#emc-devel | Logs for 2006-04-02

[03:03:07] <cradek_> cradek_ is now known as cradek
[04:11:23] <jepler> cradek: the version control system I was playing with is git, not svn. svn is less radically different from cvs---it's still centralized, for one thing.
[04:11:55] <jepler> cradek: I'm not staying long, but I wonder if you were able to make heads or tails out of that tarball that paul_c sent to the developers list...
[04:13:17] <jmk_cf> cradek got very quiet about 20 mins ago, away maybe
[04:13:54] <jepler> ok
[04:14:12] <jmk_cf> have you done much with qemu?
[04:16:05] <jepler> only a little bit. Have a specific question I can try to answer?
[04:16:28] <jmk_cf> do you have an estimate of the slowdown vs. running on the host?
[04:16:37] <jmk_cf> I was under the impression it was 2 or 3 to 1
[04:16:51] <jmk_cf> but I just did a simple test that says its more like 20 to 1
[04:17:21] <jmk_cf> time dd if=/dev/zero of=/dev/null count=1M
[04:17:30] <jmk_cf> on host and emulated system....
[04:17:43] <jepler> I don't know offhand; I thought there were some comparisons on the qemu website, though
[04:18:00] <jmk_cf> thats where I was getting the general impression of 2 or 3 to 1
[04:18:25] <jepler> http://fabrice.bellard.free.fr/qemu/qemu-accel.html
[04:18:39] <jmk_cf> which meant that running all the compile farm "slots" on qemu boxes on a SMP P3-600 machine would be a possible win vs running them on the 200MHz existing farm
[04:18:49] <jmk_cf> but at 20:1, the whole thing is a waste
[04:19:04] <jepler> they say 5 or 10 to 1 without kqemu
[04:19:30] <jmk_cf> eww
[04:20:14] <jepler> good night
[04:20:17] <jmk_cf> the download page for the accelerator says "binary archive for QEMU 0.8.0" but the tarball is named kqemu-0.7.2,tar.gz
[04:20:46] <jmk_cf> goodnight
[06:33:09] <jmkasunich> logger_devel: bookark?
[06:33:10] <jmkasunich> I'm logging. Sorry, searching removed.
[06:33:17] <jmkasunich> logger_devel: bookmark?
[06:33:17] <jmkasunich> See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emcdevel/2006-04-02#T06-33-17
[08:17:25] <anonimasu> mornin
[09:24:13] <alex_joni> morning
[14:41:01] <jepler> jmkasunich: the kqemu and qemu versions don't necessarily match
[14:59:53] <rayh-away> rayh-away is now known as rayh
[17:36:32] <jmkasunich> jepler: I tried kqemu 0.7.2 with qemu 0.7.2, and cradek tried it with qemu 0.8.0
[17:36:45] <jmkasunich> we both found execution with kqemu slower than without
[17:36:51] <jmkasunich> I'm very dissapointed
[17:44:19] <jepler> huh
[17:44:31] <jepler> I'm sorry you wasted a bunch of time.
[17:45:06] <jmkasunich> I'm gonna check out the qemu mailing list(s) and see what they have to say
[17:45:45] <jmkasunich> I did a test yesterday, running the BDI-2.20 install on the 200MHz farm slot and building emc2 took 9 minutes
[17:46:08] <jmkasunich> running qemu on my 1.6GHz sempron box took 14 mins, and kqemu on the same box took 17 mins
[17:46:44] <jepler> with your 'time dd' example, or doing something else?
[17:46:54] <jmkasunich> building emc2
[17:47:17] <jepler> oh
[17:47:36] <jmkasunich> can't get much closer to the actual application than that
[17:47:40] <jepler> yeah, you can't
[17:48:02] <jepler> if you care about the electric bill or the noise it might still be an improvement
[17:48:36] <jmkasunich> well I was actually planning on installing all the qemus on another rackmount box
[17:48:42] <jmkasunich> smaller and quieter
[17:48:51] <jmkasunich> (I don't leave my main box on 24/7)
[17:49:03] <jmkasunich> but that box is only a dual P3-600
[17:49:18] <jmkasunich> so it would be even worse than 14 mins, probably more like 30
[17:49:21] <jmkasunich> that sucks
[17:51:27] <alex_joni> hi guys
[18:30:43] <cradek> hi all
[18:30:50] <cradek> curse the time change
[18:31:01] <jmkasunich> damn you, time change!!!
[18:31:05] <jmkasunich> there, happy
[18:31:06] <jmkasunich> ?
[18:31:08] <cradek> yes
[18:31:18] <alex_joni> * alex_joni is ok with the time change
[18:31:19] <cradek> cat had 3 kittens last night
[18:31:22] <alex_joni> already adjusted :)
[18:31:25] <alex_joni> cradek: that's nice
[18:31:37] <alex_joni> sparkplug had kittens?
[18:31:42] <cradek> yes
[18:31:57] <jmkasunich> sparkplug?
[18:32:02] <alex_joni> jmkasunich: don't ask :D
[18:32:44] <cradek> she's an outdoor cat we sort of took in, don't know what to call her so we call her something different every time, and alex suggested sparkplug
[18:32:59] <alex_joni> I didn't :D
[18:33:06] <alex_joni> but I looked at a page with cat names
[18:33:10] <alex_joni> and sparkplug was there
[18:33:14] <cradek> oh
[18:33:16] <alex_joni> and it REALLY stuck out
[18:33:19] <cradek> ok then
[18:33:26] <jmkasunich> we had a cat called napalm
[18:33:38] <jmkasunich> belonged to my (rather strange) brother-in-law before he died
[18:33:39] <alex_joni> cradek: you may call her sparky
[18:34:20] <jmkasunich> orange and black longhair, he thought she looked like an explosion
[18:34:47] <alex_joni> lol, how did he call it?
[18:34:47] <cradek> haha
[18:35:06] <alex_joni> NAPALM!!! and everybody on the street took cover?
[18:37:50] <jmkasunich> you can't call a cat
[18:38:36] <jmkasunich> we usually called her fuzzbutt anyway
[18:39:05] <cradek> you sure can call a cat, but you do it by rattling two cat food cans together
[18:39:40] <jmkasunich> that works for one of ours, but not the other
[18:39:55] <jmkasunich> (not cans, we feed dry food, but pouring it into the bowl will do the trick)
[18:40:19] <skunkworks> we have trained ours to come when we call. all it takes is tunafish. (we don't even give it to them anymore)
[18:40:29] <skunkworks> lots of tunafish in the begining
[18:41:03] <rayh> hey skunkworks, how'd that wireless link work out?
[18:41:31] <skunkworks> rained. we need to run a network cable around the outside of the house under the siding.
[18:41:52] <skunkworks> it did work when I tested it between the house and the barn - maybe 300 yards.
[18:42:00] <skunkworks> (one cantinna on one end)
[18:42:16] <alex_joni> * alex_joni loves cantennas
[18:42:47] <skunkworks> through 2 windows.
[18:42:59] <rayh> Great. I'll have to do some more planning for such a thing here.
[18:43:09] <rayh> Thanks for the report.
[18:43:32] <skunkworks> I will let you know when it is officeally up and running in its final location.
[18:43:49] <alex_joni> skunkworks: windows are very poor for transmissions
[18:43:52] <rayh> Thanks. And thanks for the antenna link.
[18:44:02] <rayh> leaded glass/
[18:44:06] <jmkasunich> heh
[18:44:27] <skunkworks> alex - right that was the test as it is how it is going to be setup. inside the house to inside another house.
[18:44:29] <jmkasunich> ray: shame I didn't think of you a couple months back
[18:44:53] <skunkworks> line of site between 2 windows.
[18:44:56] <skunkworks> for now
[18:45:05] <jmkasunich> work scrapped a bunch of wavelan PCMCIA cards.. most were rather mundand, but two had connections for external antennas, and the antennas were there too
[18:45:21] <jmkasunich> 7-8 element yagis, probably good for several miles
[18:45:21] <alex_joni> skunkworks: best it would be to drill a small hole to get the RF cable and antenna out
[18:45:44] <alex_joni> jmkasunich: yagis for 2.4 aren't very good (about 10 dB I'd say)
[18:46:03] <jmkasunich> this stuff was 915Mhz
[18:46:13] <alex_joni> oh.. then it's better
[18:46:16] <skunkworks> that will be the next step if there is an issue. (we don't really want anything visable thought hint hint)
[18:47:14] <skunkworks> jmk - did you end up drawing your simple servo circuit using freqgen (I think it was freqgen)
[18:47:23] <skunkworks> h-bridge
[18:47:36] <jmkasunich> no, I never got my round tuit
[18:47:44] <jmkasunich> keep losing that darned thing
[18:47:55] <skunkworks> :) thats ok.
[18:48:05] <alex_joni> I'd recommend this for 2.4 wireless: http://www.calamp.com/pro_mmds_130094-130135.html
[18:48:16] <jmkasunich> it doesn't help that the actual unit is still at work
[18:48:20] <jmkasunich> I should bring it home
[18:48:30] <rayh> Well I sure envy your scrap heap, John.
[18:49:21] <alex_joni> rayh: those antennas (24dB) are about 40-50$ over here
[18:49:49] <skunkworks> alex - I sent ray this http://sharperconcepts.zoovy.com/ - what do you think about there parabolic antannas?
[18:50:43] <skunkworks> http://sharperconcepts.zoovy.com/product/YSC-HG2419G 19db for 38
[18:50:53] <skunkworks> $38
[18:51:09] <rayh> I was looking at these.
[18:51:27] <rayh> But also looking at the higher gain ones.
[18:51:28] <alex_joni> 19dB is kinda ok, but not ideal
[18:51:33] <alex_joni> wouldn't recommend it for ray
[18:51:43] <skunkworks> how far are we talking?
[18:51:55] <rayh> 5 miles
[18:52:31] <alex_joni> rayh: get one of these: http://www.hyperlinktech.com/web/24ghz_high_power_802.11_amplifiers.php
[18:52:34] <skunkworks> here we go http://sharperconcepts.zoovy.com/product/YSC-HG2424G 24dbi $52
[18:53:17] <alex_joni> skunkworks: that's ok, not the best deal though, I think you can get them a bit cheaper.. I mean they sell at those prices over here.. lots of miles away
[18:53:18] <skunkworks> (I have bought from them before with good luck)
[18:53:53] <alex_joni> rayh: btw, that link I sent was a joke, you're not allowed with power over 200mW on 2.4GHz in the US
[18:54:24] <alex_joni> but a 25W amplifier would certainly rock the link :D
[18:54:33] <alex_joni> if it weren't only for military usage..
[18:54:50] <skunkworks> linksys access points and routers have firmware hacks alowing for higher power out. (not legal though)
[18:54:53] <jmkasunich> microwave oven for bugs and birds
[18:55:34] <alex_joni> skunkworks: hardware is limited to something like 200mW
[18:55:39] <alex_joni> didn't see any going higher
[18:55:45] <alex_joni> the hacks refer to europe
[18:55:50] <alex_joni> over here only 50mW is allowed
[18:56:05] <alex_joni> oh, and that's EIRP power
[18:57:07] <skunkworks> radiated power. funny. we have radio stations over here now touting like 10billion mirowatts of radiated power in there ads ;)
[18:57:30] <skunkworks> their
[18:57:32] <alex_joni> yeah, but they use commercial frequencies
[18:57:38] <rayh> As a kid I ran a 25 watt carrier current am station in California. Till I got caught.
[18:57:39] <alex_joni> where they pay licenses for
[18:58:25] <skunkworks> 5 years ago the ads where saying miliwatts. (guess that wasn't exciting enough)
[19:00:02] <skunkworks> kinda like car audio amp companies rating the wattage of there amps in peak to peak power.
[19:00:12] <skunkworks> instead of rms
[19:01:09] <rayh> Image is everything! Or at least some folk, sales, and political, seem to think so.
[19:02:39] <skunkworks> ray - that is how the local public radio station got started at the university.
[19:19:50] <alex_joni> later guys, heading for my hot tub ;)
[19:25:22] <rayh> see you alex
[19:26:14] <rayh> Sure I started or built up several student radio stations while around universities.
[19:59:50] <alex_joni> * alex_joni is ack
[19:59:54] <alex_joni> back even
[20:00:17] <skunkworks> that was a quick hot tub moment.
[20:00:42] <alex_joni> skunkworks: yeah, didn't feel like soaking :D
[20:00:51] <skunkworks> shouldn't you have wireless and a portable so you can hot tub and surf? that would be the ultimate
[20:01:00] <alex_joni> actually I do
[20:01:03] <alex_joni> LOL
[20:01:15] <alex_joni> but I need a waterproof laptop to e sure
[20:01:16] <alex_joni> :)
[20:14:53] <skunkworks> is csv still down?
[20:15:34] <rayh> yep
[20:16:05] <cradek> I expect it to be fixed no earlier than tuesday
[20:16:15] <jmkasunich> sourceforge doesn't work on weekends
[20:16:31] <skunkworks> yeck.
[20:19:12] <skunkworks> I really need to get into emc2 and play with hal and such. It works great with my steppers but I want to do more. cl and such. I have no clue where to get started. I suppose I could start out with something simple like setting up a cycle start push button or such.
[20:19:28] <skunkworks> and such and such - boy that sounds funny
[20:19:54] <alex_joni> skunkworks: there's a CL config template
[20:20:01] <skunkworks> I really need to preread my posts and such
[20:23:29] <skunkworks> I will have to look around to see if I have a encoder with an index pulse. would love to test threading. (at what point is the pulses per rev set up?) Cradek: are you running your encoder into your parrallel port? 2 lines? or 3 (index)
[20:23:40] <cradek> 3
[20:23:46] <cradek> index is required for threading
[20:24:03] <cradek> yes it runs right to the parport (actually through schmitt triggers)
[20:24:14] <cradek> ppr is simply set in hal
[20:25:35] <skunkworks> cool. Going to have to play with that. would have you had to use schmitt triggers? Is there decent logic in your encoder or is what is coming out from what ever sensor you are using?
[20:26:06] <cradek> no they are probably not needed. I added them for some noise immunity since the lines run right by the spindle motor
[20:28:44] <skunkworks> crap - I should be talking in the #emc channel
[21:40:54] <alex_joni> steves_logging: are you around by any chance?
[22:09:01] <alex_joni> good night all..
[23:24:18] <jmkasunich> jmkasunich is now known as jmk_away