#milkymist IRC log for Wednesday, 2011-10-05

wolfspraulcan we have some way to put the m1 into an 'automatically boot' mode?01:08
wolfspraulit would probably require a modification in the standby bitstream01:08
wolfspraulmaybe some setting one has in the gui, to enable/disable auto-boot?01:09
wolfspraulactually thinking about this more, wouldn't most people who power-up their m1 want to boot anyway?01:10
wolfspraulmaybe the default should be to auto-boot01:10
kristianpaulyou mena avoid to press middle buttong after plug power cord?01:12
kristianpaulenable/disable may be will be achieved by flashing the current standby bitstream and the automatic boot one when disable?01:15
wolfsprauldon't understand :-)01:20
wolfspraulyes, I mean no need to press middle button01:20
wolfsprauljust skip that and boot automatically01:20
wolfspraulmaybe controlled by a setting in the gui, but now I think even by default it can just do that01:21
wolfspraulthe use cases of most people are such that they plug in/power m1 to start rendering, always01:21
kristianpaulhe, i mean, we can have two standby bitstream, one for automatically boot other for classic standby boot01:29
wolfspraulI'm mostly interested in the behavior of the boxes I ship01:41
wolfspraulright now it's just a question - wouldn't it make more sense to automatically boot by default?01:42
wolfspraulin the use cases I have seen so far, my answer would be: yes01:42
wolfspraulmaybe there is a law somewhere that makes this 'illegal' for environmental reasons?01:42
wolfspraulwasting energy is not good of course, if it ran silently that would be bad too01:43
wolfspraulwell, just thinking01:43
wpwrakwolfspraul: power on to boot makes a lot of sense to me :)06:39
wpwrakone problem may be how to distinguish power off (i.e., stay quiet until middle button) from power on06:40
wolfspraulmaybe it should just never be in this state? :-)06:45
wpwrakdunno if it has any power advantages06:47
wpwrakbut yes, i sometimes wonder myself :)06:47
wolfspraulwe do need to fix the last power-off weaknesses for that though06:48
wolfspraulwe cannot recommend a workaround (software power off), and then remove that workaround06:49
wolfspraulnobody will follow that recommendation anyway06:49
wpwrakyeah :)06:49
wpwrakof course, they have a ~one in 500 chance to get it right :)06:49
wpwrakwhich, according to murphy, means that disaster will inevitably on the first important event06:50
wpwraks/on/strike on/06:50
wpwrakand afaik, this recommendation hasn't really been communicated either :)06:50
wolfspraulI won't even try, we have zero chance to get people to follow06:51
wolfspraulI need to focus on the "press middle power button to boot" message :-)06:51
wpwrakhmm, some will. especially after the first mishap06:51
wolfspraulnah. locking makes it safe enough.06:51
wpwrakyes. "middle button to boot" is bad.06:51
wpwrakmaybe it exists to have a means to select rescue06:52
wpwrakof course, the same could be accomplished by changing the path to rescue from a button combination to "plug in power while pressing button"06:53
wpwraki would also love to have a lot more interaction on the display. OSD feedback on actions and such. of course, you'd drop all this for performance, but while getting to know the machine, it's not nice for it to operate in "expert mode"06:54
wpwrakin other words, i wish we had some people who'd like to get their hands dirty with the GUI :)06:55
sb0the main priority, usability-wise, is to get GOOD video tutorials and demos atm07:59
sb0the current videos are lousy, which both gives a bad impression about the product and doesn't explain much08:00
sb0I'm going to a small studio on Friday and try to improve that08:00
wolfspraulsb0: perfect!08:07
wolfspraulunboxing video would be great too08:07
sb0this is what I plan on doing08:08
wolfspraulI haven't checked the masala video links you posted once because you said only 1/3rd of the screen is visible and better ones are coming08:08
sb0plus footage with VGA grabber + camera setup, hopefully a good one08:08
sb0yes this masala stuff is lost in slowland08:08
wolfspraulalso agata hasn't replied about video links from masala or masecki yet, oh well08:08
wolfspraulsb0: ok I'm making some progress on review units08:27
wolfspraulfirst of all tuxbrain has agreed to be a hub for a review unit (or several if we become that popular)08:27
wolfspraulan EU hub08:27
wolfspraulso shipments will be fast and cheap, and easy for EU people to return08:27
wolfspraulthen tuxbrain still has two rc2 boards he never sold, and I will buy them back08:28
wolfspraulAdam will apply all of the important fixes to them including wolfson codec, then those 2 boards turn into review unit boards08:28
wolfspraulthat way we solve several problems at once08:28
wolfspraultuxbrain avoids a loss, sells back to me08:28
wolfspraulwe improve the quality of those 2 boards08:29
wolfspraulwe use them as review units08:29
sb0this will take weeks... can you ship a rc3 to the UK right now?08:29
wolfspraulI was planning to use the returned rc2 review unit for that08:33
wolfspraulwe cannot make it in time for China Music anyway, tuxbrain sends tomorrow with fedex, the review unit should be in the UK end of next week08:33
wolfsprauldid you demo m1 to massimo already?08:34
sb0no, it was just a quick meeting08:34
sb0he just saw it, but not running08:34
sb0if we ship the unit now instead of waiting for rc2 to come back, rework, packaging, etc. we can make it in time08:35
wolfspraulmaybe, but it feels wasteful. I am trying to speedup everything but wasting resources is not a good approach imo.08:41
wolfspraulwhat is the connection between the unit arriving in the UK to be reviewed by his partner and him traveling to China Music?08:42
wolfspraulI dont' understand how those 2 things are related.08:42
wolfspraulhow about your m1 unit? that's probably the fastest to get to the UK :-)08:42
sb0well I guess that will be the solution08:42
wolfspraulI thought you try to sell it, but if you start using it then after some time it's not in sellable condition anyway.08:42
wolfspraulwhat is the plan with that unit?08:43
wolfspraulmaybe you should have two in general08:43
wolfspraulone in perfect condition to sell, another one to show/demo or borrow out08:43
sb0I use it and demo it to people. including showing the packaging08:43
wolfspraulthe problem in borrowing out is the followup08:43
wolfsprauland getting it back in good condition or at all08:43
wolfspraulwhich is also not so much a financial problem, but the problem is unless you followup, things just disappear and *nobody* benefits08:44
wolfspraullesson #1 in economics: anything that is free will be wasted08:44
wolfspraulso that's the hard part, to followup and make sure it's used for something good08:44
wolfspraulI have a solid plan now on those 2 review units, and rushing as much as possible08:45
wolfsprauldavid will fedex tomorrow08:45
wolfsprauladam has all parts ready08:45
wolfspraulmany good problems are being taken care of at once08:45
wolfspraulwhat is the connection between the UK review unit and the China trip?08:45
sb0he just said he would prefer that. ask him for details08:47
sb0in either case, not being slow is god08:47
wolfspraulyou bet08:48
wolfspraulthat's why we send everything with fedex08:48
wolfsprauland even priority on the 10kg and 25kg boxes08:48
wolfspraulyou can send your unit08:49
wolfspraulyou should have 2 units08:49
wolfspraulpurpose: one for sale (perfect condition), one for demo/borrow08:49
wolfspraulthe reason you should have one for sale is that you can filter people :-)08:49
wolfspraulbuy or shut up08:49
wolfspraulmeeting only if you buy in cash at the meeting08:49
wolfspraulyou can do those things only if you have a perfect unit for sale08:50
sb0ok i will go to fedex this afternoon09:05
sb0technical progress? that's more a problem for xilinx11:08
wolfspraulI mean comparison with more megahertz, megabytes, mega-everything11:20
wolfspraulwpwrak: how about this CE0 and we_n pullup thing?13:05
wolfspraulis that someone we can or should rework? do any tests with?13:06
wolfspraulsb0: what do you think?13:06
wpwraksee my reply on the list ;-)13:06
wolfspraulyes I saw it13:07
wpwraki think it makes sense. don't know if this helps in this specific case, but it very well might13:07
wolfspraulmy question is more practical - can this be added on the rc3 boards?13:07
wolfspraulcan you try with your board and rerun some tests? parts missing?13:07
wolfspraulshould Adam test something?13:07
wpwraklemme have a peek ...13:07
wpwrakWE# is relatively easy to access near the NOR. CE0 harder. let's see where they are on the FPGA side.13:13
wpwrakunfriendly on the FPGA side13:17
wpwraknow, let's look at the real PCB ...13:17
wpwrakyeah, should be possible to rework it with 0402 parts. there's also access to 3.3 V nearby on both cases. connecting it will be a little messy13:20
wpwrakdo i have 4k7 at 0402 ... checking ... yes. a whole vial full of them :)13:22
wpwrakbut before trying such a rework, i want to bring the testing process more under control. it still runs away too often for my taste. more control means accelerating it to reduce the risk of getting outliers that take a week of hammering before they show up, and also controlling the temperature13:23
wpwraknot quite sure yet whether i'll try air conditioning or go straight for the fridge13:24
wpwrakprobably best to set up air conditioning and measure whether the temperature really stays reasonably constant (and low) throughout the day13:29
wolfspraulwpwrak: hmm, ok good13:56
wolfspraulso what are the next steps?13:56
wpwraktomorrow, my guest room will be available for air conditioning experiments. so i'll set the AC to 18 C, and check the temperature every hour or so.13:58
wpwrakwiring up the fridge probably has to wait until next week (or, if the air conditioning performs well, i'll then move the M1 over there).13:59
wolfspraulshould we ask Adam about trying this rework?13:59
wpwraki also need to automate the "did it boot" check, which currently involves me looking at the LEDs13:59
wolfspraulor is it not so urgent and you will investigate on your side first then we approach adam later?13:59
wpwrakhmm, you can ask him what he thinks about doing this on a number of devices14:00
wpwraki think the torture-testing is easier on my side, because i can automate it14:01
wolfspraulso the question is whether we can add a 4.7k pullup on ce0 and we# ?14:04
wpwrakyes, and whether adam things it will stay there :) (mechanically)14:06
wpwrakand the question on my side would be to see if it actually makes a difference14:06
wolfspraulwe can also just schedule it for rc414:06
wolfspraulor only one of the two, if that is easier on rc3 and still makes a positive impact14:07
wpwrakthat would depend on adam's verdict. if he doesn't like it for rework risk, overhead, or reliability, then it would become an rc4 item14:10
wpwrakif he's happy to mass-rework, we need to check if it actually accomplishes anything14:11
wpwrakwell, if he's VERY happy to mass-rework, we can just do it if it doesn't make things worse ;-)14:11
GitHub64[llvm-lm32] jpbonn pushed 390 new commits to master: http://git.io/NwEHzQ20:05
GitHub64[llvm-lm32/master] Synthesize SSE3/AVX 128 bit horizontal add/sub instructions from - Duncan Sands20:05
GitHub64[llvm-lm32/master] Add a MinNumRegs argument to MRI::constrainRegClass(). - Jakob Stoklund Olesen20:05
GitHub64[llvm-lm32/master] Constrain register classes instead of emitting copies. - Jakob Stoklund Olesen20:05
GitHub133[clang-lm32] jpbonn pushed 564 new commits to master: http://git.io/pIH7ng20:06
GitHub133[clang-lm32/master] Fix debug info encodings for char16_t and char32_t. - Devang Patel20:06
GitHub133[clang-lm32/master] [analyzer] Test for -analyze-function on ObjectiveC to accompany r139439. - Anna Zaks20:06
GitHub133[clang-lm32/master] [analyzer] Fix a new failure encountered while building Adium exposed as a result of r138196(radar://10087620). ObjectiveC property of type int has a value of type ObjCPropRef, which is a Loc.  - Anna Zaks20:06
--- Thu Oct 6 201100:00

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