00:00 | steph_ | Hi all :) |
00:03 | steph33560 | I've just received my cubox i2 ultra. Trying bootup the cubox installer but no ttyUSB found. Somebody can help ? |
00:04 | steph33560 | I've just received my cubox i2 ultra. Trying bootup the cubox installer but no ttyUSB found. Somebody can help ? |
00:04 | xraxor | what are you trying to install? |
00:04 | steph33560 | xraxor: U-Boot |
00:05 | xraxor | AFAIK the cubox installer does not work on cubox-i |
00:05 | steph33560 | HAA :) thanks |
00:05 | xraxor | did you check the forums? http://imx.solid-run.com/forums/index.php |
00:06 | steph33560 | xraxor: only the wiki :p |
00:06 | xraxor | theres a subforums about uboot http://imx.solid-run.com/forums/viewforum.php?f=10&sid=d7df8c79188f3cfcc40fd026c8114b87 |
00:06 | xraxor | make sure you are checking on right wiki |
00:07 | xraxor | here wiki http://imx.solid-run.com/wiki/index.php?title=Main_Page |
00:07 | steph33560 | Thanks xraxor :) |
00:08 | steph33560 | let's go for a jessi :) |
00:09 | xraxor | good luck |
00:09 | xraxor | :) |
01:07 | mk01 | anyone knows which .config (3.0.35kernel) parameter enables IR on cubox-i ? I have the feeling I enabled all kinds of IR or USB / whatever adapters and still no /sys/class/rc ... |
01:08 | nikitis | is rabeeh here? |
09:55 | rabeeh | mk01: ping; still online? |
09:55 | rabeeh | look at arch/arm/configs/imx6_cubox-i_hummingboard_defconfig |
09:56 | rabeeh | CONFIG_IR_GPIO_CIR is the important one in the IR core to enable IR on gpio |
09:57 | rabeeh | notice that current IR defaults to RC6 |
10:14 | mk01 | rabeeh: yes still here. thanks. I was checking GPIO stuff as well but it's too much records. overlooked the IR in name. |
10:38 | mk01 | rabeeh: if gpio-rc-recv is module, /sys/class/rc* is not created although all dependent modules are triggered and reporting no errors |
10:38 | mk01 | rabeeh: that's why I could not get it "BACK". |
10:38 | mk01 | rabeeh: but now in-kernel is ok, thank |
10:59 | rabeeh | mk01: /dev/lirc ? |
11:00 | rabeeh | on my kernel - http://pastebin.com/F45zcPvL |
11:00 | rabeeh | it's the github kernel |
11:05 | mk01 | rabeeh: no. this gets creatrd when you load lirc compatibility keymap. but "raw" (rc0,rc1,rc2..._) before keymap translation. |
11:09 | mk01 | rabeeh: but the log is from kenrel where the module is in-kernel, is it not? then it is ok. if is as module, then rcX are missing (not created). not a big issue though, just the piece of info was important |
14:31 | warped-rudi | jnettlet: ping |
15:02 | jnettlet | warped-rudi, here. Sorry dog training |
15:02 | warped-rudi | Nice ;-). |
15:03 | warped-rudi | I was kinda bored and added lmbench to the geexbox build. Now I have the tools on any platform geexbox runs on. |
15:04 | jnettlet | exhausting. half a dozen over excited 40+ kg dogs that have no boundaries set. |
15:04 | hste | jnettlet: so its you that train the dog or the dog that trains you? :) |
15:05 | jnettlet | The pack really does most the training. I just lead the pack. |
15:09 | jnettlet | warped-rudi, have you gotten any numbers yet? |
15:11 | warped-rudi | Not yet. I just successfully built the executables. The MHz sound O.K. What tests do think are most interresting ? |
15:11 | rabeeh | warped-rudi: i'm back online; i missed few of your notes |
15:11 | jnettlet | I would just run the HARDWARE test group for starters |
15:12 | rabeeh | i actually saw the micro USB connector |
15:15 | warped-rudi | rabeeh: Hi ! Part of the problem is also, that the connector is so close to the SATA port. it's pretty much impossoble to plug both. |
15:16 | rabeeh | warped-rudi: i know; but there is no place else |
15:16 | rabeeh | literally.. no other place |
15:16 | rabeeh | we used to have it beneath the SPDIF connector; which is completely ugly |
15:17 | rabeeh | warped-rudi: any idea about boot time speeds between c1 and CBi? |
15:18 | warped-rudi | I know where it is on the old CuBox ;-) |
15:19 | warped-rudi | About the speed: I'm just trying to run some benchmarks. Will report when I have it ready... |
15:20 | jnettlet | rabeeh, I asked warped-rudi to install lm-bench to get some good comparitive benchmark results on his system. |
15:20 | rabeeh | for the bogomips; if you are using interactive profile you should see between 400 to 2k bogomips (equivalent to 2x the processor speed when measured) |
15:21 | rabeeh | ok. i doubt you will see anything on the processor; but processor to ddr might show things |
15:21 | rabeeh | (i.e. lat_mem_rd, or bw_mem) |
15:22 | jnettlet | Here are my stream bandwidth and latency results. http://fpaste.org/77668/39256052/ |
15:23 | jnettlet | warped-rudi, which reminds me. You should lock the cpu-frequency before running the tests. |
15:24 | jnettlet | we don't want to test the scaling governor's performance. |
15:24 | rabeeh | jnettlet: if you warm one of the processor then it will go to highest speed |
15:25 | rabeeh | but you can do 'echo performance > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor' |
15:25 | rabeeh | that will put the processor in 1ghz |
15:25 | jnettlet | yep |
15:28 | warped-rudi | I need a bit more time since it looks like I missed some parts in my build. |
15:33 | jnettlet | no hurray |
15:46 | jnettlet | no hurray??? make that hurry :-) |
15:49 | kivutar | is there somebody from solidrun on the chan? |
16:03 | kivutar | rabeeh, maybe? |
16:07 | jnettlet | kivutar, yes but as always with IRC people are on and off. You can ask your question and then hang out for an answer, but really emailing the SolidRun support email is probably a better way to get service. |
16:10 | dv__ | warped-rudi: you were the one who added libcec support for the original cubox, right? |
16:13 | warped-rudi | dv__: Yes, but some others helped as well. |
16:14 | dv_ | tried cec with cubox-i yet? |
16:15 | warped-rudi | Very little. I got a CuBox just recently and there are so many other things... |
16:15 | dv_ | oh, i can relate |
16:16 | kivutar | jnettlet, yes, it's because I emailed them 2 weeks ago about a partnership and got no reply yet |
16:16 | dv_ | todo list is huge |
16:16 | dv_ | kivutar: iirc the guys are overloaded with requests atm |
16:17 | kivutar | dv_, yeah I can understand that, the product is interesting |
17:22 | rabeeh | kivutar: here |
17:28 | kivutar | rabeeh, hello, I sent a mail on february 8 about a partnership but got no reply |
17:28 | rabeeh | kivutar: let me check |
17:28 | kivutar | we do a linux distro and would like to support cubox-i |
17:29 | kivutar | the subject is Retro Emulation on Cubox-i |
17:29 | rabeeh | nop; checked trash too :) |
17:29 | kivutar | oh, strange, let me resend it then |
17:30 | kivutar | rabeeh, [email protected]? |
17:30 | rabeeh | oh; that's not me then; that's sales :) |
17:30 | rabeeh | they have a long long long queue :) |
17:30 | rabeeh | sorry about that; which distro is that? |
17:30 | rabeeh | please to me directly - [email protected] |
17:31 | kivutar | this one https://github.com/Niouby/Lakka/wiki |
17:31 | kivutar | it's mostly a retroarch distro, build using openelec tools |
17:31 | rabeeh | oooh; nice :) |
17:31 | rabeeh | OE based |
17:32 | kivutar | yes I like their build system |
17:32 | kivutar | it's very easy to package this way |
17:32 | kivutar | I transfered the mail to your adress |
17:34 | rabeeh | kivutar sounds very interesting |
17:35 | rabeeh | which platform are you interested in? |
17:36 | rabeeh | I can queue you for any platform you want CuBox-i versions |
17:36 | rabeeh | HummingBoard is out of reach now |
17:37 | kivutar | rabeeh, are the GC880 and GC2000 very differents? |
17:38 | kivutar | I think the CuBox-i2Ultra could allow us to add psx and gba support |
17:39 | rabeeh | gc2k has quad shader while gc880 has one |
17:40 | rabeeh | do you run under X or fb? |
17:40 | kivutar | kms |
17:40 | rabeeh | how many devs are there? |
17:40 | kivutar | we do use pixel shaders |
17:41 | rabeeh | i can queue you for more machines if you want to test gc880 and gc2k |
17:41 | rabeeh | currently we are out of stock; but i can get you one immediately from my development machines |
17:41 | kivutar | 2 devs for the system, located in france, for them we need only one device |
17:42 | kivutar | it's great! it won't be a waste |
17:42 | rabeeh | hehe.. i'm sure |
17:42 | rabeeh | i'm betting on it :) |
17:42 | rabeeh | gba mainly :) |
17:43 | rabeeh | btw - which resolution do you typical render for? |
17:43 | kivutar | it would be interesting also to send one to the retroarch team |
17:43 | rabeeh | i mean all those devices has low resolution; what do you typically choose as an output from the ARM machine? |
17:43 | kivutar | our gl backend renders up to 1080p |
17:44 | rabeeh | so all upscaling done by the gles engine |
17:44 | kivutar | yes |
17:44 | rabeeh | i was wondering how that would be vs. the i.MX6 ipu upscaler, from quality point of view since gba has really low resolution |
17:44 | rabeeh | 240x160 looking at the wiki |
17:45 | rabeeh | hmm... thinking about that; the i.MX6 has IC unit internally that can do upscaling on the fly with those resolutions |
17:45 | kivutar | rabeeh, if you join #retroarch, Squarepusher is the one who knows the gl backend I think he could reply way better than me |
17:45 | rabeeh | so no ddr bandwidth is waste for mem2mem |
17:45 | rabeeh | ok. |
17:57 | warped-rudi | jnettlet: It took a while for both, me and the CuBox, but here are teh results: http://fpaste.org/77686/ |
18:12 | mk01 | warped_rudi: isn't mac80211 compiled in kernel doing constant regulatory updates? have seen this on default kernel even with wlan0 off taking ~30% cpu cycles |
18:17 | rabeeh | cut and paste from my last speech - |
18:17 | rabeeh | warped-rudi: so it is slower |
18:17 | rabeeh | are you running SPL u-boot? or older u-boot? |
18:17 | rabeeh | i wonder if we have missed part of the ddr configurations |
18:17 | rabeeh | i would say HB is 20% faster than i4pro; where it should be the way around (given 64bit DDR and higher speed) |
18:17 | rabeeh | are you running C1 with DDR overclocking? |
18:18 | rabeeh | great. now warped-rudi is not online; damn it with freenode |
18:19 | sickness | what's hb? |
18:20 | jnettlet | sickness, hummingboard |
18:21 | sickness | k |
19:21 | warped-rudi | rabeeh: i'm back |
19:25 | warped-rudi | And in case you read the logs later: It's running lastest SPL. But I tested with both an older u-boot and a debug version sent to me from jnettlet. All with the same result. |
19:26 | warped-rudi | I'm not overlocking it. Remember, that your overclocking patch did not work on this board... |
19:45 | ssvb | warped-rudi: did you run these benchmarks in a headless configuration or with monitor connected? |
19:46 | ssvb | warped-rudi: this normally makes a huge difference, especially with 1080p monitors and 32bpp color depth |
19:46 | warped-rudi | booted up with monitor connected but detached it before testing |
19:47 | warped-rudi | btw, I'm conducting these tests because I experience unusualy slowness of this device since day one |
19:49 | ssvb | this might be not enough, it's better to explicitly blank the screen to be sure that the framebuffer scanout is not stealing the memory bandwidth |
19:49 | ssvb | or run two sets of benchmarks (with and without monitor) and then compare results |
19:58 | warped-rudi | I can re-run the test with taking special care about this. Tomorrow... |
20:41 | pepedog | Well got xbmc running in arch, vids generally play great, but |
20:42 | pepedog | Destructive blinking cursor from cursor |
20:42 | pepedog | Keyboard and mouse only work for root |
20:43 | mk01 | pepedog: /dev/input chmod |
20:43 | mk01 | seterm -cursor off |
20:43 | pepedog | Thanks |
23:33 | pepedog | xbmc on arch, where would I put setterm -blink off |