#qi-hardware IRC log for Saturday, 2011-07-09

kristianpaullol00:00
mththe last version I checked did, but I don't know if I'm fully up to date00:00
kristianpaulha, seems people here a kindle ;-)00:01
mthI'm at 7aeb7a4f0a693a1e67209e8ec3439d71326529cb00:01
mththat version I tested and have been using ever since I committed it00:02
Aylanice, it looks like kyak fixed one month ago the bug I'm having now00:02
kristianpaulbrr 23 C night, 32 C day..05:00
kristianpaulamazing how people can relly on altera megafunctions even for a 3 bits serial in/out shifter, but i guess a hurry is the main cause05:01
kristianpaulone counter left to migrate from altera megafunction to HDL..06:02
ezdagorWhy is it that with Debian, on the NanoNote there is around 2GB total drive space, but with OpenWRT there is only around 500MB?06:37
kyakezdagor: the openwrt image resides in a so called rootfs, which is 512 Mb. You have to mount the remaining space ("datafs") manually06:42
kyakit seems like debian installation is just using the whole NAND06:42
kyakif you are using one of the latest openwrt images, see the mtd.nn - it provides facilities to easily format and mount the datafs06:43
ezdagorOk. Thanks.06:43
kyaknp06:44
wpwrakkristianpaul: freezing nights, eh07:04
wpwrakkristianpaul: but just watch out, in something like 5 months, we'll catch up with you guys :)07:05
kyakwpwrak: i'm writing this message from my Ben while laying on a couch :)08:19
kyakit works better than i thought :)08:19
dvdkcool08:20
dvdktime to order one, too :)08:20
dvdkis this w/ dirtpan?08:20
kyakyeah08:20
dvdkhow does it work as a sip-phone ? :)  08:20
dvdkkidding08:20
kyaknow my screen session is always with me around the house :)08:21
kyakdvdk: it works fine when typing in ssh, but won't really work for huge data transmissions yet08:23
kyakfor example, receiving data from ftp at 2.5 K/s08:32
kyak500 b/s :)08:32
qi-botThe build has FAILED, see log here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07082011-0858/09:26
wolfspraulsip-phone, why not?09:29
wolfspraul250kbit/sec should not be the problem, more roundtrip delays, encoding, noise, protocol inefficiencies, etc.09:30
wpwrakkyak: whooo !! so cool !11:12
wpwrakkyak: that's not even supposed to be possible yet ;-))11:13
kyakwell, it is :)11:16
kyakwpwrak: what is the current consumption of atben module?11:17
wpwrakkyak: what's your ping round-trip-time ?11:18
kyakround-trip min/avg/max = 60.467/61.998/67.772 ms11:18
wpwrakkyak: current should be around 15 mA11:18
wpwrakkyak: hmm. that's atben to atusb ?11:19
kyakapproximately the same when pinging some internet host or the laptop11:19
kyakyeah, atben to atusb11:19
wpwrakkyak: very interesting. yours is about 2-3x faster than mine. what USB controller does the PC have ? UHCI or OHCI ?11:20
kyaki have both EHCI and UHCI, trying to figure out where exacly is atusb plugged in..11:23
wpwrakshould be the UHCI. i think EHCI is only used for high-speed.11:23
kyakbut it's definitely not OHCI11:23
kyakthen it's UHCI11:24
wpwrakvery interesting. something happens a lot faster in your PC than in mine.11:24
kyakmy laptop is very old11:24
kyakbut, atben and atusb are like 20 cm away from each other right now11:25
wpwrakoh, distance shouldn't matter for this11:25
kyakare you interested in my kernel config?11:25
kyakmight be something there11:25
wpwrak(distance) well, up to the range limit :)11:26
wpwrak(config) yes, please :)11:26
kyakhttp://downloads.qi-hardware.com/people/kyak/tmp/atusb-config11:27
kyakoh i must mention one thing11:27
kyakthe qi-kernel didn't build quite well, i had to use this patch:11:28
kyakhttps://lkml.org/lkml/2011/3/20/4911:29
kyakthis one here, and i had exactly the mentioned problem11:30
wpwrakah, interesting. didn't hit that one.11:31
wpwrakperhaps because i'm on 64 bit11:31
kyakyeah, i didn't hit it either when doing defconfig11:31
kyakbut then i used my distribution's kernel config as a base11:31
kyakand i hit it11:31
wpwrakmaybe it's the high-res timers11:33
kyakit could be.. btw, the kernel from defconfig didn't work at all with atusb11:34
kyakthe system was frozen11:34
wpwrakdid you see when it froze ?11:35
kyakwhen i inserted atusb11:35
kyakthen i used my distribution's config and make localyesconfig11:35
wpwrakbah :(11:35
kyakand it started to work fine11:36
kyakif i was in console when inserting atusb, it would output masses and masses of messages to tty and then freeze11:36
wpwraka yes, masses of junk are always a bit of a problem11:37
wpwraksometimes i just add code to the end of the trace to disable interrupts and loop11:38
wpwraklike this: http://pastebin.com/npRn7SxP11:39
kyakthese messages didn't end up with the "kernel panic" meesage11:40
kyakor perhaps the "kernel panic" message was earlier?..11:41
wpwrakhard to tell :)11:41
wpwrakhrtimers make it 0.3% faster. must be something else then ... your kernel is SMP mine isn't. wonder if that is the difference.11:56
wpwrakit's a scarily big switch, though11:57
wpwrakah, CONFIG_PREEMPT. let's see ...11:58
kyak# CONFIG_USB_OHCI_HCD is not set - perhaps your's is set and it is using OHCI (just as guess)12:01
wpwraknaw, i have an UHCI too12:02
wpwrakdoes your CPU have HT ?12:05
wpwrakCONFIG_PREEMPT made it some 0.06% faster :)12:06
wpwrakhmm, you have CONFIG_IEEE802154_DRIVER_DEBUG=y12:10
wpwrakthat should slow things down quite a bit12:11
wpwrakCONFIG_NO_HZ, 0.02% faster12:25
kyakyeah, i thought about disabling *_DEBUG, will do it on next rebuild12:27
kyakhttp://ark.intel.com/Product.aspx?id=2723512:27
kyakmy laptop's CPU12:27
kyakguess there is no HT12:28
wpwrakCONFIG_TREE_PREEMPT_RCU makes it 0.3% faster12:33
wpwrakoh, there is. it's SMP and HT.12:34
wpwraknot such an ancient laptop then12:34
kyakif i had four cores, would it be 4 times faster? :)12:37
wpwraktwice at best. you already have two cores :)12:38
wpwrakCONFIG_CGROUPS+CONFIG_CGROUP_SCHED make it slower again12:44
--- Sun Jul 10 201100:00

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