TC for Raspberry Pi

[caminati]

The brand new project http://olimex.wordpress.com/2012/03/07/imx233-olinuxino-development-started-today/ looks like a serious one.

Also,

If there are interested software developers please contact us at www.olimex.com in 2 weeks we will have some limited number of hardware prototypes to ship to the interested to participate in the project development.

sounds like a call to developers of TC, which could be an appropriate distro, given the limited capabilities of the thingie.
What do people here think?

[curaga]

IMHO it's far too low-end, a few notches below the Pi even. It also seems to be targetting the Arduino market, instead of the htpc/small server/etc that I see the Pi as for.

Also the perf/$ is horrible, both by itself and compared to the Pi (which also isn't that great by that metric).


With 64mb of ram it's going to be tight for many uses.

[caminati]

IMHO it's far too low-end, a few notches below the Pi even. It also seems to be targetting the Arduino market, instead of the htpc/small server/etc that I see the Pi as for.

Yes, but demonstrating that TC can run on it would be a point alone.
If Arduino has market, this probably will, too. And, betting on its success, the linux distro being available for it could receive some wordspread.
Also, if you browse Raspi forums, you will find many people having in mind uses for which the iMX233 would be fit.

Also the perf/$ is horrible, both by itself and compared to the Pi (which also isn't that great by that metric).
With 64mb of ram it's going to be tight for many uses.

I think that below a certain price threshold, $ metric dominates perf/$.
Just curious: what are good pieces of hardware maximizing perf/$ metric you know of?

[curaga]

Yes, but demonstrating that TC can run on it would be a point alone.

It's an armv5, ~400Mhz, 64mb. AFAIK the only existing distro that does compatible binaries is Debian oldstable.
Again, setting the baseline to that would then penalize the Pi, or require two ports.

Merely running TC would take over half of that ram.

Of course, just because I don't see much point doesn't mean it won't happen if someone wants to make it so.

I think that below a certain price threshold, $ metric dominates perf/$.
Just curious: what are good pieces of hardware maximizing perf/$ metric you know of?

On the arm front, looking forward to the Rhombus Tech A10 board. On "existing ARM", the Pi is about the same price while being much better hw (not counting the huge Farnell margins).

In general, right now I'm running an AMD X6, I think it cost 120e. It's been very good for my uses, I can well make use of the six cores.

[caminati]

It's an armv5, ~400Mhz, 64mb. AFAIK the only existing distro that does compatible binaries is Debian oldstable.

Again, setting the baseline to that would then penalize the Pi, or require two ports.
On the arm front, looking forward to the Rhombus Tech A10 board. On "existing ARM", the Pi is about the same price while being much better hw (not counting the huge Farnell margins).

I'vn't understood yet when, where and at what price rpi will be available, then there is the latest ethernet port problem.
This is getting tiring, after months, so I'm turning my head elsewhere.
Rhombus-tech, too seem somehow stuck, at least taking the activity on news, pcb and bugtraking pages as an assessment. irc and mailing list are discussing about taking part to gsoc, sigh.
olinuxino guys are talking of a matter of days, let's see.
Moreover, this thing will probably be quite open: I understand that they're avoiding gpu-related closures by not having a gpu
I've taken a look at the SoC massive manual, and, if I'm not wrong, it will only have basic vesa modes.

Merely running TC would take over half of that ram.
Of course, just because I don't see much point doesn't mean it won't happen if someone wants to make it so.

I will try and see if I can get one.
My point in echoing that to the forum was that maybe if the TC developers contacted the olimex guys, they could have got an alpha unit.
However, I realize that would mean a lot of work, so maybe it's best to wait the more powerful realizations.

In general, right now I'm running an AMD X6, I think it cost 120e. It's been very good for my uses, I can well make use of the six cores.

Too powerful, there's no fun running TC on it

PS: Maybe this topic should be moved to another section of the forum? I fail to see how it belongs here.

[curaga]

Too powerful, there's no fun running TC on it

On the contrary I don't believe in hardware increasing and the experience staying the same. Both hardware and software should get better. With sw such as TC you can see the potential of any hw better than running something heavy; this still stands even for powerful hw.

I think I've mentioned this many times, but getting used to extremely fast responses is a nice feeling. Then when you occasionally see much better hw, but with worse sw, you notice how it lags in every operation.

PS: Maybe this topic should be moved to another section of the forum? I fail to see how it belongs here.

Agreed, and moved.

[caminati]

On the contrary I don't believe in hardware increasing and the experience staying the same. Both hardware and software should get better. With sw such as TC you can see the potential of any hw better than running something heavy; this still stands even for powerful hw.

I think I've mentioned this many times, but getting used to extremely fast responses is a nice feeling. Then when you occasionally see much better hw, but with worse sw, you notice how it lags in every operation.

I agree, things working well is a form of beauty
However, it is quite common and plain to see that having more and more powerful hardware resulted in horrible and bloated software, which kills any joy of the progress.
On the other hand, dealing with limited resources often results in coders producing real gems.

Getting back to the topic:

The following olinuxino README update excerpt might be relevant to this discussion, giving some answer about the popint in manufacturing something less performant than raspi:

People keep comparing OLINUXINO with Raspberry Pi and BeagleBone project, so we would like to state the differences here:

Although the projects are similar there are different goals and implementations:

OLINUXINO is completely open source - including hardware and software, this means you have access to all CAD files and sources and you can reuse them for your own personal or commercial project.
There are NO restrictions to manufacture and sell these board for your own use or resale.
OLINUXINO use widely available microcontroller iMX233 which cost USD 5.50 in 100 pcs quantity, this means that people can spin off their own boards and manufacture them cheap as the processor is in TQFP easy to assembly by hobbyist package.

RASPBERRY PI have no released CAD files nor complete schematics, RPi uses processor from Broadcom which is not available for sale in small quantity, it uses BGA package which require expensive setup to assembly.
RPi is designed to be home gadget, OLINUXINO will work in industrial environment -25+85C and will be designed to be low cost but NOISE immune.

BeagleBone have open source CAD hardware files but uses BGA processor and BB board is very complex and hard to manufacture in small quantities.

OLINUXINO uses processor on 454Mhz and have less memory and will not allow fancy graphics, but this is not our intention.

I especially like the focus on openness and simplicity.

[caminati]

Interesting update on olinuxino:
 http://olimex.wordpress.com/2012/04/24/cortex-a8-in-tqfp-sure-allwinner-a13/

Boiled down:

• There is bulgarian company Olimex which have been very effective in delivering the first ten alpha iMX233-based boards, with linux kernel and busybox already booting on it (http://tech.groups.yahoo.com/group/olinuxino/message/187) and a final price already published on their page (http://www.olimex.com/dev/pricelist.html). Small talking, big facts.
  If they can hold this attitude with respect to the allwinner a13 (see the first link of this post), things will get very interesting.
  
• Then there are rhombustech/arm-netbook people, who are lost in discussion, with very few concrete (I mean, hardware) achievement. However, their contribution in knowledge dissemination and interest triggering has been fundamental, even with respect to olimex.
  
  I myself am planning to follow their information to have linux on my AllwinnerA10-based tablet, whose hardware seems quite good up to now. Sadly, I have no time at the moment.
  
• Finally, there is an olinuxino rival project at http://www.ngcoders.com/category/projects/locux-projects. They seem to have achieved the alpha stage, however they lack in communication and I won't bet about them getting there.