00:48.32 | *** join/#elinux thraxisp (n=thraxisp@24.139.16.154) |
01:24.12 | *** join/#elinux archae0pteryx (n=snewman@207.47.42.130.static.nextweb.net) |
01:51.40 | *** join/#elinux linac (n=lin@122.90.120.255) |
02:05.46 | *** join/#elinux roxfan (i=dunno@254.115-242-81.adsl-dyn.isp.belgacom.be) |
03:36.38 | *** join/#elinux surjya (n=surjya@65.6.95.219.kmr01-home.tm.net.my) |
03:37.39 | surjya | Hi geeks |
03:38.02 | surjya | I need help regarding platform selection |
03:38.36 | surjya | I planning to develop a test gear which is capable of analysing IPTV,VOIP and simple data traffic |
03:39.00 | surjya | for that i will be designing a handheld device |
03:39.18 | MonMotha | off the shelf device or custom? |
03:39.20 | surjya | I need to know which processor will be suitable for my requirements |
03:39.58 | surjya | custom |
03:40.14 | MonMotha | any reason an off-the-shelf device won't work? |
03:44.54 | surjya | i need to put some extra features in that |
03:45.02 | surjya | sorry for late reply |
03:45.40 | MonMotha | well, then the question is what kind of performance, interfaces, and power requirements do you have/ |
03:46.01 | surjya | the power is 10W |
03:46.18 | surjya | it should process 1GigE IP packets |
03:46.21 | MonMotha | 10W? that's a LOT for a handheld |
03:46.46 | MonMotha | I take it this device will be plugged in during operation? |
03:46.47 | surjya | i m planning to make the device as 7x7" |
03:47.01 | surjya | an Touch screen display will be there |
03:47.09 | surjya | so i think 10W is needed |
03:47.09 | MonMotha | or will it have a really big battery? |
03:47.25 | MonMotha | I see, so most of your power budget is being blown on the display? |
03:47.41 | surjya | yea |
03:47.53 | surjya | but it should hav its own power source |
03:48.04 | surjya | yea. that a fact also |
03:48.26 | MonMotha | There are some Marvell ARM SoCs with built-in gigE and reasonable (though not great) performance. I don't know what kind of CPU horsepower you require. |
03:48.29 | surjya | actuall it will be a GPON testing device |
03:48.37 | MonMotha | also, you'll have to sacrifice your first born to get the datasheet |
03:48.40 | surjya | we need to test 1GigE gpon links |
03:49.07 | MonMotha | you might just consider using an Atom |
03:49.26 | surjya | In applications I want to put a VLC or equivalent media mplayer for iptv testing |
03:49.41 | surjya | yea... I am just planning for atom also |
03:49.45 | MonMotha | well, most embedded SoCs have video decode acceleration of some sort |
03:49.57 | surjya | but I dont know the cost difference between ARM and Atom |
03:50.08 | surjya | the overall platform cost |
03:50.10 | MonMotha | TI has their OMAP ARM+DSP series, Freescale has i.MX with some sort of hardware decode, Marvell has hardware decode |
03:50.15 | surjya | and technical challenges |
03:50.35 | MonMotha | well, the ARM SoCs are generally easier to design with, though you can get some Atom based system-on-modules that make things easier |
03:51.30 | MonMotha | check out the Radisys Z500 |
03:51.37 | surjya | I would just need the IPTV player, a voip call analyser and simple IP data debugging application like ping, traceroute |
03:51.41 | surjya | only these 3 |
03:51.57 | MonMotha | have you considered just using a tablet PC? |
03:52.08 | MonMotha | or a netbook |
03:52.46 | surjya | and on linux platform |
03:52.52 | MonMotha | right |
03:54.06 | MonMotha | I'd re-look at off-the-shelf stuff, but if you still want to go custom, TI OMAP or Intel Atom System-on-Module is probably your best bet |
03:54.09 | MonMotha | maybe Freescale i.MX |
03:56.23 | surjya | ok... thanks a lot |
03:56.32 | surjya | I am planning to go for atom |
03:56.46 | surjya | because lots of experts are suggesting for atom only |
03:57.15 | surjya | and now u also added |
03:57.25 | MonMotha | Atom is probably the most "familiar" from a software POV |
03:57.53 | MonMotha | though you'll probably want the GMA500 (Poulsbo) SCH chipset for HD video capability, and that will require some hackery |
03:57.55 | surjya | hmmm |
03:58.20 | surjya | hav you any idea how much cost it will come for whole complete device? |
03:58.26 | surjya | with my features |
03:58.31 | MonMotha | the 945 chipset is high power (like 5W!) and doesn't have accelerated video decode |
03:58.42 | MonMotha | I think Z500 modules are about $175 |
03:58.55 | surjya | and atom? |
03:58.58 | MonMotha | that includes CPU, chipset, RAM, gigE |
03:59.05 | MonMotha | that's an Atom based system-on-module |
03:59.19 | MonMotha | I doubt you have the volume to justify dealing with the bare chips |
03:59.23 | surjya | ok |
03:59.49 | surjya | i would need around 50 |
03:59.52 | MonMotha | the ARM SoCs are much easier to deal with, and you may be able to justify bare chips |
03:59.53 | surjya | initially |
04:00.03 | surjya | ok |
04:00.03 | MonMotha | yeah, 50 isn't enough to justify dealing with the bare CPU/chipset stuff for Atom |
04:00.19 | MonMotha | the layout alone will probably drive you nuts beyond justification |
04:00.33 | surjya | means after sales increase we may increase the volume |
04:00.54 | surjya | ohhh |
04:01.41 | MonMotha | at 50, you might even consider a pre-made CPU module+memory for the ARM based stuff. There are plenty out there. |
04:02.21 | surjya | but can arm support the GigE + a video player + call processing? |
04:02.37 | surjya | i mean processing power |
04:03.02 | MonMotha | yes, with hardware acceleration |
04:03.26 | surjya | ok |
04:03.49 | MonMotha | the only CPU intensive thing there is the video playback and maybe audio processing for the VOIP, and you can use DSP functions for that. ARM926EJ-S and the common ARM11 (ARM1176JZF, I think) both have some built in DSP, and lots of ARM SoCs are mated with video decoding hardware |
04:03.56 | surjya | I am very new to dealing this type of projects |
04:04.08 | MonMotha | I'm not sure if anybody other than Marvell has onboard gigE, though. 100Mb is common. |
04:04.11 | surjya | can plz guide me how to proceed with this type of project |
04:05.17 | MonMotha | well, I'd call up your sales reps at the aforementioned companies and ask them to come up with suitable products. That's their job. |
04:06.46 | surjya | ok |
04:17.05 | *** join/#elinux m4t (i=matt@theorize.org) |
04:17.21 | m4t | anyone here done work on a vr5500 based system? |
04:21.15 | MonMotha | m4t: I've looked at a DCT5000 before; IIRC that's VR5500 |
04:21.55 | m4t | dct5000 is that a board or a soc? |
04:22.14 | m4t | stb okay |
04:22.35 | m4t | yea. nec has been reviving the vr5500 in new multimedia soc's |
04:22.52 | MonMotha | interesting |
04:23.06 | m4t | i found an old nec rockhopper board w/ a vr5500 cpu card for $0.99 |
04:23.09 | m4t | unused apparently |
04:23.13 | MonMotha | at one point, there was interest in running Linux on the DCT5000, but I think that's gone away |
04:23.18 | m4t | still has the plastic cover on the lcd |
04:23.24 | MonMotha | I'd still like to do it some time, but it's low priority |
04:24.20 | m4t | im trying to re-add rockhopper board support (deprecated ~2.6.18) to 2.6.31+ (w/enhanced vr5500 support, for the new SoC) |
04:24.49 | m4t | so far it shows up in menuconfig and selects the proper cpu |
04:24.56 | MonMotha | maybe the DCT5000 is an even older VPU |
04:24.57 | MonMotha | *CPU |
04:24.57 | m4t | but, my toolchain is still compiling |
04:25.02 | MonMotha | IIRC, it never had 2.6 support |
04:25.08 | m4t | vr5500 is from 2001 |
04:25.13 | MonMotha | well, it's old enough |
04:25.38 | MonMotha | VR5432 |
04:25.43 | m4t | very similar |
04:25.52 | MonMotha | VR5476 northbridge |
04:26.26 | m4t | ah i see the elinux page now, heh |
04:26.33 | MonMotha | yup |
04:26.49 | MonMotha | if you can get the CPU support working again, I'd look at trying to get the kernel to boot on it |
04:27.04 | MonMotha | I kinda gave up when the CPU became unsupported |
04:27.40 | m4t | i got a mips stb i'm ~working on |
04:27.52 | m4t | rebranded tatung stb-2300 |
04:28.05 | m4t | drm has it mostly locked into wince |
04:28.15 | m4t | and my jtag wasn't detected at all |
04:28.19 | m4t | so i set it aside |
04:28.39 | MonMotha | from the looks of this device, all the set-top specific stuff just sits on the PCI bus |
04:28.57 | MonMotha | so it should be possible to at least boot linux on it and get video (graphics are ATI Rage type) |
04:29.06 | m4t | that'd be cool |
04:30.16 | MonMotha | indeed |
04:30.23 | MonMotha | get me CPU support, and I'll have at :) |
04:30.44 | MonMotha | at one point, I was also stumbling over the bootloader to use, but it looks like everybody is using u-boot now |
04:30.51 | m4t | the thing is, about 5432 support |
04:31.08 | m4t | the last rockhopper targets selected CPU_VR5432 |
04:31.36 | m4t | let me check here |
04:32.33 | MonMotha | I seem to recall CPU support I needed being removed from like 2.6.8 and apparently never working at all, but I could have been mistaken |
04:34.09 | m4t | CPU_R5432 is in 2.6.18 |
04:35.10 | m4t | <PROTECTED> |
04:35.10 | m4t | <PROTECTED> |
04:35.10 | m4t | <PROTECTED> |
04:35.12 | m4t | <PROTECTED> |
04:35.42 | m4t | 2.6.18 is 'current' w/debian patches too |
04:36.32 | m4t | its in 2.6.31.6+ too |
04:38.26 | m4t | the new commits for the VR5500 are from '09 |
04:43.05 | MonMotha | hum, I wonder if it got revived |
04:44.40 | MonMotha | is there VR5476 support? |
04:47.51 | *** join/#elinux adb (n=adb@212.147.109.90) |
06:53.34 | *** join/#elinux lyakh (n=lyakh@p57BD184F.dip0.t-ipconnect.de) |
07:37.22 | *** join/#elinux hw_ (n=hw@p578b3905.dip0.t-ipconnect.de) |
08:47.51 | *** join/#elinux tsjsieb (n=chatzill@dejongbeheer.nl) |
10:15.49 | *** join/#elinux markl_ (n=mark@tpsit.com) |
11:21.55 | *** join/#elinux pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
11:38.14 | *** join/#elinux eox425 (n=root@213.133.177.11) |
11:47.53 | *** join/#elinux pirho (i=pirho@gateway/gpg-tor/key-0x2CEEC9CB) |
12:07.56 | *** join/#elinux gustavoz (n=gustavoz@host248.201-252-31.telecom.net.ar) |
12:47.19 | *** join/#elinux dijenerate (n=dijenera@64.210.44.37) |
13:16.18 | *** join/#elinux flavioribeiro (n=flavio@187.64.55.50) |
15:57.10 | *** join/#elinux ZeZu (n=zezu@c-24-14-8-186.hsd1.in.comcast.net) |
16:47.57 | *** join/#elinux gustavoz (n=gustavoz@host248.201-252-31.telecom.net.ar) |
17:05.36 | *** join/#elinux VK4MSL (n=vk4msl@gentoo/developer/redhatter) |
17:08.21 | *** join/#elinux thraxisp (n=thraxisp@century.precidia.com) |
17:47.21 | *** join/#elinux toi (n=toi@d54C2A96D.access.telenet.be) |
17:52.10 | *** join/#elinux linac (n=lin@122.90.83.149) |
18:00.21 | *** join/#elinux thraxisp (n=thraxisp@century.precidia.com) |
18:31.46 | *** join/#elinux flavioribeiro (n=flavio@187.64.44.60) |
18:58.25 | *** join/#elinux archae0pteryx (n=snewman@207.47.42.130.static.nextweb.net) |
19:33.01 | *** join/#elinux toi (n=toi@d54C2A96D.access.telenet.be) |
19:33.15 | *** join/#elinux likewise (n=chatzill@82-171-51-231.ip.telfort.nl) |
20:10.58 | *** join/#elinux geep (n=JoeBelow@bas21-toronto12-1242364132.dsl.bell.ca) |
20:45.45 | *** join/#elinux markl_ (n=mark@tpsit.com) |
21:10.10 | *** join/#elinux thraxisp (n=thraxisp@24.139.16.154) |
21:11.20 | *** join/#elinux Greyscale (n=greyscal@ks35636.kimsufi.com) |
21:53.12 | gandhijee | hi, could someone please help me? i am trying to access GPIO pins on an intel based motherboard, but i cant seem to get to them, they are on a PCI device |
21:53.45 | gandhijee | and it shows up on the bus under d31:f0 according to the documentation, but i can't figure out how to get to the pins and toggle tem |
21:53.47 | gandhijee | *them |
21:54.11 | MonMotha | should be a pretty straightforward PCI driver hack-up |
21:54.23 | MonMotha | or are you wanting to do this from userspace? |
21:58.22 | gandhijee | MonMotha: user space, right now i'm just trying to toggle them from dos's debug |
21:59.40 | MonMotha | well, you'll have to figure out what I/O or memory regions it is mapped into |
21:59.57 | gandhijee | how would i go about figuring that out? |
21:59.59 | MonMotha | lspci will tell you those things, then you can use inb/outb or memmap /dev/mem |
22:00.12 | gandhijee | lspci will tell me? |
22:00.20 | MonMotha | from DOS, I have no idea. You'll have to set your BIOS up to do the PnP stuff for you. |
22:00.25 | gandhijee | right now i've been trying to just search for certain default strings |
22:00.25 | MonMotha | yeah, lspci -v |
22:00.43 | gandhijee | that will tell me what iospace's it using under linux? |
22:00.43 | MonMotha | <PROTECTED> |
22:00.43 | MonMotha | <PROTECTED> |
22:00.56 | gandhijee | got ya |
22:01.58 | MonMotha | IIRC, that I/O port should be accessable using inb/outb, inw/outw, etc., and the memory should be mappable via /dev/mem |
22:02.12 | MonMotha | you'll need to be root for the former, and the latter depends on the permissions of /dev/mem (usually limited to root for security reasons) |
22:02.29 | gandhijee | cool |
22:03.02 | gandhijee | i've done it using inb/outb, but that was for a winbond chip, so it was pretty simple |
22:03.54 | *** join/#elinux MonMotha (n=monmotha@cpe-65-29-71-209.indy.res.rr.com) |
22:47.14 | *** join/#elinux MonMotha (n=monmotha@cpe-65-29-71-209.indy.res.rr.com) |
22:49.00 | *** join/#elinux MonMotha (n=monmotha@cpe-65-29-71-209.indy.res.rr.com) |