wolfspra1l | cladamw: good morning in the new year! ;-) | 00:46 |
---|---|---|
cladamw | wolfspra1l, hi good morning in 2012 .:-) | 01:09 |
wolfspraul | I tried building kicad on my new i7-2670QM quad-core and finally wanted to see make -j in action... | 20:35 |
wolfspraul | here's what I got: make 9.01 min; -j2 4.30 min; -j4 3.24 min; -j6 3.10 min; -j8 3.06 min | 20:36 |
wolfspraul | so I guess from now on I will always try either -j8 or -j4 first, hopefully won't run into too many build weaknesses... | 20:37 |
wpwrak | hehe :-) | 20:40 |
wpwrak | yes, make -j saves quite a lot of time. if you have plenty of RAM to cache the whole tree, even better | 20:41 |
viric | you can build with parallelism based on the system load | 20:57 |
wolfspraul | viric: how? | 21:35 |
wpwrak | with -l perhaps ? | 21:40 |
wpwrak | but most machines have no significant background load anyway. well, except for perhaps some runaway flash player ... | 21:41 |
wpwrak | i have a script called "np" that just does killall npviewer.bin nspluginviewer incredibly handy for keeping a healthy low load average :) | 21:42 |
lars_ | wolfspraul: if you have 8 cpus -j9 | 21:45 |
wolfspraul | he. thanks! trying that same build as before with -j9 brought it to 2.38 min (without any other load), quite a bit faster than -j8 | 22:01 |
wolfspraul | could be caching or filesystem related though, who knows. will go with -j9 from now on... | 22:01 |
Ayla | try with -j64 | 22:02 |
Ayla | maybe it'll compile in less than a minute :) | 22:02 |
whitequark | wpwrak: I used to have `killall -9 python' in crontab. It worked just as well | 22:49 |
whitequark | at least on ubuntu | 22:49 |
wpwrak | ;-)) | 22:50 |
whitequark | does anyone know an open standard for low-power, low-complexity wireless sensor networks? | 23:16 |
whitequark | preferably less complex than both zigbee and 6lowpan | 23:16 |
wpwrak | you can always do your own on top of 802.15.4 :) | 23:17 |
whitequark | i.e. something where all the stack can be fit in a single Arduino (no, I'm not going to use arduinos, but that describes the limit) | 23:17 |
whitequark | hm | 23:17 |
wpwrak | (like i did with dirtpan. though i wouldn't call that a "standard" :) | 23:18 |
wpwrak | (or simpler) | 23:18 |
whitequark | interesting | 23:18 |
whitequark | can you suggest any plain 802.15.4 modules? i.e. not zigbee ones | 23:18 |
whitequark | oh well, there's even a dedicated wikipedia page on the topic: http://en.wikipedia.org/wiki/Comparison_of_802.15.4_radio_modules | 23:19 |
wpwrak | define "module" :) | 23:21 |
wpwrak | there's of course the atben/atusb design ... | 23:21 |
whitequark | anything which has SPI input and ready-to-use PCB/chip antenna | 23:22 |
whitequark | _and_ has a low cost | 23:22 |
wpwrak | atben is pretty close. you'll make you own pcb anyway, right ? | 23:22 |
wpwrak | just reuse the design | 23:22 |
whitequark | how hard do you think it will be to integrate the design on my own PCB as a part? | 23:23 |
whitequark | it's RF and stuff | 23:23 |
wpwrak | depends. is your pcb 0.8 mm ? | 23:23 |
whitequark | and I have zero experience in RF and no instruments which I can use to debug RF problems | 23:23 |
whitequark | hm | 23:23 |
wpwrak | instruments are useful to have. no matter how you approach it. also pre-built modules can run into all sorts of problems. without at least a spectrum analyzer or similar, you'll never know if your design has troubles | 23:25 |
whitequark | the prototyping service I'm currently using (iteadstudio.com) has the 0.8mm variant, and I think I can adapt all other stuff to that | 23:25 |
whitequark | well, I can't afford a spectrum analyzer, I think | 23:26 |
wpwrak | if your pcb is 1.6 mm, you'd have to adjust the pcb antenna of atben. that's unfortunately a bit of trial and error (and measurements) | 23:26 |
whitequark | they're expensive as hell | 23:26 |
wpwrak | yes. a usrp is often cheaper. but still not "cheap" | 23:26 |
whitequark | Nordic semiconductor modules are cheaper than $5 | 23:26 |
wpwrak | but sometimes you can find places/friends who can let you use their equipment. universities can be great places for that. | 23:26 |
whitequark | but they're also feature vendor lock-in | 23:27 |
wpwrak | what range requirements to you have ? and how compliant to emission standards does it have to be ? | 23:27 |
whitequark | I'm looking at all of the emerging open hardware projects on the topic ("internet of things" and similar stuff), and I've finally decided that's not an option | 23:27 |
wpwrak | so you don't want to do this after all ? | 23:28 |
whitequark | err | 23:28 |
whitequark | I was talking about vendor lock-in | 23:28 |
whitequark | not the whole project | 23:28 |
wpwrak | ah, vendor lock-in always sucks :) | 23:28 |
whitequark | (range) 20m in a house with maybe one or two reinforced concrete walls | 23:29 |
wpwrak | it may make the beancounters in the back office happy, but they're not the engineers :) | 23:29 |
wpwrak | hmm. then you need a power amp. | 23:29 |
whitequark | (compliance) I don't care unless it's emitting two times more than my WiFi AP. No one will notice anyway | 23:29 |
whitequark | I am not going to sell that commercially, nor I want to do anything with FCC (yet). It's just a pet project just for me | 23:30 |
wpwrak | don't underestimate the fragility of wlan ;-) even with the meager 2 mW my atben/atusb can put into the air, i can mess up my wlan quite badly ;-) | 23:30 |
whitequark | oh | 23:30 |
whitequark | oh well. | 23:30 |
wpwrak | the solution is of course not to make it hog the same channel | 23:30 |
whitequark | my wlan disrupts the performance of my Bluetooth mouse, so I've decided to move it to 5GHz spectrum, as it's finally allowed in Russia | 23:31 |
wpwrak | but if i'm in a mean mood, i can quite easily jam my wlan with it | 23:31 |
wpwrak | okay, i'm all 2.4 GHz :) | 23:31 |
whitequark | but I have neighbors... | 23:31 |
whitequark | hm | 23:31 |
whitequark | I never thought of wlan to be so fragile. | 23:32 |
wpwrak | if you implement the usual channel access protocols, then you should be fine | 23:32 |
wpwrak | it will clash every once in a while, but nobody will notice | 23:32 |
whitequark | does 802.15.4 phy/mac include that already? | 23:33 |
whitequark | I bet it does | 23:33 |
wpwrak | my jamming happens when i deliberately make it misbehave (but still at a fraction of the power) | 23:33 |
wpwrak | yes, that's normally there | 23:33 |
whitequark | it's fine then | 23:33 |
whitequark | (range) wait, I think I screwed up | 23:33 |
whitequark | the range is about circle, not area | 23:34 |
Action: whitequark gets his tape measure | 23:34 | |
wpwrak | if you can do with 10 m free space or maybe 2-3 m through "nasty" walls, the at86rf231 is a good chip to start with. documentation is quite complete. and you have an existing project you can fork from. | 23:34 |
wpwrak | for a larger range, you'd have to add an rf amp, though | 23:34 |
wpwrak | range is radius :) | 23:35 |
whitequark | yeah :) | 23:36 |
wpwrak | (atben/atusb) plus, you can get usb dongles with known to be good specs. or, if you have a ben, atben. that's even cheaper. | 23:36 |
whitequark | I don't have any two points more distant than 6m | 23:36 |
whitequark | and there are no concrete walls inside anyway | 23:37 |
wpwrak | that way, you can quickly compare performance. the production tests process can detect some RF issues, using just atben and atusb: http://downloads.qi-hardware.com/people/werner/wpan/prod/index.html | 23:37 |
whitequark | so I think that atben-based design will work here | 23:37 |
whitequark | hm, interesting | 23:38 |
wpwrak | i got okay reception with about 5 m through a wall. so yes, it may work | 23:39 |
wpwrak | what makes a big difference is where you locate those big bags of water, called "people" | 23:39 |
--- Tue Jan 3 2012 | 00:00 |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!