| 00:30 | heap_ | ok, looks like cubox pro with FIS multiplier will perform so much better then cuboxi. https://www.sata-io.org/port-multipliers |
| 00:30 | heap_ | no idea why cubox i suppor the CBS multiplier, because purpose is for media only? |
| 00:48 | heap_ | rabeeh: ^^^? |
| 01:23 | cbxbiker61 | _heap, you have an application that requires more than 40M/sec? |
| 01:24 | cbxbiker61 | if you don't...your question is purely academic |
| 01:41 | heap_ | cbxbiker61: could you be more specific? |
| 01:41 | heap_ | cbxbiker61: im fine with 40m/sec read/write using LVM and lets say 3hdd drives |
| 01:42 | cbxbiker61 | then you have no problem |
| 01:43 | heap_ | cbxbiker61: im struggling with a compatibility answer... what cubox and what enclosure |
| 01:43 | heap_ | based on that FIS vs CBS |
| 01:44 | heap_ | another question is what is better cubox prov1 or cubox-i for nas |
| 01:45 | cbxbiker61 | technically the pro would be better, understand that nas type chips is marvell's bread and butter, whereas freescale is better in the graphics department |
| 01:46 | cbxbiker61 | so the marvell is better as a backend device and the freescale is better at the frontend |
| 01:46 | heap_ | cbxbiker61: correct |
| 01:46 | heap_ | cbxbiker61: but is one core enought? |
| 01:47 | cbxbiker61 | moving bits and bytes is not that much of a chore unless you're encrypting |
| 01:48 | heap_ | well and in case you want to use encrpytion on part of the datA? |
| 01:49 | cbxbiker61 | i wouldn't sweat it |
| 01:51 | heap_ | means it will work? |
| 01:51 | heap_ | im just worried why they dont sell old cubox anymore ;/ |
| 01:51 | cbxbiker61 | sure |
| 01:52 | cbxbiker61 | they don't have the resource to keep a lot of products that are similar going at the same time |
| 01:54 | cbxbiker61 | i have 3 or 4 of the original cubox's, i'm using one as a firewall, and shortly i'll be setting up a dedicated backup server with one attached to a jbod |
| 01:55 | cbxbiker61 | but, i could certainly substitute a cubox-i for any of those applications |
| 01:55 | heap_ | well i luckily got one too. |
| 01:55 | heap_ | cbxbiker61: but u said marell is batter for nas then freescale |
| 01:56 | cbxbiker61 | heap_ it's really academic, since for most practical purposes you wont know the difference |
| 01:57 | cbxbiker61 | the market is full of marvell dedicate nas boxes |
| 01:57 | cbxbiker61 | the market is full of marvell dedicated nas boxes |
| 02:00 | heap_ | ah oki |
| 02:00 | heap_ | synology or qnap use marvell ofthen |
| 02:00 | heap_ | often |
| 02:00 | heap_ | but none of them support LVM |
| 02:01 | cbxbiker61 | i'd probably just use btrfs rather than lvm, lvm is dead/dying |
| 02:02 | heap_ | but btrfs experiment, or not? |
| 02:02 | heap_ | hm so only bottleneck is ethernet which is 400mbit |
| 02:02 | cbxbiker61 | yeah, i've got 3 older 3TB drives i'll probably just span them with btrfs for kicks |
| 02:02 | cbxbiker61 | btrfs is fine with recent kernels |
| 02:03 | heap_ | when u plan to do it? |
| 02:03 | heap_ | and what enclosure u plan to use? |
| 02:03 | cbxbiker61 | maybe next week |
| 02:04 | heap_ | some performance result will be nice |
| 02:04 | heap_ | and what about enclosure? |
| 02:04 | cbxbiker61 | tr5m6g |
| 02:05 | heap_ | which suppor fis? |
| 02:05 | heap_ | it not possible to buy it here in europe |
| 02:05 | heap_ | better in czech/slovak |
| 02:09 | cbxbiker61 | the sans digitals are nice enclosures, maybe you could get one of their other models |
| 02:10 | heap_ | not possible here |
| 02:10 | heap_ | no seller |
| 02:11 | cbxbiker61 | easy swapping, the cheap enclosures have clumsy swap |
| 02:11 | cbxbiker61 | look at the reviews for whatever you get, and make sure people aren't complaining about swapping |
| 02:12 | heap_ | just wondering if there is any marvell with 1gbit eth + esata |
| 02:13 | cbxbiker61 | that's what the original cubox is |
| 02:15 | heap_ | cbxbiker61: oh then why you mentioned 40mb/s limit?\ |
| 02:20 | cbxbiker61 | that's about what the cubox-i will do, the original cubox is probably faster, but i haven't benchmarked it |
| 10:59 | topi` | is there OpenCV libraries available for i.MX6? |
| 10:59 | topi` | I'm trying to figure out if Hummingboard would be good enough for real-time video analysis (using a sufficiently small resolution video) |
| 11:57 | obinou | topi`: I ran OpenCV on imx6 a few years ago |
| 11:57 | obinou | (on a 4-core sabre-lite board) |
| 11:57 | obinou | So yes, it can be compiled, it even exists as a working package in yocto |
| 11:58 | obinou | concerning the performances , I think you'll have to try - it looks to me that it's highly dependant on the calculation yu do & the size of the images |
| 12:42 | jnettlet | topi`, the new V5 binaries are slightly better with OpenCL. They are only a bit faster, but they do add support for things like atomics |
| 14:21 | MikeSeth | hey all! |
| 14:55 | MarcusVinter | Hey |
| 15:00 | MarcusVinter | I have an idea to reduce the image size. Do you think it is possible to lower the u-boot resolution to 800x600 so that I can reduce the bmp size and have it fullscreen still? |
| 15:03 | MarcusVinter | I'm thinking changing CONFIG_ENV_SIZE and CONFIG_ENV_OFFSET and CONFIG_SYS_CBSIZE to 800 would do it but that's just a guess. Any insight from you guys? |
| 15:03 | MarcusVinter | I've got the demo this device in a few hours to a client and it would be nice to get the splash working in time. |
| 15:09 | Artox | MarcusVinter, wouldnt it be na alternative to just move the image to a regular filesystem on sdcard and load it from there? |
| 15:10 | Arto | 15:10 * Artox just wonders if there is a way not to mess with uboot at all |
| 15:10 | Artox | anotehr thing that might work: use uboot.img on first partition, instead of putting it on that weird sector 42 |
| 15:10 | Artox | would need somebody who knows something about SPL |
| 15:11 | Artox | it probably can only load from fat |
| 15:11 | Arto | 15:11 * Artox is just throwing in dreams |
| 15:15 | MarcusVinter | ^ That's a great idea. I just don't know how to accomplish it :P |
| 15:16 | vpeter | http://boundarydevices.com/displaying-a-splash-screen-with-u-boot/ |
| 15:16 | vpeter | But probably depends on a u-boot version. |
| 15:21 | MarcusVinter | Thanks vpeter :) |
| 15:22 | curlymo | if you want an animated splash, you can look as my implementation for XBian. Although it can only be started as early as the kernel is loaded. |
| 15:22 | MarcusVinter | I've reduced the bmp size to 800x600, left the u-boot resolution at 1024x760 and aligned the splash to the centre and it looks fine, and leave 4 lines at the bottom so I can see what's going on. Its not ideal but hopefully the entire image will be less than 1MB now and leave enough space for the kernel without overwriting it. |
| 15:23 | MarcusVinter | curlymo, That sounds ideal, I wouldn't mind an animated splash as the kernel loads too. Do you have a link to it on the forum or a blog or anything so I can check that out? |
| 15:24 | curlymo | this is how it looks (raspberry pi): https://www.youtube.com/watch?v=HKfc5_Hizj0 |
| 15:25 | curlymo | it has a progres bar, not visible on this video |
| 15:26 | MarcusVinter | curlymo, That looks great! How might I achieve something like this? |
| 15:27 | curlymo | by programming the framebuffer directly |
| 15:27 | curlymo | or by using my package ;) |
| 15:27 | MarcusVinter | Ouch that sounds painfully difficult. Where might I acquire your package? |
| 15:28 | curlymo | https://github.com/CurlyMoo/Splash |
| 15:28 | curlymo | has been around for quite some time. |
| 15:30 | curlymo | and if you do fixes, please let me know or do pull-requests |
| 15:30 | curlymo | we have the same splash running with XBian on the imx6 |
| 15:30 | MarcusVinter | Much appreciated! I'll check it out tomorrow when I get a bit more time and see if I can implement it. muchas gracias |
| 15:30 | MarcusVinter | Absolutely, thanks :D |
| 15:32 | curlymo | i asked our developer if can we have that xbian still in uboot as well (with that link vpeter posted) so we have the xbian logo there from beginning till end. |
| 15:33 | MarcusVinter | curlymo, Thats a good idea :P |
| 15:34 | curlymo | yes, thanks for the head start ;) |
| 15:36 | vpeter | What about having animated gif :) |
| 15:36 | curlymo | animated gif is an ass to work with in C |
| 15:36 | curlymo | just sequenced drawing of jpg / png is much easier |
| 15:36 | vpeter | animaated bmp then :-) |
| 15:37 | curlymo | looking forward to your additions ;) |
| 15:37 | vpeter | looking sources from previous link. |
| 16:02 | MarcusVinter | Well reducing the image size worked, I didn't overwrite the Kernel. |
| 16:02 | MarcusVinter | So thats nice. |
| 19:06 | jnettlet | MarcusVinter, okay I am available. Do you still have questions? |
| 19:08 | MarcusVinter | jnettlet, Hi, thanks for getting back to me. I ended up lowering the image resolution and it's turned out quite well and brought the u-boot size down enough so I don't overwrite the kernel :P I'm good for now i think :) |
| 19:09 | jnettlet | MarcusVinter, okay great. I think the best solution for you down the road is switching you to use directboot and then loading your splash screen under linux |
| 19:10 | jnettlet | you should also merge in the latest u-boot changes I just pushed as it cuts some time off the SPL boot time. |
| 19:10 | MarcusVinter | jnettlet, Yeah, I think that's a good idea. curlymo mentioned his kernel splash that I'm going to take a look at, I may even do both. |
| 19:10 | MarcusVinter | jnettlet, Oh great, I will. thanks :D |
| 19:11 | MarcusVinter | The kernel boots pretty fast at the moment anyway, I stripped out any bloating that we didn't need, you hardly notice the kernel load at all. |
| 19:12 | jnettlet | do you have a custom kernel commandline? |
| 19:13 | MarcusVinter | Not at the moment no :P |
| 19:15 | MarcusVinter | I think the idea of directboot with a kernel splash will be the way to go in the long run. Good idea. In the long run it should shave about 10 seconds off the boot time. |
| 19:16 | MarcusVinter | We currently have a bootdelay of 7 anyway so they get to see our splash. So yeah, directboot will shave the 7 seconds off plus the SPL update perhaps another seconds or two. |
| 19:16 | jnettlet | yeah, let your logo play over the kernel and init, so the machine is doing something while you are advertising |
| 19:22 | MarcusVinter | Yeah, that's ideal. It's not wasting any time then. I then eliminate the image size problem and can animate the boot logo :D Sounds great! |
| 19:23 | curlymo | animated boot splash can be done from an initram |
| 19:24 | curlymo | that's when we trigger it @XBian |
| 19:24 | MarcusVinter | Even if its not animated though, still a great idea though. |
| 20:32 | Exaga | is the solid-run.com site down??? |
| 20:33 | Exaga | i cannot access solid-run.com/wiki - it times out |
| 20:42 | vpeter | Exaga: works here |
| 20:43 | Exaga | it's up now |
| 20:43 | Exaga | must be my isp |
| 20:43 | vpeter | it's down for me now :) |
| 20:43 | Exaga | thanks vpeter |
| 20:43 | vpeter | http://www.downforeveryoneorjustme.com/ |
| 20:43 | Exaga | perhaps i spoke too soon |
| 20:44 | Exaga | it's down for me too now |
| 20:44 | vpeter | Yes :-) |
| 20:44 | vpeter | Under attack again maybe. |
| 20:45 | Exaga | oh it's up again |
| 20:48 | vpeter | Maybe someone is restoring something. |