits promise to be in the same vein22:09
wpwrakof course, we've been promised "3 hot young girls" who were supposedly "literally the center [...] of the party" (the one in warsaw), so that video ought to be interesting :)22:11
wpwraknice. my usb-midi dongle works, too. we can drop MIDI IN/OUT now. (just kidding :)22:23
wpwrakbtw, did we ever get a reaction on the NOR flawed unlocking protocol used by urjtag ?22:38
wpwraks/NOR flawed/flawed NOR/22:38
wpwrakand why isn't qi-bot editing ? :-)22:38
--- Fri Dec 2 201100:00

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