mth | NO_HZ only applies when the system is idle, not with one task running | 00:00 |
---|---|---|
DocScrutinizer06 | U might e right | 00:00 |
DocScrutinizer06 | o/ | 00:01 |
mth | Ayla: can the USB controller be suspended while the freq change happens? | 00:02 |
mth | like we do for the MSC | 00:02 |
DocScrutinizer06 | if you wanna drive this ahead, w/o me playing advocado diabolo(?), provide CPU cycles breakdowns, for the IRQ handler invoked every other time, vs the one task running. Give a ratio | 00:03 |
Ayla | I don't see a reason why it couldn't | 00:03 |
DocScrutinizer06 | Ayla: suspending USB is a friggin complicated and annoying ting to do | 00:06 |
DocScrutinizer06 | and takes tiiiiime | 00:06 |
Ayla | how much time? | 00:07 |
DocScrutinizer06 | but depending on controller you don't need to suspend | 00:07 |
DocScrutinizer06 | (howm much) milliseconds | 00:07 |
DocScrutinizer06 | several | 00:07 |
DocScrutinizer06 | you wanna switch clock so fast it "doesn't even matter" | 00:08 |
DocScrutinizer06 | ideally clock of usb uart is independent | 00:08 |
Ayla | ideally. | 00:09 |
mth | Ayla: I don't think we should be doing PLL changes on demand anyway | 00:09 |
mth | only divider changes | 00:09 |
DocScrutinizer06 | everything after uart is basically not related to clock | 00:09 |
mth | and then only the CPU divider has to be changed | 00:09 |
DocScrutinizer06 | mth: +++ | 00:09 |
Ayla | yes | 00:09 |
DocScrutinizer06 | thats the way | 00:10 |
DocScrutinizer06 | you need to consider other timeframe based hw IF too | 00:10 |
DocScrutinizer06 | like rs232 | 00:11 |
Ayla | but that differs from how the "ondemand" governor works | 00:11 |
DocScrutinizer06 | basically all async hw if | 00:11 |
mth | we can write our own governor | 00:11 |
mth | I'm not even sure how much is gained by running at a lower freq | 00:11 |
mth | vs running fast and then being idle for a while | 00:11 |
DocScrutinizer06 | the trick is to provide an API that changes CPU clock *only* | 00:12 |
DocScrutinizer06 | mth: nuttin | 00:12 |
DocScrutinizer06 | speed to idle | 00:12 |
Ayla | we'd gain a couple of milliwatts | 00:12 |
DocScrutinizer06 | of course depending on overhead | 00:12 |
DocScrutinizer06 | for switching | 00:12 |
mth | Ayla: how did you measure that? | 00:13 |
Ayla | with a multimeter :p | 00:13 |
DocScrutinizer06 | ~listvalues to idle | 00:13 |
mth | I mean, how did you generate the load? | 00:13 |
infobot | Factoid search of 'to idle' by value returned no results. | 00:13 |
DocScrutinizer06 | ~listvalues idle | 00:13 |
infobot | Factoid search of 'idle' by value (18 of 52): /proc/uptime ;; 50 fps ;; atrap a285 ;; autolog ;; cmd: grope (.*?) ;; condor ;; disk striping with parity ;; efnet server ;; gpppkill ;; grope ;; hlt ;; idle ;; jamuraa ;; jargon finger ;; killer ;; loll ;; note: the point of the game ;; now kirma's task-bootstrap upload. | 00:13 |
Ayla | using "yes" IIRC | 00:14 |
mth | but that generated an unlimited load | 00:14 |
DocScrutinizer06 | meh | 00:14 |
mth | in which case an ondemand-style governor won't scale down | 00:14 |
DocScrutinizer06 | shouldn't | 00:15 |
Ayla | I didn't measure with the ondemand governor | 00:15 |
mth | you'd need a load of about 30% or so to get a useful measurement | 00:15 |
DocScrutinizer06 | calculate prime factors of 8744 every hsync | 00:15 |
mth | let's say you compute primes, then computing all primes will simply run faster on a higher clock, so you do more work while consuming more energy | 00:15 |
DocScrutinizer06 | err vsync | 00:15 |
DocScrutinizer06 | ;-) | 00:15 |
mth | computing a fixed number of primes per second would be a fair load to compare on | 00:16 |
DocScrutinizer06 | yep | 00:16 |
Ayla | I'd say the latter is better | 00:16 |
mth | so you'd get to see the effect of a slow clock being busy 60% of the time vs a fast clock busy 30% of the time | 00:16 |
DocScrutinizer06 | rush to idle | 00:16 |
DocScrutinizer06 | is it | 00:16 |
DocScrutinizer06 | usually | 00:16 |
Ayla | judging from the general performance gaps between frequencies, and http://wiki.dingoonity.org/index.php?title=OpenDingux:Development#Power_consumption_measurements | 00:17 |
DocScrutinizer06 | unless idle is <10% | 00:17 |
mth | in PC overclocking, lower freqs save power because the high freqs need a higher voltage supplied to the CPU | 00:17 |
mth | but we're not changing any voltages | 00:17 |
DocScrutinizer06 | where "10" is a guestimatedepending on overhead o switch clock speeds | 00:17 |
mth | that is with the PLL at 34 vs 408 MHz though | 00:18 |
DocScrutinizer06 | idiots at #maemo do OC with UNDERvolting ;-D | 00:18 |
mth | could you also measure the PLL at 408 MHz and a high CPU divider? | 00:18 |
mth | DocScrutinizer06: well, if overheating is the enemy and not instability, it might work | 00:19 |
Ayla | how could I force a high divider? | 00:19 |
whitequark | DocScrutinizer06: on jz4740 everything is bound to master clock | 00:19 |
mth | hardcode it in the cpufreq driver is the easiest way, I guess | 00:19 |
whitequark | hence you cannot easily scale cpufreq if you want your peripherals to work | 00:19 |
whitequark | lcd, usb, ram... everything | 00:19 |
DocScrutinizer06 | overheating is 3rd or 5th concern though | 00:19 |
whitequark | exceptionally stupid design imo | 00:20 |
DocScrutinizer06 | whitequark: we had same on S3C2410 iirc | 00:20 |
mth | yes, you can only scale cpufreq drastically, like 1/3 or 1/2 of the PLL freq, but not 80% | 00:20 |
DocScrutinizer06 | on s2c2442 iirc it been decoupled | 00:20 |
whitequark | s3c? sounds samsung, hmm | 00:20 |
DocScrutinizer06 | samsung S3C, Neo | 00:21 |
DocScrutinizer06 | O | 00:21 |
mth | I even had to patch the cpufreq infrastructure to get this PLL-changing approach to work at all | 00:21 |
DocScrutinizer06 | M | 00:21 |
whitequark | yeah, already found | 00:21 |
DocScrutinizer06 | OM even | 00:21 |
whitequark | mth: wasn't there also some other issue, like cpu clock being bound to ram/bus/dma clock? | 00:22 |
whitequark | where you could theoretically scale it 1/3 or 1/2, but it won't work in practice due to bugs or some other issues | 00:22 |
DocScrutinizer06 | RAM bus, sure | 00:22 |
mth | there are various restrictions about the dividers | 00:22 |
whitequark | I recall reading the datasheet and thinking that one cannot really do *any* cpufreq scaling on that chip | 00:22 |
mth | also, there is the case where raising the PLL freq means you have to increase the memory divider and you end up with a slower system overall | 00:22 |
DocScrutinizer06 | that's why many devices don't run RAM at max speed | 00:23 |
mth | since slower SDRAM has more impact than a slightly faster CPU | 00:23 |
DocScrutinizer06 | incl gta02 iirc | 00:23 |
whitequark | yeah | 00:23 |
whitequark | jz4740 divider system isn't suited to scaling at all | 00:24 |
mth | we should be testing CL=2 vs CL=3 for the SDRAM, it's possible CL=3 aligns better with the high PLL values we'd like for the CPU | 00:24 |
whitequark | it's a set-and-forget system | 00:24 |
whitequark | I'd say that proper wait states would give more in that case | 00:24 |
mth | except that our users like to overclock a bit if it means getting 60 fps instead of 45 fps from an emulator or game | 00:24 |
whitequark | well, it | 00:25 |
whitequark | 's pretty much the same set-and-forget | 00:25 |
mth | CL is a kind of wait state, isn't it? | 00:25 |
whitequark | mhm... CL? | 00:25 |
mth | cas latency | 00:25 |
whitequark | ah, yes | 00:25 |
whitequark | well, I'm not sure how exactly it affects power consumption | 00:26 |
Ayla | it does align better with the CPU, that's for sure | 00:26 |
whitequark | with static circuits it's the count of cycles which matters, and (I hope) the CPU is pipelined anyway | 00:26 |
mth | probably CL=2 is better for that; fewer clock pulses can only be good | 00:26 |
mth | but I don't know if the difference is significant or not | 00:26 |
Ayla | can't we reconfigure the RAM on demand? | 00:27 |
whitequark | pretty perversive thing to do | 00:27 |
mth | we can write to its config regs at any time afaik, but I don't know if it will have side effects | 00:27 |
whitequark | I'd say that it will introduce general system instability through transition effects | 00:28 |
whitequark | i.e. if the cpu will decide it's time to fill its cache exactly at the time another cycle would be inserted | 00:28 |
whitequark | and we don't know much about 4740 internals | 00:29 |
mth | Ayla: see it as a selling point for ubiboot ;) | 00:29 |
Ayla | it wouldn't be hard to patch u-boot | 00:29 |
whitequark | u-boot entirely consists of patches, no? :) | 00:29 |
whitequark | of the vendor quality, mostly | 00:30 |
mth | with jzboot you can simply set the CL value in the config file | 00:30 |
Ayla | and currently ubiboot sets CL=2 too | 00:30 |
mth | I don't know what u-boot consists of exactly, I got completely lost in its sources | 00:30 |
whitequark | pure shit meters deep | 00:31 |
whitequark | just like C is a pdp-11 assembler dreaming it's a programming language, u-boot is a yet another secondary bootloader which thinks it's an OS | 00:32 |
whitequark | shut up and load the damned sectors! | 00:34 |
whitequark | or pages | 00:34 |
whitequark | or whatever unit your non-volatile storage has | 00:34 |
whitequark | fucking insomnia. | 00:35 |
Ayla | without insomnias, I wouldn't go very fast on my projects ;) | 00:36 |
mth | not for me: I still sleep 7 hours, just weird hours | 00:37 |
mth | so there is no time gained | 00:37 |
mth | and quite a lot of time lost if I go to bed and don't fall asleep | 00:37 |
whitequark | mth: yeah, I understand you well | 00:37 |
whitequark | if I'm let to myself I'd go to bed at 8AM and wake up after 7-12 hours | 00:38 |
whitequark | but unfortunately every other person in this world lives in exactly invert intervals | 00:38 |
Ayla | how can you sleep 7 hours when you go to sleep at 5am? You don't go to work? :) | 00:38 |
mth | if you work at home, you can pick any hours of the day you like | 00:39 |
whitequark | you can have different kinds of work :) but yes, I don't work in plain old office | 00:39 |
Ayla | ah, ok | 00:39 |
whitequark | well, sometimes I do, when all else fails to give profit :) | 00:39 |
whitequark | not quite often through | 00:39 |
Ayla | do you enjoy working at home? | 00:39 |
mth | yes and no | 00:40 |
Ayla | I don't think I would like it, I'd prefer to meet some other people from time to time | 00:40 |
mth | what I miss most is a "sparring partner" to discuss design ideas | 00:40 |
whitequark | I generally live with my own projects, whatever they are | 00:40 |
whitequark | through working at home means more procrastination | 00:40 |
mth | not necessarily in the same room, but more than one person on a project | 00:40 |
whitequark | mth: +1000 | 00:41 |
whitequark | public irc channels often help, through | 00:41 |
mth | I do meet other people almost every day, but they're not colleagues | 00:41 |
whitequark | also, when I do work at the daytime, I often do that in a cafe | 00:42 |
whitequark | provides a nice background noise | 00:42 |
DocScrutinizer06 | duh, shutup and don't awake me :-P | 00:42 |
Action: DocScrutinizer06 scrolls back to see how loing he slept | 00:42 | |
whitequark | DocScrutinizer06: when I can't sleep, no one can :D | 00:42 |
Ayla | whitequark: hey, nice | 00:42 |
DocScrutinizer06 | wow, 15min | 00:42 |
mth | I use a different computer with no mail or IRC or anything else distracting running | 00:43 |
whitequark | DocScrutinizer06: do you sleep with your n900? :) | 00:43 |
mth | and put on some music that I've listened to before, so it doesn't draw much attention | 00:43 |
DocScrutinizer06 | nah, we got separate beds | 00:43 |
DocScrutinizer06 | this is lenovo t500 though | 00:43 |
mth | I have done meetings in a cafe though | 00:44 |
whitequark | mth: I just run irc in irssi. it doesn't capture my attention by its own then, same for email. And when I need to get distracted, I just switch desktops | 00:44 |
whitequark | also, yes, music | 00:44 |
mth | in theory my mail doesn't have to distract me, but if I see unread mail I want to check it, even though I know there is no reason to | 00:44 |
mth | but without a mail counter I actually check it more often | 00:44 |
whitequark | I had some of my most productive hours when I were sleep deprived and listening to the same track for 200th or so time | 00:44 |
whitequark | er, I was | 00:45 |
Action: DocScrutinizer06 killed mailcounter systray icon, whe the 5 digits didn't fit anymore | 00:45 | |
whitequark | e.g., yesterday was the first time I truly lead one of my major projects to 1.0.0 point | 00:46 |
whitequark | at the last 8h I did more than in previous four days | 00:46 |
mth | I excluded most mailinglists from the counter, it's more like an inbox counter now | 00:46 |
DocScrutinizer06 | I googled my name yesterday, amazing experience | 00:46 |
DocScrutinizer06 | esp since there are still others like me who do firebrigade, gardening... | 00:47 |
DocScrutinizer06 | and one even does linux kernel (NIC et al) | 00:47 |
DocScrutinizer06 | dunno how to exclude dirs from kmail counter | 00:48 |
Action: whitequark never had mailcounter. too lazy to set up | 00:48 | |
whitequark | through I have kind of a mailcounter on my phone, when it's on wifi | 00:48 |
mth | DocScrutinizer06: I'm using kmail as well, but I can't find the option anymore | 00:50 |
mth | I think it used to be in the context menu for a mail folder | 00:50 |
whitequark | DocScrutinizer06: http://talk.maemo.org/member.php?s=78b7634e2cefca7edf93bb87f77ef9c7&u=35572 is that your photo on the right? :) | 00:51 |
mth | by the way, I've had major problems after switching to akonadi storage, so if you haven't done so yet, I'd recommend not doing that for a long time | 00:51 |
mth | I once programmed a KDE2 panel applet to show new mail status (no count, just a flag) | 00:58 |
mth | before that became standard functionality | 00:58 |
Ayla | KDE 2? :) | 01:03 |
Ayla | sounds like prehistoric | 01:03 |
mth | I started with KDE 1.1, I think | 01:25 |
DocScrutinizer06 | whitequark: guess ;-) | 01:35 |
DocScrutinizer06 | http://talk.maemo.org/member.php?u=35572 | 01:38 |
Action: DocScrutinizer06 wonders what that "s=<randome large numer>" came from | 01:39 | |
DocScrutinizer06 | yeah and kde3 was so good kde4 is still struggling to catch up | 01:43 |
DocScrutinizer06 | wtf is akonadi? | 01:44 |
mth | it's some kind of storage backend separate from the mail reader itself | 01:52 |
mth | not a bad idea in itself, imo, but the implementation is very immature | 01:52 |
qi-bot | [commit] Xiangfu: milkymist: files: automatic mount exroot at /dev/sda1 (master) http://qi-hw.com/p/openwrt-packages/2fbc9e8 | 03:32 |
qi-bot | [commit] Xiangfu: urjtag: minor fix, create usr/lib folder (master) http://qi-hw.com/p/openwrt-packages/935266e | 03:40 |
qi-bot | The build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.minimal-20120619-1445 | 17:00 |
whitequark | LOL | 17:29 |
whitequark | SGS2 RIL has a SecSetIMEI call. | 17:30 |
whitequark | does what it says on the box. | 17:30 |
whitequark | DocScrutinizer06: shouldn't radios be certifier or something like that? ;) | 17:31 |
whitequark | *certified | 17:31 |
wpwrak | maybe it doesn't work ? :) | 18:17 |
wpwrak | e.g., on an RSX-11 system (RSX-11 was relatively VMS-like a PDP-11 operating system) i was once shown not only SET PASSWORD, but also a command SHOW PASSWORD. not sure if it came like that from DEC or is someone had added it a s a joke. the output suggested the latter :) | 18:30 |
larsc | were you able to run it without having to enter the password first? | 18:41 |
wpwrak | oh yes. it output several diagnostics. accessing passowrd database, then some problem, ... then it suggested to try to decode it for you, worked on it for a while, gave up, | 18:51 |
wpwrak | then it showed a world map, "war games" like. wrote some more stuff. until you eventually got a final fatal error. | 18:52 |
larsc | hehe :) | 19:08 |
viric | mh these loongson fpu things run super bad. | 19:19 |
viric | ffmpeg fails, lame fails, libvorbis fails... | 19:20 |
whitequark | wpwrak: it has been verified to work correctly | 19:59 |
whitequark | that's the worst part | 19:59 |
wpwrak | perhaps they'll try to make it less obvious the next time :) | 20:01 |
kristianpaul | cool cisco routers have altera, lattice fpga's and cpld | 20:18 |
kristianpaul | their main chip seems to be self brandend.. | 20:18 |
kristianpaul | wonder what hapennded with that network procesor core announced on open cores time ago.. | 20:20 |
wpwrak | anyone knows a nice palindrome made of up 6-10 left-right symmetrical upper-case characters ? (AHIMOTUVWXY) | 20:29 |
whitequark | wpwrak: grep thinks there are no such words in /usr/share/dict/american-english | 20:34 |
whitequark | the best match was | 20:34 |
whitequark | 1.9.3-p194 :012 > words.grep /([AHIMOTUVWXY])([AHIMOTUVWXY])([AHIMOTUVWXY])\3\2\1/i => ["Chattahoochee\n", "Chattahoochee's\n"] | 20:34 |
wpwrak | naw, thre must be some. e.g., "aha", "tit", ... | 20:35 |
whitequark | you said 6-10 | 20:35 |
wpwrak | ah, you already took that into account | 20:35 |
whitequark | yup | 20:36 |
wpwrak | didn't find any in the usual palindrome lists either. one with six characters was listed ("MAXXAM") but that seems to be a name | 20:38 |
wpwrak | well. it'll have to do for testing. thanks ! | 20:42 |
--- Thu Jun 21 2012 | 00:00 |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!