• [00:02:00] * jrmuizel (n=jrmuizel@CPE001f5be79d0f-CM0017ee62f8b0.cpe.net.cable.rogers.com) has joined #beagle
  • [00:02:17] * jrmuizel (n=jrmuizel@CPE001f5be79d0f-CM0017ee62f8b0.cpe.net.cable.rogers.com) Quit (Remote closed the connection)
  • [00:04:14] * phil_ca (n=philburk@c-76-126-101-224.hsd1.ca.comcast.net) Quit ()
  • [00:07:04] * scrp3l (n=pablo@201.250.180.40) Quit ("Ex-Chat")
  • [00:12:07] * maluta (i=bd6b8b50@gateway/web/freenode/x-a5daae059562b92a) Quit ("Page closed")
  • [00:14:52] * Batko_Marto (n=Batko_Ma@CPE001346f996d2-CM001e6b1a8d1e.cpe.net.cable.rogers.com) has joined #beagle
  • [00:19:37] * Wiedi (n=wiedi@newton-air.w.fruky.net) Quit ("^C")
  • [00:25:08] * j_ack (n=jack@p57A40218.dip0.t-ipconnect.de) has joined #beagle
  • [00:29:14] * rsalveti (n=rsalveti@200.184.118.130) Quit (Remote closed the connection)
  • [00:33:14] * brolin (n=brolin@190.158.89.183) has joined #beagle
  • [00:42:17] * hochopeper (n=Unknown@ppp118-208-39-18.lns3.bne1.internode.on.net) has joined #beagle
  • [00:47:25] * alecrim (n=alecrim@189.2.128.130) Quit (Read error: 113 (No route to host))
  • [00:49:58] * rsalveti (n=rsalveti@189.70.35.140) has joined #beagle
  • [01:00:05] * j_ack (n=jack@p57A40218.dip0.t-ipconnect.de) Quit ("Verlassend")
  • [01:13:26] * javaJake (n=javaJake@unaffiliated/javajake) Quit (Read error: 110 (Connection timed out))
  • [01:18:32] * greyback_ (n=greyback@ip-83-147-165-234.dub-3rk1.metro.digiweb.ie) Quit (Read error: 60 (Operation timed out))
  • [01:21:20] * tegila (n=tegila@189-015-135-057.xd-dynamic.ctbcnetsuper.com.br) Quit ("Leaving.")
  • [01:27:24] * billenium is now known as Pizza
  • [01:29:37] * Pizza is now known as billenium
  • [01:40:42] * brolin (n=brolin@190.158.89.183) Quit (Client Quit)
  • [01:50:18] * javaJake (n=javaJake@unaffiliated/javajake) has joined #beagle
  • [02:08:19] * javaJake (n=javaJake@unaffiliated/javajake) Quit ("night")
  • [02:34:27] <mru> omapfbplay has learned a new skill
  • [02:34:38] <mru> it can now output to Xvideo
  • [02:34:56] <mru> thus making the name a bit of a misnomer
  • [02:36:04] * Animule (n=Animal@70-56-35-197.hlrn.qwest.net) Quit ("Not all men are created equal. Only the finest become Cowboys.")
  • [02:37:07] <muriani> haahahaha
  • [02:37:15] <muriani> I went googling
  • [02:37:39] <muriani> came upon a page of build stats, and who's listed but my old buddy from genesi
  • [02:37:46] <muriani> heh, small world
  • [02:53:01] <billenium> Hmm, does anyone with a beagleboard want to do me a favor?
  • [02:53:10] * denix0 (n=denix@pool-71-251-63-232.washdc.east.verizon.net) Quit ("http://arago-project.org")
  • [02:53:13] <muriani> Depends on the favor.
  • [02:53:38] <billenium> I need to see if ecereC works on ARM/Beagleboard.
  • [02:53:57] <billenium> http://www.ecere.com/
  • [02:54:30] <billenium> If you could be so kind as to download and compile it, and test out maybe compiling a "hello world" program... I would be VERY thankful!
  • [02:56:28] <muriani> Hmm, just from a cursory glance, it seems it would be difficult to run this on the beagle directly, without some heavy modification
  • [02:56:47] <muriani> but I may be wrong. Let's give the source a try.
  • [02:56:56] <billenium> :)
  • [02:57:23] <muriani> or I would, if I could connect to my beagle
  • [02:57:33] <muriani> my network's being screwy lately :/
  • [03:00:37] <billenium> good luck if you can
  • [03:01:13] <muriani> yeah,I don't have a USB mouse, so I have to rely on VNC
  • [03:01:26] <muriani> bah
  • [03:01:30] <muriani> I'll have to check it out later
  • [03:02:04] <billenium> If you get anything /query me the results
  • [03:40:04] * lifeeth (n=praneeth@unaffiliated/lifeeth) Quit ("Up .....Up and Awayyyyyyyy... Atom Atom Ant....")
  • [03:51:55] * BThompson (n=Bernie@69.7.114.75) Quit (Read error: 110 (Connection timed out))
  • [04:03:43] * BThompson (n=Bernie@69.7.114.75) has joined #beagle
  • [04:05:35] * scruggs (n=chris@72-161-110-72.dyn.centurytel.net) Quit ("leaving")
  • [04:07:22] * dirk2 (n=dirk@p5B040F89.dip0.t-ipconnect.de) has joined #beagle
  • [04:33:28] * Animule (n=Animal@70-56-35-197.hlrn.qwest.net) has joined #beagle
  • [04:50:50] * dirk2 (n=dirk@p5B040F89.dip0.t-ipconnect.de) Quit ("Konversation terminated!")
  • [05:14:27] <L84Supper> anyone using QT4 for development on beagleboard?
  • [05:19:05] * dirk2 (n=dirk@p5B040F89.dip0.t-ipconnect.de) has joined #beagle
  • [05:20:34] * koen__ (n=koen@188.90.70.231) has joined #beagle
  • [05:20:46] <koen__> good morning all
  • [05:21:02] <_av500_> hi koen__
  • [05:21:10] <sakoman_> good morning koen__
  • [05:21:23] <sakoman_> are you out of the hospital yet?
  • [05:21:53] <koen__> should get out later today
  • [05:22:05] <sakoman_> how are you feeling?
  • [05:22:31] <sakoman_> sorry to hear that this happened to you!
  • [05:22:59] <koen__> a lot better than yesterday morning
  • [05:23:23] <sakoman_> do you know how you caught this virus?
  • [05:23:26] <koen__> sakoman_: it's viral and should be over very soon
  • [05:24:03] <koen__> nope, and the docs don't know which bug it is
  • [05:24:15] <sakoman_> pretty scarey!
  • [05:24:23] <koen__> yeah
  • [05:24:59] <koen__> but i haven't been so rested in months
  • [05:25:02] <sakoman_> you're too young for that kind of thing!
  • [05:25:42] <koen__> the other young person on this ward is 55
  • [05:25:43] * _av500_ agrees
  • [05:27:34] <koen__> i wonder if ffmpeg can read the xray movie they gave me
  • [05:27:51] <_av500_> hehe
  • [05:27:52] <sakoman_> :-)
  • [05:28:01] <sakoman_> time for you to get out of there!
  • [05:28:18] <_av500_> upload to mplayer samples and it will
  • [05:28:22] <koen__> indeed
  • [05:28:49] <koen__> the doc Will do rounds in about 2 hours from now
  • [05:29:24] <_av500_> at least they let you go on a weekend
  • [05:29:39] <_av500_> used to be here that nothing happened before monday morning...
  • [05:30:23] <koen__> sakoman_: i broke my angioseal yesterday, your wife Will prolly know why that sucks
  • [05:31:18] <sakoman_> I'll ask her when she wakes up :-) It sure doesn't sound good!
  • [05:32:47] <koen__> these smartphones help a lot with boredom
  • [05:33:18] <koen__> i suspect that an android phone would try to mate with the ECG machine
  • [05:33:22] <_av500_> if they let you multitask :-)
  • [05:33:59] <koen__> :)
  • [05:34:52] <_av500_> i have to ask the connectbot author to make urls in the ssh session clickable...
  • [05:35:00] <sakoman_> koen__: I have ereader on my iPhone and always have 5 or 6 books with me
  • [05:35:36] <koen__> i should try ebooks again
  • [05:36:06] <sakoman_> I got addicted to reading ebooks about 8 years ago when I travelled a lot
  • [05:36:26] <sakoman_> the selection is really quite good now
  • [05:36:43] <koen__> i just buy a lot of books :)
  • [05:36:45] <sakoman_> I have a hard time reading paper books now!
  • [05:37:12] <sakoman_> after 8 years I think I have like 500 ebooks
  • [05:37:20] <sakoman_> so I buy a lot of books too :-)
  • [05:37:35] <koen__> i was thinking about one of those cheapy-cheapy e-ink devices
  • [05:37:44] <sakoman_> iphone works great
  • [05:37:50] <_av500_> koen__: dont
  • [05:37:54] <sakoman_> and it is always with you
  • [05:38:08] <Pavlov> the kindle is the best thing since sliced bread
  • [05:38:14] <koen__> you have a point there
  • [05:38:28] <_av500_> and still the page flippin takes ages
  • [05:38:30] <sakoman_> I got my wife addicted back then too
  • [05:39:11] <sakoman_> fortunately we have similar taste in books so we share
  • [05:39:12] <Pavlov> makes traveling so much nicer
  • [05:40:30] <_av500_> sakoman_: what is your local time?
  • [05:40:39] <sakoman_> 10:40 pm
  • [05:40:56] <_av500_> west coast?
  • [05:41:06] <sakoman_> yes, far northern CA
  • [05:41:15] <_av500_> thx
  • [05:41:18] <koen__> 740 over here
  • [05:42:51] <koen__> sakoman_: i worked for a few years in a supermarkets which played a cd in a loop with "never rains in socal" on it
  • [05:43:35] <sakoman_> heh, I've heard that song more times that I care too :-)
  • [05:43:40] <koen__> hearing 8 times a day marks you for life ;)
  • [05:43:47] <sakoman_> I'm glad I don't live there!
  • [05:44:07] <sakoman_> this area isn't really CA :-)
  • [05:44:45] <koen__> i might be visiting St. Barbara after the summer
  • [05:44:50] <sakoman_> people say they are from "the north state"
  • [05:45:02] <sakoman_> that's a nice area to visit
  • [05:46:25] <koen__> 15 years ago it looked nice
  • [05:46:39] <sakoman_> it's about an 8 hour drive from here
  • [05:46:42] <koen__> haven't been there since
  • [05:46:59] <sakoman_> been about that long since I've been there too
  • [05:47:29] <koen__> don't you have a solar powered jet?
  • [05:48:19] <sakoman_> no, but I put in solar hot water heating this week :-)
  • [05:48:56] <sakoman_> it works amazingly well
  • [05:48:57] <_av500_> hot tub beats jet
  • [05:49:40] <koen__> are your regular panels photo-voltaic or using a stirling engine?
  • [05:50:42] <sakoman_> PV
  • [05:51:12] <sakoman_> nothing exotic, just kyocera panles
  • [05:51:32] <koen__> with batteries?
  • [05:52:15] <koen__> sakoman_: you know that kyocera is yashica after ?? takeover/merger?
  • [05:53:14] <sakoman_> yeah, with batteries. I haven't followed the kyocera/yashica thing. interesting!
  • [05:56:02] <koen__> does the i2c on the expansion connector need level conversion?
  • [05:56:24] <sakoman_> normally, yes. it is 1.8V
  • [05:56:30] <koen__> ok
  • [05:56:52] <sakoman_> (for both beagle and overo)
  • [05:57:19] <koen__> btw, specialcomp is sending me a prototype palo43 case
  • [05:58:32] <sakoman_> I just got one today
  • [05:58:42] <sakoman_> cute! haven't instaled the board yet
  • [05:59:11] <koen__> that reminds me, i need to steal the pandora pwm code
  • [05:59:31] <koen__> so i can battery power it
  • [05:59:59] <sakoman_> do they have a real driver or just a quick hack?
  • [06:00:31] <koen__> haven't looked at it that closely
  • [06:03:05] <koen__> av500: that c67x link you posted was a Nice read
  • [06:03:20] <sakoman_> time for sleep. take care koen! hope you escape soon :-)
  • [06:03:40] <koen__> thanks
  • [06:03:43] <koen__> good might
  • [06:04:22] <koen__> stoopid autocorrection
  • [06:04:55] <sakoman_> good night :-)
  • [06:13:23] <koen__> later all
  • [06:13:24] * koen__ (n=koen@188.90.70.231) Quit ("Get Colloquy for iPhone! http://mobile.colloquy.info")
  • [06:31:15] * koen___ (n=koen@188.89.152.227) has joined #beagle
  • [06:31:28] <koen___> re
  • [06:32:30] <koen___> hmmmm, my tail is getting longer and longer
  • [06:33:05] <koen___> time to reconnect
  • [06:33:06] * koen___ (n=koen@188.89.152.227) Quit (Client Quit)
  • [06:35:33] * KosiNuss (n=tom@R3e1a.r.pppool.de) has joined #beagle
  • [06:36:02] <_av500_> no ssh client?
  • [06:44:28] * dirk2 (n=dirk@p5B040F89.dip0.t-ipconnect.de) Quit ("Konversation terminated!")
  • [06:57:42] * Xenion (n=robert@p579FC340.dip.t-dialin.net) has joined #beagle
  • [07:00:29] * koen___ (n=koen@95.99.232.193) has joined #beagle
  • [07:01:11] <koen___> re
  • [07:01:41] <koen___> xv support in omapfbplay, nice
  • [07:04:03] <_av500_> koen___: no ssh client?
  • [07:04:40] <koen___> yes, but the osk is painfull
  • [07:04:58] <_av500_> osk?
  • [07:04:59] * ArteK (n=Artur@81.15.241.96) has joined #beagle
  • [07:05:03] <_av500_> kbd
  • [07:05:14] <koen___> yes
  • [07:05:18] <koen___> on screen KBC
  • [07:05:37] <koen___> kbd
  • [07:05:54] * valhalla (n=valhalla@81-174-24-42.dynamic.ngi.it) has joined #beagle
  • [07:05:57] <koen___> not the 5912 ;)
  • [07:10:58] * Xenion (n=robert@p579FC340.dip.t-dialin.net) Quit ("Verlassend")
  • [07:15:33] * koen___ (n=koen@95.99.232.193) Quit ("Get Colloquy for iPhone! http://mobile.colloquy.info")
  • [07:20:58] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) has joined #beagle
  • [07:24:29] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) Quit (Read error: 104 (Connection reset by peer))
  • [07:26:05] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) has joined #beagle
  • [07:29:23] * jipi (n=jipi@bb116-14-207-189.singnet.com.sg) has joined #beagle
  • [07:36:55] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) Quit (Read error: 104 (Connection reset by peer))
  • [07:38:16] * ThomasEgi (n=thomas@pppdyn-03.stud-ko.rz-online.net) has joined #beagle
  • [07:47:28] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) has joined #beagle
  • [08:02:19] <mru> morning
  • [08:05:30] * felipec (n=felipec@a91-153-253-80.elisa-laajakaista.fi) Quit ("Leaving")
  • [08:08:46] <_av500_> gm
  • [08:10:57] * zedstar (n=john@fsf/member/zedstar) has joined #beagle
  • [08:12:40] * Animule (n=Animal@70-56-35-197.hlrn.qwest.net) Quit (Read error: 110 (Connection timed out))
  • [08:19:56] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) Quit ("leaving")
  • [08:24:38] * oespirit is now known as oespirit|away
  • [08:31:55] * koen__ (n=koen@s55917625.adsl.wanadoo.nl) has joined #beagle
  • [08:33:05] * felipec (n=felipec@a91-153-253-80.elisa-laajakaista.fi) has joined #beagle
  • [08:42:11] * robclark_ (n=robclark@adsl-76-255-67-10.dsl.rcsntx.sbcglobal.net) has joined #beagle
  • [08:43:08] <mru> has anyone tried omapfbplay w/ xv output on beagle?
  • [08:46:24] * koen (n=koen@s55917625.adsl.wanadoo.nl) Quit (Read error: 110 (Connection timed out))
  • [08:54:51] <_av500_> this xv put frame looks expensive...
  • [08:56:43] <mru> in the X driver?
  • [08:59:21] <_av500_> ah no, you changed that
  • [08:59:25] * robclark (n=robclark@ppp-70-253-147-142.dsl.rcsntx.swbell.net) Quit (Read error: 110 (Connection timed out))
  • [08:59:30] <mru> oh, that
  • [08:59:38] <_av500_> you now decode into the xv buffersm right?
  • [08:59:40] <mru> that was just temporary
  • [08:59:54] <mru> yes, it's decoding directly into shared buffers
  • [09:00:28] <_av500_> so xv does planar to 422?
  • [09:00:48] <mru> on beagle?
  • [09:00:53] <_av500_> yes
  • [09:00:56] <mru> no idea, no X on my beagle
  • [09:01:10] <_av500_> :-)
  • [09:01:39] * oespirit|away is now known as oespirit
  • [09:02:13] <mru> I added xv support so I can run it on a pc
  • [09:03:30] <_av500_> ah ok
  • [09:04:12] <_av500_> what about tcvp?
  • [09:04:37] <mru> what about it?
  • [09:04:42] <_av500_> on beagle
  • [09:05:02] <mru> some bizarre libs it uses need porting
  • [09:05:02] <_av500_> i gather this is your player
  • [09:05:09] <_av500_> libtc
  • [09:05:11] <mru> yeah, I wrote those bizarre libs
  • [09:05:42] <_av500_> didnt look unportable to me
  • [09:05:53] <mru> it's not unportable
  • [09:06:01] <mru> but it's not just a matter of recompiling
  • [09:06:41] <mru> it was originally developed on Alpha
  • [09:07:51] <mru> the tricky part in libtc is the macros for constructing a va_list
  • [09:08:18] <_av500_> which is not portable bz default
  • [09:08:29] <mru> it can't be done portably
  • [09:08:39] * BThompson1 (n=Bernie@69.7.114.75) has joined #beagle
  • [09:08:56] <_av500_> zes
  • [09:08:58] <_av500_> yes
  • [09:09:03] <_av500_> damn french kbd
  • [09:09:18] <_av500_> id better paint all keys white...
  • [09:09:32] * mru has an all-black keyboard
  • [09:09:49] * _av500_ knows
  • [09:10:58] <mru> but any keyboard can be used with any mapping
  • [09:11:42] <_av500_> i still have to learn the frogpad...
  • [09:11:48] <_av500_> one day...
  • [09:12:07] <mru> is that what you call french keyboards?
  • [09:12:24] <_av500_> hehe no
  • [09:12:31] * Xenion (n=robert@ip-90-187-43-73.web.vodafone.de) has joined #beagle
  • [09:12:39] <_av500_> http://www.frogpad.com/
  • [09:12:43] <mru> would have been a fitting name
  • [09:12:51] * Xenion (n=robert@ip-90-187-43-73.web.vodafone.de) has left #beagle
  • [09:12:54] <_av500_> yes
  • [09:13:08] <mru> no offence to the french, but their keyboards are *weird*
  • [09:13:20] <_av500_> yes, numbers with shift....
  • [09:14:43] * ogra (n=ogra@ubuntu/member/ogra) Quit (Read error: 104 (Connection reset by peer))
  • [09:15:12] * ogra (n=ogra@ubuntu/member/ogra) has joined #beagle
  • [09:18:24] * jipi (n=jipi@bb116-14-207-189.singnet.com.sg) Quit ("Leaving")
  • [09:21:43] * BThompson (n=Bernie@69.7.114.75) Quit (Read error: 110 (Connection timed out))
  • [09:32:20] * oespirit is now known as rkirti|zZz
  • [09:43:45] * koen (n=koen@95.99.232.59) has joined #beagle
  • [09:46:11] <koen> so, any exciting stuff happen today?
  • [09:47:12] * Xenion (n=robert@ip-90-186-187-83.web.vodafone.de) has joined #beagle
  • [09:47:46] * koen is bored
  • [09:50:54] <_av500_> xvplay
  • [09:51:04] <_av500_> try it and report
  • [09:51:38] <koen> i already noticed xvplay :)
  • [09:53:19] <koen> i'll be here till at least monday, but my parents Will pickup the pma430
  • [09:53:39] <koen> what can the 430 handle?
  • [09:53:57] <_av500_> mpeg4 + mp3 in avi
  • [09:54:11] <_av500_> vga
  • [09:54:37] <koen> those 350mb per episode internet backups?
  • [09:55:15] <_av500_> yep
  • [09:56:08] <koen> awesome
  • [09:56:52] <koen> now i just need to supress the gagreflex when using qte2 ;)
  • [09:58:33] <koen> it's running oabi software?
  • [10:01:48] <koen> mru: xvplay looks really exciting :)
  • [10:02:34] <koen> i wonder if they allow beagle in the hospital
  • [10:06:30] <ssvb> koen: it would be nice if somebody could try/comment: http://www.nabble.com/-PATCH-0-1--ARM%3A-NEON-optimized-implementation-of-memcpy.-td24328820.html
  • [10:07:30] <koen> ssvb: mail me and i'll try it when i get back home
  • [10:07:36] <mru> using neon for less than a cacheline is suicide
  • [10:07:47] <koen> any benchmarks to run?
  • [10:07:56] * Xenion (n=robert@ip-90-186-187-83.web.vodafone.de) Quit (Read error: 110 (Connection timed out))
  • [10:08:02] * Xenion (n=robert@ip-90-186-20-65.web.vodafone.de) has joined #beagle
  • [10:08:40] <ssvb> mru: yes, the crossover point between arm and neon parts needs to be carefully selected
  • [10:09:17] <mru> I would only use neon for full cachelines
  • [10:09:22] <koen> i wonder if they allow laptops in here
  • [10:09:29] <mru> anything else will kill A8 performance
  • [10:10:16] <ssvb> mru: a good benchmark is needed (preferably some real application using lots of memcpy calls for small blocks)
  • [10:10:38] <mru> small memcpy calls should ideally be inlined
  • [10:10:46] <mru> and long ones should be rare
  • [10:10:56] <mru> if memcpy is a bottleneck, the app is poorly designed
  • [10:11:38] <ssvb> mru: yes, but I tried to experiment with gcc a bit, and it generates calls to memcpy even for very small known at compile time block sizes
  • [10:11:46] <ssvb> mru: this does not look good
  • [10:11:49] <mru> I know
  • [10:11:54] <mru> file a bug report
  • [10:12:08] <mru> the time is better spent fixing gcc than optimising memcpy
  • [10:12:40] <koen> :)
  • [10:13:15] <mru> and the gains from fixing apps overusing memcpy are also greater than optimising memcpy
  • [10:14:05] <mru> ssvb: why are you using vld4/vst4?
  • [10:14:23] <ssvb> mru: in order to have 8 bit element size
  • [10:14:31] <mru> that's a bad idea
  • [10:14:54] <mru> e.g. line 68-69 would be better to use vld1.32 {d0[0]}
  • [10:14:59] * notaz (n=notaz@82-135-242-201.static.zebra.lt) Quit (Read error: 110 (Connection timed out))
  • [10:15:27] <mru> much faster
  • [10:15:32] <ssvb> mru: it works just as fast in practice
  • [10:15:34] <mru> vld4 takes at least 3 cycles
  • [10:15:41] <mru> vld1 can run in one
  • [10:15:51] * notaz (n=notaz@82-135-242-201.static.zebra.lt) has joined #beagle
  • [10:16:11] <mru> if you don't see a difference, you should find what's throttling it and fix that
  • [10:16:15] <mru> maybe use some pld
  • [10:16:29] <ssvb> I initially had this one, but then read ARM manual about strongly ordered/device memory
  • [10:16:44] <mru> you'd be a fool to use memcpy on that
  • [10:17:04] <mru> the C standard gives no guarantees for what memcpy might do
  • [10:17:30] <koen> i remember heated duscussions about so memory
  • [10:17:41] <ssvb> mru: what's wrong with pld? it's already used
  • [10:18:03] <mru> sorry, I missed it
  • [10:18:07] <koen> it involved rmk, so of course the sky was falling down
  • [10:18:28] <mru> SO is the big mallet
  • [10:18:41] <mru> it will always behave exactly like the code looks
  • [10:18:56] * Xenion (n=robert@ip-90-186-20-65.web.vodafone.de) Quit (Read error: 60 (Operation timed out))
  • [10:19:06] <mru> but it's slow as hell
  • [10:19:24] <mru> every load/store has to go all the way to L3
  • [10:20:06] <mru> device memory is usually correct for memory mapped registers and such
  • [10:20:18] <ssvb> mru: at least changing some vst2/vst4 may be really useful, but I need to check how it would work with big endian data (try with 'setend be' / 'setend le' around the function body)
  • [10:20:29] <mru> it should just work
  • [10:20:51] <mru> that's my reading of the manual at least
  • [10:20:52] <ssvb> ok
  • [10:21:21] <mru> a load followed by a matching store should always be a copy
  • [10:21:48] <ssvb> I mean vld4.8 + vst1.32 pair
  • [10:21:54] <mru> that won't work
  • [10:21:54] * Wiedi (n=wiedi@newton-air.w.fruky.net) has joined #beagle
  • [10:22:01] <mru> don't do that
  • [10:22:03] <ssvb> destination buffer is guaranteed to be aligned there
  • [10:22:10] <ssvb> but not the source
  • [10:22:19] <mru> so use an unaligned vld1
  • [10:22:52] <mru> and if can choose, having the source unaligned is much better than unaligned dest
  • [10:22:59] <mru> of course both aligned is best
  • [10:22:59] <ssvb> in practice vld4 vs. vld1 does not make any visible difference and vld4 is just a little bit safer
  • [10:23:11] <mru> it's not safer at all
  • [10:23:25] <mru> the end result is exactly the same
  • [10:24:01] <ssvb> except for the theoretical case of using strongly ordered memory :-)
  • [10:24:12] <mru> but that's a bug in the first place
  • [10:24:35] <mru> anything mapped as SO probably has specific requirements on access size anyway
  • [10:24:42] <mru> and you can't possibly know what the correct one is
  • [10:24:48] <ssvb> mru: not really, it could be considered a platform specific extension
  • [10:24:50] <ssvb> :)
  • [10:25:25] <ssvb> mru: speaking about standards, C standard also does not cover things like threads, etc.
  • [10:25:25] <mru> I don't understand
  • [10:25:31] <ssvb> what exactly?
  • [10:25:34] <mru> no, but posix does
  • [10:25:42] <mru> and memcpy *is* C
  • [10:26:03] <mru> if you use memcpy, you have to be prepared for any of the allowed implementations
  • [10:26:44] <mru> about SO memory, memmapped registers typically must be accessed as their native size
  • [10:26:55] <mru> you can't read a 32-bit reg as 4 byte accesses
  • [10:27:06] <mru> so in that case your code would fail
  • [10:27:26] <mru> otoh, if the correct access size is 8 or 16 bits, using a 32-bit load will fail
  • [10:27:54] <mru> so don't bother trying to do the right thing when there is no way to know
  • [10:28:03] <mru> just make it as fast as possible for normal memory
  • [10:28:13] <mru> let people shoot themselves in the foot if they like
  • [10:28:43] <ssvb> mru: another pitfall is if somebody for whatever reason will enable alignment exceptions (A bit), of course this is also a very theoretical issue
  • [10:29:06] <mru> the A bit doesn't affect neon
  • [10:29:12] <ssvb> are you sure?
  • [10:29:36] <mru> no
  • [10:29:51] <mru> it does
  • [10:29:52] <mru> my bad
  • [10:30:05] <mru> anyway, you'd be stupid to run with that set
  • [10:30:13] <ssvb> agreed :)
  • [10:30:21] <mru> stupid or rmk...
  • [10:31:47] <mru> one more thing: it's good to put at least one instruction between a flag-setting instruction and a conditional branch
  • [10:31:50] <mru> saves one cycle
  • [10:32:51] <mru> your code looks arm-bound actually, which is bad
  • [10:33:19] <mru> if you can make the inner loop copy more in each iteration, the branches become essentially free
  • [10:33:40] <ssvb> mru: hmm, I thought that handling of conditional branches is shifted one stage later in the pipeline, so that setting a condition and branching can be executed simultaneously
  • [10:33:57] <mru> that's not what the cycle counter says
  • [10:34:03] <ssvb> mru: you are free to tweak the code, let me know if some of the changes can make it faster
  • [10:34:59] <ssvb> I'll try some of the changes too
  • [10:35:01] <ssvb> thanks
  • [10:35:10] <mru> and as I said, try writing only full cachelines with neon
  • [10:35:41] <mru> mixing neon and arm accesses within a cacheline, and especially within a 16-byte block, hurts
  • [10:36:06] <mru> and under no circumstances do an unaligned neon store spanning a 16-byte boundary
  • [10:36:38] <_av500_> koen: sorry bout qte2 :-)
  • [10:36:42] <ssvb> mru: yes, it's *very* visible in benchmarks :)
  • [10:36:50] <ssvb> really hard not to notice
  • [10:37:07] <mru> I guess you've seen my blog post on that topic
  • [10:37:11] <_av500_> trolltech gave qtopia to us and ony later admitted it was a joke
  • [10:37:12] <ssvb> yes
  • [10:37:21] <mru> the good news is that A9 is much better
  • [10:37:46] <mru> you never lose more than 4 cycles iirc
  • [10:37:51] <ssvb> of course I verified this information, I never trust anyone by default, even vendor docs :)
  • [10:38:00] <mru> this stuff isn't in the docs
  • [10:38:12] <mru> not the ones I see at least
  • [10:38:25] <ssvb> mru: by the way, have you experimented with write-allocate stuff?
  • [10:38:32] <mru> no
  • [10:38:55] <mru> there are a few situations where turning it off would help
  • [10:39:07] <mru> but for typical code it's beneficial
  • [10:40:34] <ssvb> mru: some of the info mentions that allocating cache line should be delayed so that fast filling memory with some data will not result in any reads
  • [10:40:54] <ssvb> mru: there are also some bits which look like they might affect this behavior
  • [10:41:23] <mru> no-write-allocate is good if you're doing lots of scattered writes and rarely read the same address again soon
  • [10:41:35] <mru> i.e. you'd have misses most of the time regardless
  • [10:41:51] <mru> in all other cases it's good
  • [10:44:09] <ssvb> mru: I got some weird result trying to compare contiguous write of data to memory (memset) and writing every second byte, the performance was about the same
  • [10:44:37] <ssvb> mru: I'm not sure if I did it right, but still I'm a bit suspicious about filling memory performance
  • [10:45:03] <mru> large blocks?
  • [10:45:15] * pH5 (n=ph5@p5485C2FC.dip.t-dialin.net) has joined #beagle
  • [10:45:34] <ssvb> mru: yes, memory throughput limited
  • [10:45:41] <mru> that explains it
  • [10:46:06] <mru> you still have to load/writeback the entire cachelines
  • [10:46:17] <ssvb> mru: for memset too?
  • [10:46:25] <mru> write-allocate
  • [10:46:45] <ssvb> mru: I was trying to check if this 'delayed allocation' works
  • [10:46:51] <mru> oh
  • [10:47:24] <ssvb> mru: I would assume that the data should be accumulated in the write buffer, and if it covers full cache line, there should be no read
  • [10:47:26] <mru> well, writing every second byte is probably no different to the memory system than writing all of them
  • [10:47:31] <mru> try words instead
  • [10:47:44] <mru> the write buffer is 16 bytes
  • [10:47:57] <mru> cache line is 64 bytes
  • [10:47:57] <ssvb> hmm, this seems to be too small
  • [10:49:05] <mru> nevertheless, that's what it is
  • [10:49:23] <mru> hence the 16-byte block quirks with neon/arm mixed accesses
  • [10:49:41] <ssvb> maybe it is split into 16-byte parts?
  • [10:51:38] <ssvb> writing 16-byte aligned blocks was also very beneficial for performance on OMAP1 and OMAP2, memset performance could be doubled this way
  • [10:51:55] <mru> that's not surprising
  • [10:51:58] <ssvb> I had a feature request in maemo bugzilla in the old 770 days: https://bugs.maemo.org/show_bug.cgi?id=733
  • [10:52:03] * Christos (n=chatzill@ppp-94-66-21-252.home.otenet.gr) has joined #beagle
  • [10:52:54] <ssvb> though I was very new to embedded development and core system components development those days, I should have tried to contact glibs-ports instead :)
  • [10:53:36] <ssvb> and my interpretation of TRMs was a bit off, so I posted lots of crap there :)
  • [10:53:45] <mru> ;-)
  • [10:53:51] <mru> we all have to start somewhere
  • [10:53:57] * rkirti|zZz is now known as rkirti
  • [10:56:18] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) has joined #beagle
  • [11:03:58] * oespirit (n=oespirit@117.254.9.34) has joined #beagle
  • [11:04:11] * Christos is now known as christoS
  • [11:04:28] * christoS is now known as chriStoS
  • [11:04:52] * chriStoS is now known as ChRiStOs
  • [11:05:20] * ChRiStOs is now known as cHrIsToS
  • [11:06:13] * cHrIsToS is now known as Christos_N
  • [11:09:10] * Xenion (n=robert@ip-77-24-170-119.web.vodafone.de) has joined #beagle
  • [11:09:11] <ssvb> mru: tried to search gcc bugzilla
  • [11:09:33] <ssvb> mru: LOL - http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29174
  • [11:10:10] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) Quit ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
  • [11:10:55] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) has joined #beagle
  • [11:12:37] <ssvb> mru: IMHO if the user did not insert a call to memcpy explicitly, he probably had reasons to do so (maybe the number of iterations is known to be very small)
  • [11:15:49] <mru> indeed
  • [11:16:00] <mru> that loop specicially is not equivalent to memcpy
  • [11:16:11] <mru> it has defined semantics on overlapping source and dest
  • [11:16:27] <mru> oh wait, it uses restrict
  • [11:16:36] <mru> then memcpy should be safe
  • [11:16:41] <mru> from that POV
  • [11:18:58] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) Quit ("ChatZilla 0.9.85 [Firefox 3.0.11/2009060215]")
  • [11:19:21] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) has joined #beagle
  • [11:21:23] * rkirti (n=oespirit@117.254.13.113) Quit (Read error: 110 (Connection timed out))
  • [11:21:25] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) Quit (Client Quit)
  • [11:21:43] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) has joined #beagle
  • [11:23:54] * Christos_N (n=chatzill@ppp-94-66-21-252.home.otenet.gr) Quit (Client Quit)
  • [11:24:12] * Christos_N (n=Christos@ppp-94-66-21-252.home.otenet.gr) has joined #beagle
  • [11:43:42] * Xenion (n=robert@ip-77-24-170-119.web.vodafone.de) Quit (Read error: 110 (Connection timed out))
  • [11:43:56] * Xenion (n=robert@ip-90-187-214-46.web.vodafone.de) has joined #beagle
  • [11:44:10] * booxter (n=booxter@241-243-252-87-dynamic-pool.gprs.mts.by) Quit (Read error: 110 (Connection timed out))
  • [11:44:36] * magnet is now known as Guest31411
  • [11:47:39] * javaJake (n=javaJake@unaffiliated/javajake) has joined #beagle
  • [11:49:35] * koen (n=koen@95.99.232.59) Quit (Read error: 60 (Operation timed out))
  • [11:53:45] * Christos_N (n=Christos@ppp-94-66-21-252.home.otenet.gr) Quit (Read error: 60 (Operation timed out))
  • [11:58:36] <ssvb> mru: tried to benchmark conditional branches: http://pastebin.ca/1483772
  • [12:00:46] <ssvb> mru: how did you do these measurements? ("<mru>: that's not what the cycle counter says")
  • [12:01:24] <mru> http://unicorn.mansr.com/~mru/bp.S
  • [12:02:18] <mru> how are your gnu as skills?
  • [12:02:40] * robclark (n=robclark@ppp-70-253-39-72.dsl.rcsntx.swbell.net) has joined #beagle
  • [12:04:06] <ssvb> mru: should be ok, when also peeking into 'info as' periodically
  • [12:05:20] <ssvb> mru: why are you asking?
  • [12:05:26] <mru> hmm, I ran the test again, and now it doesn't seem to make any difference where the flags are set
  • [12:05:47] <mru> because that file I linked is hard to understand otherwise
  • [12:06:55] <ssvb> mru: I'm fine with any macros, had to use lots of different assemblers
  • [12:07:50] * robclark_ (n=robclark@adsl-76-255-67-10.dsl.rcsntx.sbcglobal.net) Quit (Read error: 60 (Operation timed out))
  • [12:08:01] <ssvb> mru: ok, I thought that you might have checked the code in the cycle accurate simulator and it reported problems with instructions pairing
  • [12:08:14] <mru> I don't trust simulators
  • [12:08:36] <ssvb> I haven't used the simulator, and don't have one yet
  • [12:08:37] <mru> unless they're using the actual vhdl or whatever
  • [12:08:56] <mru> I only have whatever comes with rvds
  • [12:10:45] <ssvb> with ARM11 it's pretty easy to get away without using any kind of simulator
  • [12:11:04] <ssvb> it has data dependency stall performance monitoring unit event
  • [12:12:27] <ssvb> so one can just insert kind of 'nop' instructions and get rid of these events, then think about how to rearrange the code in order not to have neither 'nops' nor these events in the code
  • [12:12:32] <ssvb> was pretty simple
  • [12:13:39] <ssvb> with Cortex-A8 I did not find a 'silver bullet' yet :) so getting the simulator may be really needed
  • [12:14:09] <mru> there are various pmon events you can use
  • [12:14:26] <mru> keep messing with the code until it doesn't get faster
  • [12:15:48] * Animule (n=Animal@71.33.160.80) has joined #beagle
  • [12:32:22] * oespirit (n=oespirit@117.254.9.34) Quit (Remote closed the connection)
  • [12:33:50] * docelic__ (n=docelic@78.134.196.188) has joined #beagle
  • [12:35:40] * docelic_ (n=docelic@78.134.197.40) Quit (Read error: 60 (Operation timed out))
  • [13:07:58] * Xenion (n=robert@ip-90-187-214-46.web.vodafone.de) Quit (Read error: 110 (Connection timed out))
  • [13:09:58] * lifeeth (n=praneeth@unaffiliated/lifeeth) has joined #beagle
  • [13:10:11] * zman (n=Unknown@ppp118-208-43-24.lns3.bne1.internode.on.net) has joined #beagle
  • [13:10:39] * zman is now known as Guest25061
  • [13:11:03] <ssvb> mru: about that NEON memcpy, inner loop has 8 instructions (so takes 4 cycles to decode), NEON theoretically takes 3+2=5 cycles, ARM code is well pairable and should be faster than 5 cycles, reordering ARM instructions does not change anything for L1 cache test
  • [13:11:32] <ssvb> mru: so theoretically NEON is the bottleneck, but unrolling loop really helps a bit :)
  • [13:12:35] <ssvb> mru: in any case, performance is a bit far from 128-bit per cycle, so there must be something else
  • [13:13:13] <mru> try preloading further ahead
  • [13:13:17] <ssvb> probably cpu can't sustain 128-bit per cycle rate continuously
  • [13:14:23] <mru> try adding various multiples of 64 bytes and see what's fastest
  • [13:14:57] <ssvb> mru: that a test for L1 cached data, and preloading distance already seems to be optimal for large transfers
  • [13:15:01] <ssvb> testing that
  • [13:15:30] <mru> I've found the optimal distance generally hard to predict
  • [13:15:44] <mru> and in general the A8 memory system is much suckier than the manual lets on
  • [13:16:02] * hochopeper (n=Unknown@ppp118-208-39-18.lns3.bne1.internode.on.net) Quit (Read error: 60 (Operation timed out))
  • [13:16:44] * s27 (i=7aa60de8@gateway/web/freenode/x-d81442a76a4c4687) has joined #beagle
  • [13:17:31] <ssvb> mru: I suspect that LSU also has its own queue, could overflow and fail to handle 128-bit per cycle
  • [13:17:58] <mru> there is definitely a load/store queue
  • [13:18:09] <mru> remember that erratum about it leaking entries?
  • [13:18:34] <ssvb> mru: so short bursts are fine, but not continuous transfers
  • [13:19:06] <mru> ultimately you're obviously bound by dram and/or l3
  • [13:26:05] <mru> xvplay is no more, omapfbplay can have multiple output options instead
  • [13:27:14] <s27> i use tomi display frame work on a custom board. i get lots of sync_lost errors in the kernel
  • [13:27:42] <s27> i have enabled display in u-boot. this seems like an issue now
  • [13:33:17] * Guest25061 (n=Unknown@ppp118-208-43-24.lns3.bne1.internode.on.net) Quit ("Leaving")
  • [13:45:30] <_av500_> mru: go out get some sun
  • [13:46:17] <_av500_> mru: go out get some sun, errr oracle
  • [13:54:20] * rkirti (n=oespirit@117.254.9.215) has joined #beagle
  • [13:55:16] <tomba> s27: update to newer DSS2 version
  • [14:08:00] * tzhau_ (n=tzhau@dsl-245-76-118.telkomadsl.co.za) has joined #beagle
  • [14:08:31] * robclark_ (n=robclark@ppp-70-129-133-149.dsl.rcsntx.swbell.net) has joined #beagle
  • [14:08:44] <mru> _av500_: ;-)
  • [14:09:47] * robclark (n=robclark@ppp-70-253-39-72.dsl.rcsntx.swbell.net) Quit (Read error: 104 (Connection reset by peer))
  • [14:10:06] * tzhau (n=tzhau@dsl-245-76-118.telkomadsl.co.za) Quit (Read error: 110 (Connection timed out))
  • [14:11:41] * robclark (n=robclark@ppp-70-129-129-179.dsl.rcsntx.swbell.net) has joined #beagle
  • [14:13:52] * Xenion (n=robert@ip-90-186-254-228.web.vodafone.de) has joined #beagle
  • [14:14:59] * Xenion (n=robert@ip-90-186-254-228.web.vodafone.de) Quit (Client Quit)
  • [14:15:18] * rsalveti_ (n=rsalveti@189.70.35.140) has joined #beagle
  • [14:21:59] * greyback (n=greyback@ip-83-147-165-234.dub-3rk1.metro.digiweb.ie) has joined #beagle
  • [14:25:17] * robclark_ (n=robclark@ppp-70-129-133-149.dsl.rcsntx.swbell.net) Quit (Read error: 110 (Connection timed out))
  • [14:25:25] * javaJake (n=javaJake@unaffiliated/javajake) Quit (Read error: 110 (Connection timed out))
  • [14:28:05] <s27> tomba: do i need to reset the dss in the kernel?
  • [14:28:22] <ssvb> mru: well, finally looks like relative placement of pld, vld1 and vst1 instructions affects L1 and L2 cached block copies in a somewhat erratic way, so a good timing is needed for feeding LSU unit with requests and LSU is the bottleneck here. normal memory performance is more or less consistent
  • [14:28:44] * ArteK (n=Artur@81.15.241.96) has left #beagle
  • [14:28:59] <mru> remember that the neon pipeline is 16 instructions behind the arm pipeline
  • [14:29:11] <mru> if the neon instruction fifo is full
  • [14:29:17] <ssvb> mru: so it's just a matter of finding best permutation, and pray that it will remain good at other cpu clock frequencies :)
  • [14:29:32] <mru> you might want to test it on a9 too
  • [14:29:45] <mru> do you have an a9?
  • [14:30:11] <ssvb> no
  • [14:30:36] <mru> if you make a statically linked executable I can test it for you
  • [14:48:24] * greyback (n=greyback@ip-83-147-165-234.dub-3rk1.metro.digiweb.ie) Quit ()
  • [14:55:41] * rkirti (n=oespirit@117.254.9.215) Quit (Remote closed the connection)
  • [14:57:09] <mru> anyone care to test omapfbplay with xv output on a beagle?
  • [15:10:38] * ds2 (i=noinf@netblock-66-245-251-24.dslextreme.com) Quit (Read error: 110 (Connection timed out))
  • [15:10:57] * javaJake (n=javaJake@unaffiliated/javajake) has joined #beagle
  • [15:11:25] * felipec (n=felipec@a91-153-253-80.elisa-laajakaista.fi) Quit (Remote closed the connection)
  • [15:28:55] * djlewis (n=djlewis@75.15.65.232) has joined #beagle
  • [15:29:14] <djlewis> gm mru
  • [15:29:27] <javaJake> Morning djlewis ;)
  • [15:29:45] <djlewis> Hey!, how the heck are you javaJake?
  • [15:29:58] <javaJake> Good
  • [15:30:06] <javaJake> Relaxing before some afternoon fun. :)
  • [15:30:15] <javaJake> What's on your TODO list for the next week?
  • [15:30:33] <djlewis> yep, I have female company coming.. had to pick up a few things..
  • [15:30:59] * javaJake will not comment
  • [15:31:32] <djlewis> the next week.... get up, get out of bed, run a come across my head, sip some morning coffee and head out to do some computer and network repairs..
  • [15:32:25] <javaJake> Huh. Nothing to do with #beagle eh? :)
  • [15:32:33] * rkirti (n=oespirit@117.254.12.176) has joined #beagle
  • [15:33:11] <djlewis> well, mru put out a call for testors and i'll volunteer for that.
  • [15:33:45] * greyback (n=greyback@dhcp-892b7be8.ucd.ie) has joined #beagle
  • [15:34:04] <djlewis> Last night I was going to try some things on BB but I tried first on mt wkstn and made no progess, where it should have easily worked :P
  • [15:34:17] * kozak (n=subbu@122.166.48.72) has joined #beagle
  • [15:34:28] <djlewis> :-@ == better face for my last line.
  • [15:35:12] * tegila (n=tegila@189-015-135-057.xd-dynamic.ctbcnetsuper.com.br) has joined #beagle
  • [15:37:28] <djlewis> javaJake: I have had my BB for going on 4 months now. Kinda trying to catch my next wind on it..
  • [15:42:01] <javaJake> Oh, OK :)
  • [15:42:23] <javaJake> Sorry, cat distracted me there for a moment
  • [15:45:04] * rkirti (n=oespirit@117.254.12.176) Quit (Remote closed the connection)
  • [15:45:25] <djlewis> javaJake: company has arrived, gotta go for now dude...
  • [15:45:39] <javaJake> Just going to say similar things myself
  • [15:45:41] <javaJake> Catch you later
  • [15:46:00] * djlewis (n=djlewis@75.15.65.232) Quit ("Kopete 0.12.4 : http://kopete.kde.org")
  • [15:46:03] * greyback (n=greyback@dhcp-892b7be8.ucd.ie) Quit ()
  • [15:46:59] * rkirti (n=oespirit@117.254.9.233) has joined #beagle
  • [15:47:49] * javaJake (n=javaJake@unaffiliated/javajake) Quit ("gone for the afternoon")
  • [16:00:44] * valhalla (n=valhalla@81-174-24-42.dynamic.ngi.it) Quit ("Leaving")
  • [16:04:23] * BThompson1 (n=Bernie@69.7.114.75) Quit (Read error: 110 (Connection timed out))
  • [16:19:26] * risal_xyz (n=risal@155.69.162.56) has joined #beagle
  • [16:20:05] <risal_xyz> hai.... my usb0 net on the beagleboard doesnt show up on the linux host .... how do I enable the usbnet feature?
  • [16:20:52] <jkridner|work> you compile it into the kernel or load the kernel module.
  • [16:21:08] <jkridner|work> for it to show up on the host...
  • [16:21:15] <jkridner|work> you have to enable it on the beagleboard first.
  • [16:22:06] <risal_xyz> i am using the koen's latest beagle demo image...
  • [16:22:25] <jkridner|work> i think his latest has a composite driver by default, not usbnet.
  • [16:22:35] <risal_xyz> what is e_ether.ko and e_cdc.ko ....?? i have been searching for about 2 days on this and no lead .... till now
  • [16:22:36] <jkridner|work> but, that should work fine with a linux host.
  • [16:23:03] <jkridner|work> what does 'ifconfig' show on your beagle?
  • [16:23:26] <risal_xyz> root@beagleboard:~# ifconfig
  • [16:23:26] <risal_xyz> lo Link encap:Local Loopback
  • [16:23:26] <risal_xyz> inet addr:127.0.0.1 Mask:255.0.0.0
  • [16:23:26] <risal_xyz> inet6 addr: ::1/128 Scope:Host
  • [16:23:26] <risal_xyz> UP LOOPBACK RUNNING MTU:16436 Metric:1
  • [16:23:26] <risal_xyz> RX packets:12 errors:0 dropped:0 overruns:0 frame:0
  • [16:23:28] <risal_xyz> TX packets:12 errors:0 dropped:0 overruns:0 carrier:0
  • [16:23:30] <risal_xyz> collisions:0 txqueuelen:0
  • [16:23:32] <risal_xyz> RX bytes:600 (600.0 b) TX bytes:600 (600.0 b)
  • [16:23:34] <risal_xyz>
  • [16:23:36] <risal_xyz> usb0 Link encap:Ethernet HWaddr C2:E0:04:8C:D1:22
  • [16:23:38] <risal_xyz> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
  • [16:23:40] <risal_xyz> RX packets:0 errors:0 dropped:0 overruns:0 frame:0
  • [16:23:42] <risal_xyz> TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
  • [16:23:44] <risal_xyz> collisions:0 txqueuelen:1000
  • [16:23:46] <risal_xyz> RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
  • [16:23:48] <risal_xyz> ooops...
  • [16:24:18] <risal_xyz> i do ifconfig on the host... and usb0 is not seen there....
  • [16:24:28] <jkridner|work> oh well. so, you need to do an ifconfig on the beagleboard first.
  • [16:24:32] * felipec (n=felipec@a91-153-253-80.elisa-laajakaista.fi) has joined #beagle
  • [16:24:39] <jkridner|work> if you have /etc/network already setup....
  • [16:24:47] <jkridner|work> then, you can do 'ifup usb0'.
  • [16:25:35] <risal_xyz> but on the beagle usb0 is already up right....
  • [16:25:45] <jkridner|work> it doesn't have an ip address yet.
  • [16:26:15] * greyback (n=greyback@ip-83-147-165-234.dub-3rk1.metro.digiweb.ie) has joined #beagle
  • [16:26:23] <risal_xyz> usb0 Link encap:Ethernet HWaddr C2:E0:04:8C:D1:22
  • [16:26:24] <risal_xyz> inet addr:192.168.2.15 Bcast:192.168.2.255 Mask:255.255.255.0
  • [16:26:24] <risal_xyz> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
  • [16:26:24] <risal_xyz> RX packets:0 errors:0 dropped:0 overruns:0 frame:0
  • [16:26:24] <risal_xyz> TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
  • [16:26:24] <risal_xyz> collisions:0 txqueuelen:1000
  • [16:26:26] <risal_xyz> RX bytes:0 (0.0 b) TX bytes:0 (0.0 b)
  • [16:26:31] <risal_xyz> yea done... now it has an ip address
  • [16:26:41] <jkridner|work> k. now do you see it on the host?
  • [16:26:47] <risal_xyz> nope.....
  • [16:26:53] <risal_xyz> i do ifconfig on host
  • [16:27:01] <risal_xyz> but usb0 is not displayed...
  • [16:27:17] <mru> you need to configure usb0 on both ends
  • [16:27:21] <jkridner|work> do you see other adapters?
  • [16:27:25] <mru> with different IP addresses in the same subnet
  • [16:28:00] <risal_xyz> on /etc/sysconfig/network-scripts/ifcfg-usb0 ,.......... i have configured like this
  • [16:28:15] <risal_xyz> DEVICE=usb0
  • [16:28:16] <risal_xyz> BOOTPROTO=static
  • [16:28:16] <risal_xyz> ONBOOT=no
  • [16:28:16] <risal_xyz> USERCTL=no
  • [16:28:16] <risal_xyz> IPV6INIT=no
  • [16:28:16] <risal_xyz> PEERDNS=yes
  • [16:28:18] <risal_xyz> TYPE=Ethernet
  • [16:28:20] <risal_xyz> NM_CONTROLLED=no
  • [16:28:22] <risal_xyz> IPADDR=192.168.2.14
  • [16:28:24] <risal_xyz> BROADCAST=192.168.2.255
  • [16:28:24] <mru> means nothing to me
  • [16:28:26] <risal_xyz> NETMASK=255.255.255.0
  • [16:28:28] <risal_xyz> NETWORK=192.168.2.0
  • [16:28:30] <risal_xyz> ~
  • [16:28:35] <mru> except that you're probably running a redhatish system
  • [16:28:40] <mru> which to me equals trouble
  • [16:28:40] <risal_xyz> yea
  • [16:28:45] <risal_xyz> hmm....
  • [16:29:02] <jkridner|work> probably want to stop pasting that much. just want to know if 'ifconfig' on the host shows other adapters.
  • [16:29:12] <mru> try ifconfig -a
  • [16:29:21] <risal_xyz> okk
  • [16:29:23] <mru> plain ifconfig doesn't list adapters that are "down"
  • [16:29:55] <jkridner|work> k
  • [16:30:00] <risal_xyz> it only shows eth0,irda0 and lo
  • [16:30:09] <risal_xyz> no usb0....
  • [16:30:24] <mru> which end?
  • [16:30:31] <risal_xyz> on the linux host
  • [16:30:35] <mru> lsusb?
  • [16:31:03] <risal_xyz> should i paste it?
  • [16:31:14] <mru> use pastebin.com
  • [16:31:18] <jkridner|work> long pastes should... yeah.
  • [16:32:18] <risal_xyz> http://pastebin.com/md7c0c6e
  • [16:32:36] <mru> well, it ain't there
  • [16:32:40] <risal_xyz> yea
  • [16:32:57] <mru> Bus 001 Device 035: ID 0525:a4a1 Netchip Technology, Inc. Linux-USB Ethernet Gadget
  • [16:33:05] <mru> that's what you should see
  • [16:33:14] <mru> bus and device numbers can differ of course
  • [16:33:31] <risal_xyz> i found many sources telling to enable g_ether ... i dont think i have it ... is it the problem??
  • [16:33:57] <mru> yeah, that would be a problem....
  • [16:34:42] <risal_xyz> but i do ...... modprobe g_ether on host and it gives...[root@ris_fedora8 network-scripts]# modprobe g_ether
  • [16:34:42] <risal_xyz> FATAL: Error inserting g_ether (/lib/modules/2.6.23.1-42.custom_ris_1.fc8/kernel/drivers/usb/gadget/g_ether.ko): No such device
  • [16:35:01] <mru> g_ether goes on the beagle
  • [16:35:31] <risal_xyz> in beagle it says........... FATAL: Module g_ether not found.
  • [16:35:52] <mru> do you have a usb0 device on the beagle?
  • [16:35:57] <risal_xyz> yea
  • [16:36:05] <mru> hmm
  • [16:36:36] <risal_xyz> http://pastebin.com/m8d99b66
  • [16:36:42] <mru> do you see something like this in the beagle kernel log? "g_ether gadget: high speed config #1: CDC Ethernet (ECM)"
  • [16:36:43] <risal_xyz> this is the ifconfig of beagleboard
  • [16:37:57] <risal_xyz> i see this..... http://pastebin.com/m2ed63523
  • [16:38:20] <risal_xyz> instead of g_ether... i get some messages with g_cdc...??
  • [16:38:42] <jkridner|work> that is the composite driver, i believe.
  • [16:38:51] <jkridner|work> you could try removing it and installing the ether driver.
  • [16:39:11] <risal_xyz> even then.... it should have come up in the linux host ... isnt it??
  • [16:39:22] <jkridner|work> it should have....
  • [16:39:49] <risal_xyz> i am powering the beagleboard through OTG port.. is it a problem..??
  • [16:39:54] <jkridner|work> was just wondering if enumerating as a usb ethernet would show up better than the composite driver.
  • [16:40:01] * jipi (n=jipi@bb116-14-207-189.singnet.com.sg) has joined #beagle
  • [16:40:02] <jkridner|work> should not be a problem.
  • [16:40:55] <risal_xyz> but how do i remove it and install g_ether...?? shoud i build the kernel myself....??
  • [16:41:14] <risal_xyz> i am a newbie in linux.... :)
  • [16:41:43] <jkridner|work> you can just do it live with 'rmmod' and 'insmod'.
  • [16:43:22] <risal_xyz> http://pastebin.com/mc9add5a
  • [16:43:43] <risal_xyz> unable to insert the modules...
  • [16:44:19] <jkridner|work> you'll need to go to the directory to where the .ko files live.
  • [16:44:41] <jkridner|work> lsmod
  • [16:45:23] <risal_xyz> lsmod http://pastebin.com/m55fffbf6
  • [16:46:06] * rkirti (n=oespirit@117.254.9.233) Quit (Remote closed the connection)
  • [16:46:37] <jkridner|work> k. go to where g_ether.ko is to do the insmod.
  • [16:46:50] <jkridner|work> looks like g_cdc isn't loaded.... could be compiled into the kernel.
  • [16:48:06] <risal_xyz> hmm... i see inside /lib/modules/2.6.29-omap1/kernel/drivers/usb .. but those .ko are nowhere to be found
  • [16:48:45] * Cru_N_cher (n=luls_lol@dslb-084-058-147-244.pools.arcor-ip.net) has joined #beagle
  • [16:49:38] <jkridner|work> cd /lib/modules; find . | grep g_ether
  • [16:49:56] <risal_xyz> root@beagleboard:/# find /lib/* -name "g_ether.ko" doesnt return any match
  • [16:50:37] <risal_xyz> .... no match...
  • [16:50:47] <risal_xyz> i think g_ether is not present....
  • [16:50:51] <risal_xyz> :(
  • [16:51:19] <jkridner|work> k. i thought you had said earlier that you had found the g_ether.ko file.
  • [16:51:47] <risal_xyz> no... i saw many articles about g_ether ... thats why i asked....
  • [16:51:57] <jkridner|work> ah, k.
  • [16:52:13] <jkridner|work> as far as you did get, I am surprised that 'lsusb' on the host showed nothing.
  • [16:52:56] <risal_xyz> i actually thought g_ether.ko was needed in the host and so i made a custom kernel with g_ether.ko..... it took me 3 hours to build the kernel..... :( .. its only now that i know that g_ether is needed in the beagleboard....
  • [16:53:30] <jkridner|work> ah. probably a standard kernel on the host would be just fine.
  • [16:53:53] <mru> jkridner|work: btw, did you hear that the TI compiler can't compile ffmpeg properly?
  • [16:54:30] <jkridner|work> i did not. you talking to someone on the TI compiler team so that we can fix it?
  • [16:54:36] <risal_xyz> and one more intriguing thing... in my windows system..earlier the "Found new wizard" used to pop up when the beagleboard was connected, but now its not showing anymore and so i cannot use the RNDIS featture...:(
  • [16:54:50] <mru> jkridner|work: roger forwarded them some test cases
  • [16:55:19] <mru> jkridner|work: most importantly, c99 variable length arrays are not supported
  • [16:55:22] <jkridner|work> in koen's newer kernels, RNDIS/Windows won't work because he's using the composite driver, not g_ether.
  • [16:55:31] <mru> it pretends to compile the code with the --gcc flag
  • [16:55:46] <risal_xyz> i see....
  • [16:56:03] <risal_xyz> but i can always make a custom kernel to be run in beagleboard .. right...??
  • [16:56:08] <risal_xyz> i will try that..... :)
  • [16:56:18] <jkridner|work> you could recompile the beagle kernel and switch from g_cdc to g_ether. that is becoming a FAQ.
  • [16:56:18] <mru> but the actual code doesn't allocate any space at all for the arrays...
  • [16:56:35] <mru> this happens on both arm and dsp
  • [16:56:48] <risal_xyz> okk.... thanks a lott..... :D
  • [16:56:50] <mru> same compiler frontend apparently
  • [16:56:55] <jkridner|work> I'd be inclined to use koen's defconfig and patches and change just that one bit.
  • [16:57:06] <jkridner|work> mru: as I would expect.
  • [16:57:29] <jkridner|work> mru: I expect them to share a lot of code.
  • [16:57:50] <mru> jkridner|work: speaking of dsp, where are things like function calling conventions documented?
  • [16:58:14] <jkridner|work> btw, we have an internal C6000 compiler built to run on the ARM. I'm thinking it could be really useful to build DSP/Link under the native build environment preferred by the Canonical-types.
  • [16:58:30] <mru> maybe for them
  • [16:58:36] <jkridner|work> mru: last I saw the compiler conventions, it was in "yellow books"...
  • [16:58:39] <mru> I couldn't care less
  • [16:58:45] <jkridner|work> but, I suspect it is simply in the compiler user's guide.
  • [16:58:56] <mru> ok, I'll look there
  • [16:59:59] <mru> hmm, what users guide?
  • [17:00:03] <mru> it didn't install one
  • [17:00:07] <jkridner|work> http://focus.ti.com/lit/ug/spru187o/spru187o.pdf
  • [17:00:23] <jkridner|work> Looks like it is in Chapter 7.
  • [17:00:40] <jkridner|work> has register conventions and 7.4 has calling conventions.
  • [17:00:49] <mru> thanks
  • [17:03:24] <mru> hmm, I can see opportunities for clever function argument ordering here...
  • [17:03:54] <jkridner|work> lots of fun things for compiler folks to do. :)
  • [17:04:13] <jkridner|work> sometimes, I think the architecture was done for the entertainment of the compiler writers. ;)
  • [17:04:19] <mru> hehe
  • [17:04:32] <mru> or crazy people like me...
  • [17:04:33] * rkirti (n=oespirit@117.254.12.241) has joined #beagle
  • [17:04:49] <jkridner|work> will you hand-schedule your assembly too?
  • [17:04:54] <jkridner|work> that is a really fun process.
  • [17:05:04] <mru> some of my arm assembler uses its own unique calling conventions
  • [17:05:06] <jkridner|work> the linear assembler is good.
  • [17:05:59] * KosiNuss (n=tom@R3e1a.r.pppool.de) Quit (Read error: 110 (Connection timed out))
  • [17:06:27] * risal_xyz (n=risal@155.69.162.56) has left #beagle
  • [17:06:31] <jkridner|work> mru: seeing you at about 6 minutes in on a LinuxTag2009 video. :)
  • [17:06:43] <mru> url?
  • [17:07:25] <jkridner> http://translate.google.com/translate?sl=auto&tl=en&u=http%3A%2F%2Fwww.linux-magazin.de%2FNEWS%2FVideo-Das-war-der-Linuxtag-2009
  • [17:07:32] <jkridner> sorry for the translation URL.
  • [17:10:33] <mru> ah yes, I remember that shoot
  • [17:20:57] * djlewis (n=djlewis@75.15.65.232) has joined #beagle
  • [17:21:06] <mru> I have two evil plans for omapfbplay: 1) use dsp for yuv conversion, and 2) sync playback over network for video wall
  • [17:21:45] <billenium> Could anyone do me a small favor on their beagleboard?
  • [17:21:51] <mru> what?
  • [17:22:17] <billenium> I need someone to try out a program that may or may not work on (ARM)/beagleboard.
  • [17:22:21] <billenium> http://ecere.com/
  • [17:22:46] <mru> hmm, not so small
  • [17:22:52] <billenium> Yeah i know :P
  • [17:23:04] <djlewis> mru: was that you in the middle giving the ffmpeg presentation?
  • [17:23:23] <mru> djlewis: I'm the one talking the most, yes
  • [17:23:58] <djlewis> cool.. I had seen you on a presentation linked from beagleboard.org from last year.
  • [17:24:15] <mru> yeah, the lugradio event probably
  • [17:24:27] <djlewis> mru: btw, i offered to test your app.
  • [17:24:35] * jipi (n=jipi@bb116-14-207-189.singnet.com.sg) Quit (Remote closed the connection)
  • [17:25:01] <djlewis> if it is something my hardware can do...
  • [17:25:25] <mru> djlewis: do you have X on your beagle?
  • [17:25:56] <djlewis> mru: i'm glad you spoke it in english. I am running sakoman's Angstrom desktop 2.6.30
  • [17:26:02] <djlewis> and 2.6.29
  • [17:26:12] <mru> djlewis: I speak english and swedish
  • [17:26:17] <mru> swedish isn't very useful
  • [17:26:40] <djlewis> but you type a real good american dialect ;)
  • [17:27:21] <djlewis> I can't hear your accent in irc...
  • [17:27:29] <mru> hehe
  • [17:29:07] <mru> do you have ffmpeg libs built for the beagle?
  • [17:29:16] * jkridner|work (n=a0321898@nat/ti/x-6cfc38cc5ab53400) has left #beagle
  • [17:29:31] <djlewis> ffmpeg runs under my mplayer so yes.
  • [17:29:41] <mru> good
  • [17:29:57] <mru> then you can git pull git://git.mansr.com/omapfbplay
  • [17:30:35] <djlewis> I dont GIT. is it something I can install?
  • [17:30:50] <mru> hmm, what do you then?
  • [17:31:01] <djlewis> build locally on B
  • [17:31:18] <mru> that's not related
  • [17:31:19] <djlewis> or install pre package
  • [17:31:23] <mru> how do you get source code?
  • [17:31:44] <djlewis> so to svn with linux wkstn
  • [17:31:46] <jkridner> you can 'git' source code from the board.
  • [17:32:22] <mru> you can grab a tar.gz from my gitweb at http://git.mansr.com/?p=omapfbplay;a=summary
  • [17:32:40] <mru> use the 'snapshot' link of the most recent revision
  • [17:32:42] <djlewis> I would have to opkg git too ?.
  • [17:33:03] <jkridner> that is a way to get both 'git' and 'omapfbplay'.
  • [17:33:24] <jkridner> pre-built binaries, that is.
  • [17:33:25] <mru> opkg doesn't have the version of omapfbplay I committed today
  • [17:33:56] <jkridner> so opkg 'git' then pull the sources and build cross or native.
  • [17:34:28] <mru> djlewis: just leave it if it's too much trouble
  • [17:34:47] <jkridner> what is new in omapfbplay today?
  • [17:34:48] <mru> koen__ is bound to try it out when he gets a chance
  • [17:34:58] <djlewis> I already have the tar, I'll give it a shot this aft.
  • [17:35:07] <mru> jkridner: omapfbplay has xvideo output now
  • [17:35:28] <jkridner> ah, nice.
  • [17:35:33] * jkridner|work (n=a0321898@nat/ti/x-6cfc38cc5ab53400) has joined #beagle
  • [17:35:38] <mru> djlewis: for a native compile: make OMAPFB=y XV=y
  • [17:35:52] <djlewis> unless it needs libs I dont have it should go well.
  • [17:36:03] <mru> if ffmpeg libs and headers are installed it should be fine
  • [17:36:15] <mru> and X libs for the X bits of course
  • [17:39:41] <djlewis> mru: are you testing with any particular input file type?
  • [17:39:50] <mru> anything ffmpeg will decode
  • [17:40:45] <djlewis> ok.
  • [17:45:01] * ArteK (n=Artur@81.15.241.96) has joined #beagle
  • [17:49:39] <ssvb> mru: another fun thing - now I need to reconsider the impact of mixing ARM and NEON instructions, apparently I have fallen into RMK's trap when doing some of the benchmarks
  • [17:49:54] <mru> what trap is that?
  • [17:50:06] <mru> believing everything is armv4?
  • [17:50:06] <ssvb> mru: the one with /proc/cpu/alignment being set to 2 (fixup) by default
  • [17:50:32] <ssvb> it also affects NEON instructions which have strict alignment modifiers
  • [17:50:48] <mru> they always trap
  • [17:51:04] <mru> that's the purpose of the alignment modifier
  • [17:51:13] <mru> sort of
  • [17:51:22] <ssvb> yes, but these traps are hard to notice with this RMK's setup
  • [17:52:09] <ssvb> and looks like I got this crap triggered in one of my preliminary benchmarks
  • [17:52:27] <mru> that's ungood
  • [17:53:46] <ssvb> surely it also produces invalid result as the output, but I mostly cared about checking performance and incorrectly attributed it to the effect of ARM/NEON hazards :)
  • [17:54:57] <mru> there's a reason I don't even build the fixup code
  • [17:55:42] <ssvb> I'm going to remove it from my kernel now
  • [17:56:40] <mru> word of caution though: disabling that code entirely on armv5 gives unexpected results
  • [18:00:51] * Crofton (n=balister@89-26-120-19.saalf.stat.salzburg-online.at) has joined #beagle
  • [18:01:44] <mru> hi Crofton
  • [18:01:51] <mru> still in salzburg?
  • [18:01:59] <Crofton> mru, I just saw a patch for a NEON memcpy on libc-ports
  • [18:02:05] <Crofton> Maria Alm
  • [18:02:06] <ssvb> :)
  • [18:02:08] <Crofton> last night
  • [18:02:17] <mru> Crofton: yeah, we've been discussing it here
  • [18:02:20] <Crofton> ok
  • [18:02:26] <Crofton> wanted to make sure you saw it
  • [18:02:38] <Crofton> it was near the top of my gmail
  • [18:02:50] <Crofton> and I was looking at it to get addresses for postcards :)
  • [18:02:50] <mru> I am mostly of the opinion that optimising memcpy is wasted effort
  • [18:03:01] <Crofton> heh
  • [18:03:32] <mru> the correct way to optimise memcpy is by not calling it
  • [18:03:36] <Crofton> I knew a guy who used memcpy to benchmark memory access speed, then playd with BIOS settings to get best results
  • [18:03:42] <Crofton> hahaha
  • [18:03:48] <mru> a well-written app spends <5% time in memcpy
  • [18:04:01] <Crofton> then I found out that the libc on that particular os sucked
  • [18:04:06] <ssvb> mru: there are not too many well written apps
  • [18:04:22] * mru should write more apps...
  • [18:04:36] <ssvb> mru: and things like gstreamer like to move data around
  • [18:04:45] <mru> look at omapfbplay for instance, not one memcpy call
  • [18:05:07] <mru> in video processing, memcpy is your worst enemy
  • [18:05:14] <Crofton> Holy crap, Sarah Palin is running for president
  • [18:05:23] <mru> president of what?
  • [18:05:25] <ssvb> mru: this thing also likes to overuse threads and mutexes, so having a look at NPTL code may be a good idea
  • [18:05:26] <Crofton> mru, agreed :)
  • [18:05:29] <Crofton> heh
  • [18:05:32] <Crofton> US politics
  • [18:06:44] <mru> ssvb: don't you know the steps of optimisation? 1) optimise the design, 2) optimise algorithms, 3) optimise implementations
  • [18:06:51] <mru> most apps fail at 1
  • [18:07:12] <ssvb> mru: unfortunately I can't do anything about step 1
  • [18:08:11] <mru> we could try to make programming so hard that only the best are able to do it at all
  • [18:08:12] <ssvb> mru: not that I I'm not capable to, but these things are decided by other guys
  • [18:08:20] <mru> yeah, I know
  • [18:08:23] <mru> reality sucks
  • [18:18:53] * tzhau_ (n=tzhau@dsl-245-76-118.telkomadsl.co.za) Quit (Read error: 110 (Connection timed out))
  • [18:22:49] <Crofton> I am starting to move back to reality
  • [18:23:10] * BThompson (n=Bernie@69.7.114.75) has joined #beagle
  • [18:28:11] * Nicce (i=3e48e2f9@gateway/web/freenode/x-c32e90ba6adf834f) has joined #beagle
  • [18:29:29] * Nicce (i=3e48e2f9@gateway/web/freenode/x-c32e90ba6adf834f) Quit (Client Quit)
  • [18:33:08] * djlewis (n=djlewis@75.15.65.232) Quit (Read error: 110 (Connection timed out))
  • [18:37:37] <ssvb> mru: ok, after all NEON seems to be actually a perfect choice for memcpy when processing leading and trailing bytes with ARM code
  • [18:38:09] * j_ack (n=jack@p57A40AD0.dip0.t-ipconnect.de) has joined #beagle
  • [18:38:51] <ssvb> mru: there are no unaligned NEON stores at all, also no write accesses need to be done within 16-byte block mixing ARM and NEON
  • [18:40:23] <ssvb> mru: and penalty for writing data with ARM after NEON in the same cache line is rather small
  • [18:41:45] <ssvb> will post an updated patch tomorrow :)
  • [18:50:28] * kozak (n=subbu@122.166.48.72) Quit ("Leaving.")
  • [18:51:06] <_av500_> Crofton: how long do you stay in EU?
  • [18:51:49] * docelic__ is now known as docelic
  • [18:52:39] * KosiNuss (n=tom@R3e1a.r.pppool.de) has joined #beagle
  • [18:53:02] <Crofton> about 26 more hours :(
  • [18:54:01] <Crofton> l8r all
  • [18:54:10] * Crofton (n=balister@89-26-120-19.saalf.stat.salzburg-online.at) Quit ("Leaving")
  • [18:54:11] <jkridner|work> travel safe.
  • [18:56:11] <_av500_> mru: we forgot something for the BeaglePack, they have to takl to each other over radio so we also draw in the ham radio crowd...
  • [18:56:44] <_av500_> and over shortwave the pack can coordinate long distance :-=
  • [19:03:33] <_av500_> jkridner|work: you are in TX?
  • [19:05:04] <mru> _av500_: right
  • [19:05:12] <mru> and project video on the moon
  • [19:06:54] <mru> anyway, I've now reworked omapfbplay to the point that I can start working on the timer sync
  • [19:07:16] <mru> I'm using a kind of static plugin scheme, if that makes any sense
  • [19:07:16] * tegila (n=tegila@189-015-135-057.xd-dynamic.ctbcnetsuper.com.br) Quit ("Leaving.")
  • [19:10:57] * zedstar (n=john@fsf/member/zedstar) Quit (Read error: 110 (Connection timed out))
  • [19:12:28] * Xenion (n=robert@p579FC340.dip.t-dialin.net) has joined #beagle
  • [19:13:05] <Xenion> Guten Abend alle miteinander / Good evening all together :-)
  • [19:14:24] * mranostay (n=mranosta@cpe-009117.dhcp009.wadsnet.net) Quit ("This computer has gone to sleep")
  • [19:16:42] <jkridner|work> _av500_: I'm based out of Houston.
  • [19:17:02] <muriani> jkridner|work: ..really?
  • [19:17:05] <muriani> what area?
  • [19:17:30] <jkridner|work> Northwest area. TI office is to the Southwest (Stafford).
  • [19:17:34] <muriani> yeah
  • [19:17:42] <muriani> I know where the TI office is :)
  • [19:17:51] <muriani> I've driven by it quite a bit.
  • [19:18:17] <muriani> coolio, I'm off westheimer, just outside the belt
  • [19:18:33] <jkridner|work> excellent.
  • [19:19:21] <jkridner|work> there were a couple good Linux get-togethers at the Coffee Groundz near downtown.
  • [19:19:30] * koen (n=koen@188.90.234.229) has joined #beagle
  • [19:19:36] <jkridner|work> I think they stopped them. They were hosted by the Linux Journal.
  • [19:19:47] <muriani> Indeed.
  • [19:19:54] <mru> hi koen, how are you doing?
  • [19:19:54] <muriani> It's the Geek Gathering nowadays
  • [19:20:04] <jkridner|work> Also did some BeagleBoard demos at BarCamp.
  • [19:20:04] <muriani> well, that's the big nerd event there at least
  • [19:20:14] <muriani> Not familiar with BarCamp
  • [19:20:15] <jkridner|work> k. I'll need to look into that.
  • [19:20:28] <muriani> yeah, it's first fri. of every month
  • [19:20:31] <koen> mru: good, but being kept for more tests...
  • [19:20:37] <muriani> so... would have been yesterday
  • [19:20:41] <muriani> I think they didn't do it then
  • [19:20:48] <mru> koen: they're turning you into a labrat ;-)
  • [19:20:58] <muriani> the Groundz does make some damn good coffee though.
  • [19:21:15] <koen> yeah :)
  • [19:21:24] * jkridner|work isn't that into coffee and drinks beer there. :)
  • [19:22:06] <muriani> hehe, yeah. I do that too.
  • [19:22:09] <muriani> coffee, then beer.
  • [19:22:10] <mru> muriani: do you mean *good* or good by american standards?
  • [19:22:19] <muriani> mru: starbucks is pure trash.
  • [19:22:21] <koen> they only seem to have decaff here
  • [19:22:25] <muriani> I can actually drink the coffee here.
  • [19:22:43] <mru> what has starbucks got to do with anything?
  • [19:22:43] * Xenion (n=robert@p579FC340.dip.t-dialin.net) Quit (Read error: 104 (Connection reset by peer))
  • [19:22:44] <muriani> what most of the US considers coffee is completely unpalatable to me.
  • [19:22:48] <mru> we were talking about coffee
  • [19:22:52] <muriani> hahaha
  • [19:22:55] <muriani> Exactly :p
  • [19:23:15] <mru> good coffee is hard to find in the states
  • [19:23:15] * Xenion (n=robert@p579FC340.dip.t-dialin.net) has joined #beagle
  • [19:23:27] <muriani> good coffee shops are hard to find.
  • [19:23:28] <mru> much easier here, I just go to my kitchen
  • [19:23:32] <muriani> Indeed.
  • [19:24:32] * koen has a moka stovetop unit
  • [19:24:34] <muriani> I enjoyed being an independent barista, so I had to have a setup at home for coffee :P
  • [19:27:43] * koen (n=koen@188.90.234.229) Quit ("Get Colloquy for iPhone! http://mobile.colloquy.info")
  • [19:29:59] * _av500_ smells coffee
  • [19:30:26] * billenium smells _av500_
  • [19:30:41] * mru smells
  • [19:30:55] <mru> no, not really (I hope)
  • [19:31:52] * Xenion (n=robert@p579FC340.dip.t-dialin.net) has left #beagle
  • [19:33:33] <shell_> any news on igepv2?
  • [19:42:01] * mranostay (n=mranosta@cpe-009117.dhcp009.wadsnet.net) has joined #beagle
  • [19:43:57] * JDuke128 (n=kadir@212.174.90.98) has joined #beagle
  • [19:44:12] <JDuke128> hi , did someone add ccd camera on beagleboard ? is it possible ? and easy ?
  • [19:44:30] <mru> you can use a usb webcam
  • [19:44:54] <_av500_> JDuke128: direct is not easy, the ccdc pins are not routed
  • [19:44:55] <JDuke128> can i use 4 usb at the same time ?
  • [19:45:05] <JDuke128> i want direct yes
  • [19:45:08] <JDuke128> better then usb
  • [19:45:48] <_av500_> direct is a nogo
  • [19:46:02] <JDuke128> nogo ?
  • [19:46:21] <_av500_> no.go
  • [19:46:35] <JDuke128> usb faster ?
  • [19:46:36] <_av500_> no-go, it does not work
  • [19:46:50] <_av500_> the ccdc pins of the omap3 are not routed on the BB
  • [19:47:11] <JDuke128> any hardware h264 or on2 vp6 codec beagleboard has for cameras ?
  • [19:47:39] <_av500_> hw no, sw yes
  • [19:48:01] <JDuke128> hmm
  • [19:48:39] <_av500_> if you want hw and ccd look at that dm355 based board
  • [19:48:45] <JDuke128> so when i do realtime encoding it slows much
  • [19:48:57] <JDuke128> did u heard about leopardboard ?
  • [19:49:19] <_av500_> yesm thats what i meant
  • [19:49:48] <JDuke128> ahh but it doesnt have opengl es
  • [19:49:57] <JDuke128> i want to mix opengl es with h264 codec
  • [19:50:07] <_av500_> mix?
  • [19:50:16] <JDuke128> augmented reality
  • [19:50:38] <JDuke128> like android vision
  • [19:51:03] * _av500_ has sometimes trouble to copy with normal reality...
  • [19:51:07] <_av500_> cope
  • [19:51:22] <mru> nice freudian slip...
  • [19:51:50] <_av500_> mru: no still damn frogpad...
  • [19:52:04] <JDuke128> btw , its better to use j2se or j2me on beagleboard ?
  • [19:52:05] * mru wants a hud like in those games where enemies are indicated in flashing colours
  • [19:52:23] <mru> it's better to not use java at all
  • [19:52:43] <JDuke128> c++ linux...
  • [19:52:55] <mru> assembler !
  • [19:53:01] * _av500_ wanted to type "mru: hit him" but was too slow...
  • [19:53:23] <JDuke128> i was thinking about installing OpenSolaris ARM port to beagle board with j2se and javafx
  • [19:53:28] * mru agrees with the guy who said assembler is the most fun you can have with your clothes on
  • [19:53:43] <JDuke128> assembler really takes much time to develop
  • [19:53:50] <mru> I guess SM is possible with clothes on....
  • [19:55:38] <_av500_> JDuke128: could it be that your reality is already slightly augmented?
  • [20:01:01] * KosiNuss (n=tom@R3e1a.r.pppool.de) Quit (Read error: 60 (Operation timed out))
  • [20:01:32] * zedstar (n=john@fsf/member/zedstar) has joined #beagle
  • [20:04:05] * j_ack (n=jack@p57A40AD0.dip0.t-ipconnect.de) Quit ("Verlassend")
  • [20:08:38] * rsalveti (n=rsalveti@189.70.35.140) Quit (Remote closed the connection)
  • [20:08:51] * rsalveti_ (n=rsalveti@189.70.35.140) Quit (Remote closed the connection)
  • [20:09:11] * rsalveti (n=rsalveti@189.70.35.140) has joined #beagle
  • [20:11:13] <muriani> mru: there's one here working on some AR stuff, and I'll be working on HUD once the vuzix wrap920 glasses are released
  • [20:11:45] <mru> how will it know who's the enemy?
  • [20:12:01] <mru> can you make it detect people wearing a java t-shirt and target-designate them?
  • [20:12:16] <muriani> if you rig a camera to it, I'm sure that's posible :p
  • [20:12:30] <muriani> my implementation won't have a camera, at least at first
  • [20:12:37] <mru> then we just need to arm the beaglepack with usb missile launchers
  • [20:12:43] <muriani> hahahah
  • [20:14:56] <Downix> don't tempt me
  • [20:17:50] * rsalveti (n=rsalveti@189.70.35.140) Quit (Read error: 104 (Connection reset by peer))
  • [20:20:07] <_av500_> mru: yeah, the hunter-killer approach
  • [20:22:19] <_av500_> actually, usb missile lauchers could be a cheap platform to aim the picos
  • [20:23:32] <ShadowJK> the picos?
  • [20:25:03] * florian (n=fuchs@f054222069.adsl.alicedsl.de) has joined #beagle
  • [20:26:07] <_av500_> project BeaglePack
  • [20:33:00] <mru> _av500_: good idea
  • [20:33:25] <mru> and leave the missiles on for a more menacing look
  • [20:33:29] <_av500_> yes
  • [20:33:36] <_av500_> or for a davinciHD...
  • [20:34:26] <mru> right, davincihd is the bad guy
  • [20:35:34] <_av500_> if we repeat it often enough :-)
  • [20:41:10] * rsalveti (n=rsalveti@200.184.118.130) has joined #beagle
  • [20:42:03] <Downix> ...and the true origin of Skynet is now known, the Beagleboard....
  • [20:42:04] <Downix> 8)
  • [20:44:52] * LubeckTech (i=d814858a@gateway/web/freenode/x-76255c744252daa3) has joined #beagle
  • [20:51:47] * _av500_ googles for cheap 2nd hand nuke proof bunker
  • [20:53:43] <Downix> _av500_, I know of a few for sale
  • [20:54:03] <Downix> there's a thriving market for de-militarized former ICBM silos
  • [21:00:48] <ThomasEgi> hm? wlan-paranoid?
  • [21:02:11] <_av500_> Downix: I take the one with Claire Danes preinstalled...
  • [21:17:22] * Wiedi (n=wiedi@newton-air.w.fruky.net) Quit ("^C")
  • [21:44:51] * Frisco (i=4cb768a3@gateway/web/freenode/x-a5e400d95ba7c49c) has joined #beagle
  • [21:45:17] * zedstar (n=john@fsf/member/zedstar) Quit (Read error: 110 (Connection timed out))
  • [22:00:29] * sakoman_ (n=sakoman@static-74-41-60-154.dsl1.pco.ca.frontiernet.net) Quit (Remote closed the connection)
  • [22:10:16] * florian (n=fuchs@f054222069.adsl.alicedsl.de) Quit ("Verlassend")
  • [22:18:46] * ArteK (n=Artur@81.15.241.96) has left #beagle
  • [22:22:48] * Frisco (i=4cb768a3@gateway/web/freenode/x-a5e400d95ba7c49c) Quit ("Page closed")
  • [22:38:50] * ThomasEgi (n=thomas@pppdyn-03.stud-ko.rz-online.net) Quit (Remote closed the connection)
  • [23:00:19] * sakoman (n=sakoman@static-74-41-60-154.dsl1.pco.ca.frontiernet.net) has joined #beagle
  • [23:09:15] * pH5 (n=ph5@p5485C2FC.dip.t-dialin.net) Quit ("Leaving")
  • [23:40:29] * djlewis (n=djlewis@75.15.65.232) has joined #beagle
  • [23:42:19] * tzhau (n=tzhau@dsl-245-76-118.telkomadsl.co.za) has joined #beagle
  • [23:51:55] <djlewis> so quiet . . .
  • [23:52:05] <mru> saturday night...
  • [23:52:22] <jkridner|work> holiday in US.
  • [23:52:42] <jkridner|work> only those without any life whatsoever around the house still. :)
  • [23:52:52] <djlewis> Hey, thats me :)
  • [23:52:57] <mru> like jkridner|work ...
  • [23:53:04] <jkridner|work> :)
  • [23:53:08] <mru> note the |work suffix...
  • [23:53:23] <djlewis> but you have your work name up and at home?
  • [23:53:28] <djlewis> I did but too late.
  • [23:53:57] <jkridner|work> I work from my laptop and it is up in front of me.
  • [23:54:11] <jkridner> I am at home too. :)
  • [23:54:15] <djlewis> i'm tryint task-native-sdk install on this 2.6.30
  • [23:54:32] <mru> "working from home" is the corporate version of "dog ate my homework"
  • [23:54:48] <jkridner|work> odd analogy.
  • [23:55:07] <djlewis> but my dog did.. it was a cake...
  • [23:55:09] <jkridner|work> feels more like dog did my homework.
  • [23:55:15] <mru> beagle...
  • [23:55:36] <djlewis> no, the shepard/collie mix 100lb'er
  • [23:56:02] <djlewis> this Beaagle eats my lunch.
  • [23:56:37] <mru> my sister's dog will find mushrooms for her
  • [23:57:53] <djlewis> I dont know a good shroom from poison.
  • [23:58:07] <mru> hopefully the dog does...
  • [23:59:02] <djlewis> mru: this 2.6.30, I forgot I had not put the build tools on it yet. Its working on them though.
  • [23:59:34] <djlewis> opkg is pulling down unstable