| 00:01.25 | provel_ | ssvb: it's just emulator that's more than 50%cpu without rescaling |
| 00:01.34 | provel_ | ssvb: then you need to scale |
| 00:02.10 | provel_ | ssvb: and likely change color space/apply shaders |
| 00:02.11 | libv | provel_: which is what overlays could be used for |
| 00:02.21 | libv | again, overlays. |
| 00:02.43 | provel_ | there's some doc about overlays? some example code? |
| 00:02.46 | libv | it should be a few days work for anyone to figure out how the overlays work on the disp engine |
| 00:02.58 | libv | and hooking things into Xv should be easy |
| 00:03.05 | ssvb | libv: yep, doing this now :) |
| 00:03.20 | libv | but then, very few existing game backends actually know about xv. |
| 00:03.37 | provel_ | xv = xvideo? |
| 00:03.45 | libv | i doubt that sdl uses it for rescaling/colourspace conversion |
| 00:03.49 | libv | yes |
| 00:03.57 | ssvb | provel_: scaling should be also reasonably fast on the CPU |
| 00:04.14 | provel_ | oh god.... also xvideo is good |
| 00:04.43 | provel_ | ssvb: scaling 250x220 to 1280x720? |
| 00:05.23 | provel_ | however retroarch has xvideo backend too |
| 00:05.28 | *** join/#arm-netbook jlj (~jlj@109-104-24-9.customers.ownit.se) |
| 00:06.51 | provel_ | bye |
| 00:06.57 | ssvb | provel_: pixman can do ~90MPix/s for bilinear scaling, which means ~100FPS and somewhat faster for nearest (this is memory bandwidth bound) |
| 00:07.13 | libv | but that eats 100% cpu |
| 00:07.22 | ssvb | yes |
| 00:07.56 | ssvb | he has 50% spare CPU, so we are down to just 50FPS |
| 00:08.16 | ssvb | that is unless there is something else doing stupid things in the pipeline |
| 00:09.13 | libv | i am amazed that this retroarch thing depends on openvg |
| 00:09.55 | ssvb | it might be not a bad idea for the platforms where openvg actually works and is hardware accelerated |
| 00:13.39 | libv | this might be worth a go https://github.com/micahpearlman/MonkVG |
| 00:31.45 | *** join/#arm-netbook freakazoid0223 (~matt@pool-173-75-233-172.phlapa.fios.verizon.net) |
| 00:55.39 | *** join/#arm-netbook Entropy512 (~quassel@cpe-69-207-202-50.stny.res.rr.com) |
| 01:06.47 | *** join/#arm-netbook L84Supper (~Larch@unaffiliated/l84supper) |
| 01:10.52 | *** join/#arm-netbook stefanro1 (~stefan@p57B94DF2.dip.t-dialin.net) |
| 02:05.53 | *** join/#arm-netbook Jef91 (~Instantbi@bodhilinux/team/Jef91) |
| 02:39.02 | *** join/#arm-netbook itamarjp (~itamar@cl-24.udi-01.br.sixxs.net) |
| 02:39.02 | *** join/#arm-netbook itamarjp (~itamar@fedora/itamarjp) |
| 03:47.24 | *** join/#arm-netbook IEF (ief@shadowchild.nl) |
| 03:47.27 | *** join/#arm-netbook ganbold__ (~Ganbold@202.21.108.231) |
| 03:47.33 | *** join/#arm-netbook jelly (~jelly@pdpc/supporter/active/jelly) |
| 03:47.34 | *** join/#arm-netbook lkcl (~lkcl@host-92-26-19-91.as13285.net) |
| 04:40.54 | *** join/#arm-netbook toxicpsion (~toxicpsio@h216-45-98-25.dynamic.platinum.ca) |
| 04:52.31 | *** join/#arm-netbook bsdfox (~Bob@unaffiliated/bsdfox) |
| 04:54.50 | *** join/#arm-netbook aholler (~aholler@p57B20211.dip0.t-ipconnect.de) |
| 05:42.22 | toxicpsion | PMU is AXP right? @ i2c bus 0: addr: 0x34 on mele right? |
| 06:48.58 | *** join/#arm-netbook Avernos_ (~avernos@111.194.208.98) |
| 06:59.08 | *** join/#arm-netbook robclark` (~robclark@ec2-50-19-223-142.compute-1.amazonaws.com) |
| 07:01.44 | *** join/#arm-netbook Quarx (~Quarx@109.120.46.213) |
| 07:22.28 | *** join/#arm-netbook gimli (~gimli@xbmc/staff/gimli) |
| 08:01.20 | *** join/#arm-netbook bsdfox (~Bob@unaffiliated/bsdfox) |
| 08:28.32 | *** join/#arm-netbook Gujs_ (~Gujs@cpe-92-37-29-204.dynamic.amis.net) |
| 08:31.55 | *** join/#arm-netbook eFfeM (~frans@c73189.upc-c.chello.nl) |
| 08:44.11 | *** join/#arm-netbook rellla (~quassel@p4FE5639E.dip0.t-ipconnect.de) |
| 08:50.35 | *** join/#arm-netbook provel_ (~matte@host34-135-dynamic.18-79-r.retail.telecomitalia.it) |
| 08:50.37 | provel_ | hi |
| 08:51.57 | provel_ | still trying to make EGL work under framebuffer..... still one question |
| 08:52.28 | provel_ | I have an egl surface rendering.... but it doesn't seems to "collate" with the framebuffer |
| 08:53.04 | provel_ | is there a way I can get an handle/element from framebuffer to pass it as EGLNativeWindow to the EGL? |
| 08:54.15 | *** join/#arm-netbook popolon (~popolon@og-free.planet-service.fr) |
| 08:56.26 | *** join/#arm-netbook rellla (~Thunderbi@p4FE5639E.dip0.t-ipconnect.de) |
| 08:56.48 | provel_ | I'm trying open("/dev/fb0", O_RDRW) but I doubt... |
| 09:00.51 | provel_ | segmentation fault, as expected :-) |
| 09:05.23 | provel_ | hum, looking at eglplatform.h I swear it's impossible |
| 09:05.56 | provel_ | there should be an element field there |
| 10:05.46 | orly_owl | this is probably old news in here http://www.tgdaily.com/hardware-features/67971-allwinner-rolls-out-new-arm-powered-chips-for-android-devices |
| 10:05.51 | orly_owl | a20 and a31 |
| 10:10.29 | *** join/#arm-netbook Quarx (~Quarx@109.120.46.213) |
| 10:27.52 | *** join/#arm-netbook vgrade (~root@static.43.45.47.78.clients.your-server.de) |
| 10:33.32 | rellla | using stage/sunxi-3.0 and the latest hdmi-commits i'm wondering if i have to give some arguments to boot.scr or uEnv.txt to get hdmi running? dmesg give some edid-output, but screen doesn't come up!? |
| 10:49.13 | *** join/#arm-netbook gimli (~gimli@xbmc/staff/gimli) |
| 10:58.16 | *** join/#arm-netbook hansg (~hans@ip32-174-211-87.adsl2.static.versatel.nl) |
| 11:21.41 | *** join/#arm-netbook stefanro (~stefan@p57B95FC1.dip.t-dialin.net) |
| 11:30.16 | *** join/#arm-netbook andi2342346 (~none@e178186193.adsl.alicedsl.de) |
| 11:39.29 | *** join/#arm-netbook lerc (~quassel@121.75.134.249) |
| 11:47.27 | rm | mnemoc, hey |
| 11:47.30 | *** join/#arm-netbook provel_ (~matte@host34-135-dynamic.18-79-r.retail.telecomitalia.it) |
| 11:47.52 | rm | which kernel branch should I use if I want both latest Mali, and the HDMI stuff? |
| 11:51.00 | provel_ | 3.0 in my setup is not compatible with r3p1 x11 dmali drivers |
| 11:52.07 | provel_ | are shaders supposed to work with Mali drivers? |
| 12:15.00 | *** join/#arm-netbook penguin42 (~dg@tu006.demon.co.uk) |
| 12:18.58 | techn_ | provel_: that's right |
| 12:19.21 | techn_ | you need r3p1 kernel stuff also |
| 12:26.44 | provel_ | but 3.4 seems unchanged since months, while 3.0 gets the updates (for example for usb otg port..) |
| 12:28.10 | provel_ | is r3p1 working better/faster than 3.0 in xorg? in my setup xorg eats lots of CPU time (even more than 50% some times, with opengl) |
| 12:28.48 | techn_ | opengl is not hw accelerated what I know |
| 12:32.14 | provel_ | sorry... it's gl |
| 12:32.24 | provel_ | gles |
| 12:32.44 | provel_ | still xorg consumes lot of cpu time |
| 12:33.07 | techn_ | what is your cpu frequency? |
| 12:33.26 | techn_ | https://github.com/linux-sunxi/linux-sunxi/tree/sunxi-3.4 |
| 12:33.42 | techn_ | last update on 3.4 is from 2 days ago |
| 12:33.46 | provel_ | oh |
| 12:34.05 | provel_ | and 3.4 is faster? |
| 12:34.58 | techn_ | most likely not.. no one has done any benchmarking |
| 12:35.12 | techn_ | but I think differences are only marginal |
| 12:36.44 | provel_ | hum... some benchs here :http://linux-sunxi.org/Benchmarks |
| 12:37.38 | provel_ | but I don't know were to get things like es2_gears to test my own |
| 12:41.04 | *** join/#arm-netbook jas-hacks (~mtx512@188-223-215-73.zone14.bethere.co.uk) |
| 12:42.22 | mnemoc | rm: latest hdmi stuff is in stage/sunxi-3.x |
| 12:43.13 | jas-hacks | provel_ : You compile es2_gears (part of mesa-demo) from here http://cgit.freedesktop.org/mesa/demos, need to compile yourself |
| 12:46.47 | ganbold__ | where can I find about fex file syntax? |
| 12:47.11 | provel_ | jas-hack: thank you |
| 12:48.55 | jukivili | how often latest stable(s) is merged to sunxi trees? |
| 12:49.57 | jas-hacks | provel_: I suggest you try razor-qt desktop, I found it be snappy considering no hw acceleration. |
| 12:54.56 | jas-hacks | ganbold__: google for 'crane_fex_guide' you'll find a pdf |
| 12:55.54 | *** join/#arm-netbook Turl_ (~Turl@cyanogenmod/maintainer/Turl) |
| 13:02.59 | ganbold__ | jas-hacks: ok, the problem is I can't give power to USB while booting other OS |
| 13:05.23 | jas-hacks | ganbold__: so want to disable the gpio pin for USB port ? |
| 13:06.37 | ganbold__ | jas-hacks: maybe other way, driver can't find USB since it is not powered |
| 13:11.13 | provel_ | still... I can't believe that my a10 it's slower than a raspberry Pi |
| 13:11.54 | jas-hacks | ganbold__: You can try setting 'usb_drv_vbus_gpio' but you would need to know the gpio pin. |
| 13:12.02 | jukivili | provel_: do you have cpufreq enabled? |
| 13:12.34 | provel_ | jukivili : think so... 1000 bogomips |
| 13:12.57 | provel_ | on load |
| 13:13.10 | techn_ | provel_: which benchmark you are using? |
| 13:13.28 | provel_ | retroarch emulator (ssnes9x and psnes) |
| 13:14.10 | provel_ | I compiled all with ffast-math , unrolls, neon, etc.etc. |
| 13:14.20 | techn_ | x or framebuffer? |
| 13:14.56 | *** join/#arm-netbook Skoti (589cdff4@gateway/web/freenode/ip.88.156.223.244) |
| 13:15.35 | techn_ | someone benchmarked x with mali drivers.. and performance was killed by malloc |
| 13:16.03 | provel_ | malloc? O__o |
| 13:16.14 | techn_ | [01:57:53] <rz2k> provel_: ssvb has done some measurements on xf86-video-mali. https://groups.google.com/d/msg/linux-sunxi/6lLMjfvZj44/lH5kFRB0bN8J |
| 13:17.13 | ganbold__ | jas-hacks: but gpio pin is defined in script.bin, right? |
| 13:18.26 | provel_ | techn_: very nice thanks..... now need just to figure out how to speedup |
| 13:18.37 | jas-hacks | ganbold__: Yes. |
| 13:18.41 | *** join/#arm-netbook hg_5 (~chatzilla@unaffiliated/hg-5/x-8664886) |
| 13:21.53 | ssvb | techn_: the performance is killed by both lack of cpu caching for the ump allocated pixmap memory, and the expensive ioctls doing these ump allocations |
| 13:23.26 | ssvb | provel_: just profile the code to know *what* to speedup, the answer *how* to speedup it may be sometimes obvious after the profiling is done |
| 13:25.20 | techn_ | ssvb: what profiling method you used? |
| 13:26.12 | techn_ | or is that embedded to that cairo |
| 13:26.30 | techn_ | since I'd like to profile xbmc sometime |
| 13:27.10 | ssvb | techn_: "perf" tool, but oprofile should also work fine if you prefer it |
| 13:28.53 | ssvb | techn_: some hints are here for example: http://www.pixelbeat.org/programming/profiling/ (just a randomly googled link) |
| 13:30.02 | techn_ | I tried couple tools ( which doesnt need compiling with profiling hooks).. but valgrind kills performane.. kernel debugfs didn't work right away.. strace-like tool crashed |
| 13:32.18 | ssvb | I don't remember if I had to enable extra some kernel options to get the performance counters support for perf |
| 13:32.39 | ssvb | but this should be quite easy |
| 13:36.38 | ssvb | provel_: "in my setup xorg eats lots of CPU time (even more than 50% some times, with opengl)" - yes, because the cpu is busy copying data from the buffers where mali does all the rendering to the framebuffer |
| 13:37.29 | ssvb | provel_: and these buffers are of course uncached, which means several times worse performance than you would expect for memcpy with normal cached memory |
| 13:39.25 | ssvb | provel_: for your emulator thing, a quick check would be to just use xf86-video-fbdev driver to see if it performs better (without gles, just using software rendering) |
| 13:39.51 | ganbold__ | jas-hacks: doesn't work for me, maybe I'm missing something, probably need to get linux-sunxi source and try to build kernel with some debug printf and see |
| 13:41.38 | jas-hacks | ganbold__: What device are you testing on? |
| 13:43.43 | ganbold__ | cubieboard |
| 13:45.42 | ganbold__ | so something like for sata: volatile uint32_t *sata_reg = (uint32_t *) 0xe1c20828; *sata_reg |= (1 << 0); /* set for output */ |
| 13:45.56 | ssvb | provel_: obviously the high cpu usage for gles is not normal, dri2 mali integration needs some fixes to get the best performance out of it |
| 13:46.57 | ganbold__ | maybe I'm missing something, but after that I supposed to get voltage on sata connector, am I right? |
| 13:50.31 | jas-hacks | ganbold__: Looking a cubie schematic, I guess power is controller through PH6? |
| 13:51.05 | ganbold__ | jas-hacks: for usb1 I think ph06, but for sata it is PB08 |
| 13:53.18 | jas-hacks | ganbold__: The quickest way to test the pins is to use hno's uboot. You can then set each pin to determine if it is the right one or not. |
| 13:53.27 | ganbold__ | jas-hacks: so pb08 means pio_base + Offset: 0x28 and 0-2 bits and 001 sets for output |
| 13:53.45 | ganbold__ | jas-hacks: where to find hno uboot |
| 13:54.35 | jas-hacks | ganbold__: https://github.com/hno/uboot-allwinner |
| 13:57.10 | jas-hacks | ganbold__: There is also a utility in https://github.com/linux-sunxi/sunxi-tools - pio.c to do something similar if you don't want to use uboot. |
| 13:57.26 | ganbold__ | jas-hacks: does hno uboot have commands for gpio testing |
| 13:58.59 | jas-hacks | ganbold__: You can do something like this 'gpio set ph19', just type help for more info. |
| 14:00.06 | ganbold__ | jas-hacks: I see, how to use sunxi-tools? |
| 14:03.13 | libv | provel_: have you tried looking closely at our mali-libs tree? |
| 14:03.44 | jas-hacks | ganbold__: sunix-tools is similar |
| 14:06.00 | ganbold__ | jas-hacks: do I need to go to FEL mode? or make some image and write to sd card and boot? |
| 14:09.11 | jas-hacks | ganbold__: for uboot, yes need to create a bootable SD with just hno uboot on it and serial console attached. It may be quicker to just compile sunix-tools and run that from your linux distro. |
| 14:12.26 | ganbold__ | jas-hacks: for uboot right, for sunxi-tools I thought totally different thing :) |
| 14:14.33 | *** join/#arm-netbook andi2342346 (~none@e178186193.adsl.alicedsl.de) |
| 14:29.03 | *** part/#arm-netbook jas-hacks (~mtx512@188-223-215-73.zone14.bethere.co.uk) |
| 15:00.38 | provel_ | huf... still struggling for better gpu... anyone here used fb drivers with openVG ? should I revompile everything? |
| 15:00.56 | libv | provel_: no. |
| 15:01.09 | libv | provel_: there is no openvg available for mali at this time. |
| 15:01.32 | provel_ | ok |
| 15:10.52 | provel_ | I'll likely give up.... if no-one else has already a decently working port of Retroarch.... mine's still going at half speed for snes games |
| 15:51.55 | *** join/#arm-netbook rz2k (~rz2k@128-69-239-76.broadband.corbina.ru) |
| 16:23.32 | *** join/#arm-netbook itamarjp (~itamar@cl-24.udi-01.br.sixxs.net) |
| 16:23.32 | *** join/#arm-netbook itamarjp (~itamar@fedora/itamarjp) |
| 16:29.22 | *** join/#arm-netbook bsdfox (~Bob@unaffiliated/bsdfox) |
| 16:43.17 | *** join/#arm-netbook Jef91 (~Instantbi@bodhilinux/team/Jef91) |
| 17:06.07 | *** join/#arm-netbook andi23423467 (~none@e179074092.adsl.alicedsl.de) |
| 17:47.26 | *** join/#arm-netbook andi2342346 (~none@e179140144.adsl.alicedsl.de) |
| 18:35.25 | *** join/#arm-netbook Workboot (~gumboot@rev.bovine.muck.net.nz) |
| 18:59.26 | *** join/#arm-netbook mdasoh_ (~mdasoh@253-175-174-206.cust.wirelessbeehive.com) |
| 19:19.48 | *** join/#arm-netbook ph0en1x (~ph0en1x@c-98-210-135-53.hsd1.ca.comcast.net) |
| 19:20.52 | ph0en1x | hey all. what's the easiest way to identify if my generic mk802 is an 802ii or a regular 802? |
| 19:21.36 | ph0en1x | looking online at images of the pcb, it looks more like an 802ii than an 802 |
| 19:22.00 | ph0en1x | i see the power protect IC on the board |
| 19:22.47 | ph0en1x | and mine has the Uniciron AA029? antenna chip rather than a wire antenna |
| 19:33.38 | *** join/#arm-netbook slapin_nb (~slapin@2a02:578:5002:8016:3238:55ff:fe02:6bad) |
| 19:34.58 | rm | not the first one then |
| 19:35.01 | rm | why do you care anyways |
| 19:37.39 | ph0en1x | getting a development environment set up |
| 19:39.02 | ph0en1x | i noticed linaro doesnt have any precise rootfs images up anymore |
| 19:41.23 | mnemoc | it does, but not under the "latest" symlink |
| 19:41.44 | ph0en1x | really? i poked around and all the directories were empty |
| 19:42.15 | ph0en1x | everything under here only has a latest: |
| 19:42.16 | ph0en1x | https://snapshots.linaro.org/precise/images |
| 19:43.47 | mnemoc | http://releases.linaro.org/12.11/ubuntu/precise-images/alip |
| 19:44.23 | ph0en1x | awesome |
| 19:45.52 | ph0en1x | guess i'll have to update the link on the wiki |
| 19:57.36 | mnemoc | that's why tutorials on forums stink and tutorials on wikis are cool :) |
| 20:01.29 | penguin42 | you mean the way the links are wrong the instant after you post something to a forum? |
| 20:04.40 | mnemoc | yeah :) |
| 20:05.48 | slapin_nb | Happy New Year!!!! |
| 20:18.30 | mnemoc | happy 2013 slapin_nb |
| 20:19.20 | *** join/#arm-netbook darkvlad (~ph0en1x@91.215.159.37) |
| 20:22.27 | *** join/#arm-netbook ccssnet (~ccssnet@c-98-216-141-157.hsd1.ma.comcast.net) |
| 20:28.40 | *** join/#arm-netbook Jef91 (~Instantbi@bodhilinux/team/Jef91) |
| 20:39.22 | *** join/#arm-netbook ccssnet (~ccssnet@c-98-216-141-157.hsd1.ma.comcast.net) |
| 21:02.49 | *** join/#arm-netbook jelly (~jelly@pdpc/supporter/active/jelly) |
| 21:06.39 | *** join/#arm-netbook servili007 (~servili00@unaffiliated/servili007) |
| 21:11.06 | *** join/#arm-netbook WarheadsSE (~WarheadsS@c-174-59-252-104.hsd1.pa.comcast.net) |
| 21:18.27 | *** join/#arm-netbook marcan (marcan@marcansoft.com) |
| 21:40.59 | *** join/#arm-netbook Jef91 (~Instantbi@bodhilinux/team/Jef91) |
| 21:41.54 | ManoftheSea | I know I've asked for (and received) a lot of help. Could I beg help in getting my audio to work on this mk802ii? |
| 21:42.09 | ManoftheSea | I expect it to go through the hdmi, and the TV speakers to play sound. |
| 21:45.42 | specing | swings his magic wand of sound |
| 21:46.46 | jelly-home | is that what you call it |
| 21:48.31 | ManoftheSea | actually, yeah, if I had a starting place that'd be nice. I have pulseaudio on Debian Wheezy on an mk802ii hooked to a tv through hdmi. |
| 21:50.25 | ManoftheSea | I have a /dev/snd/by-path/{platform-soc-audio.0,platform-sun4i-codec} |
| 21:55.24 | ManoftheSea | hmm, some clicks and pops. That's... better than nothing. |
| 21:55.36 | ManoftheSea | How about a way to test that cedarx is working? |
| 21:57.33 | *** join/#arm-netbook WarheadsSE (~WarheadsS@c-174-59-252-104.hsd1.pa.comcast.net) |
| 22:07.32 | *** join/#arm-netbook XenGi (~XenGi@2a02:748:a800:149:154:158:199:1) |
| 22:13.15 | *** part/#arm-netbook nulluser (~Justin@99-46-220-68.lightspeed.livnmi.sbcglobal.net) |
| 22:17.45 | *** join/#arm-netbook XenGi_ (~XenGi@2a02:748:a800:149:154:158:199:1) |
| 22:36.37 | *** join/#arm-netbook popolon (~popolon@og-free.planet-service.fr) |
| 22:46.35 | *** join/#arm-netbook arete74 (~arete74@net-93-64-241-175.cust.dsl.vodafone.it) |
| 22:46.53 | *** join/#arm-netbook gzamboni_ (~gzamboni@94.191.15.109.rev.sfr.net) |
| 22:48.40 | *** join/#arm-netbook WarheadsSE (~WarheadsS@c-174-59-252-104.hsd1.pa.comcast.net) |
| 22:51.25 | *** join/#arm-netbook XenGi (~XenGi@2a02:748:a800:149:154:158:199:1) |
| 23:22.56 | rz2k | you now recieve knight status for writing bad APIs and providing zero customer support. http://www.eetimes.com/electronics-news/4403955/Imagination-Technologies-CEO-knighted- |
| 23:25.46 | penguin42 | rz2k: It's better than it used to be; many years ago you had to conquer far away lands and slaughter their people |
| 23:27.02 | rz2k | lol |
| 23:27.04 | rz2k | :) |
| 23:27.42 | rz2k | hardkernel gave us HNY present - they've finally pushed updated 3.0.51 for odroid-x2 |
| 23:44.23 | *** join/#arm-netbook freakazoid0223 (~matt@pool-173-75-233-172.phlapa.fios.verizon.net) |
| 23:52.58 | ssvb | rz2k: do you also have odroid-x2? |
| 23:56.37 | rz2k | yeah |