${OS}Box - Wishlist

Since in this thread there was expressed the wish for some non-x86 hardware to run *BSD on, I think we should pool the ideas and wishes here so that it is a central point for people to look wishes up and maybe report on actual hardware found.

For starters, I would like the following things to be part of that HW:

• ECC RAM option
  May come later, but needs to be part of the path so no road blocks are build by accident
• SMP capable, maybe later on
  See above, you can mess up your system controller logic in so many ways
• Simple graphics for video, OpenGL maybe later
  For starters (and for servers) - who needs more than VGA B&W or maybe 256 colors?
• PCI / PCIe interfaces
  Expansion is crucial, see "lessions learned" from the good old PC
• USB
  Did I mention "cheap expansion"?
• Ethernet, 100MBit minimum
• Mass storage, SAS/SCSI/SATA

Nice to have would be

• Card reader to boot from
• OpenGL hardware
• multiple GBit interfaces
• Several CPUs with multiple cores (the Godson with 64 cores is in the queue, 4 cores are available)

Architecture (in my order of preference, not neccesarily your order, but let's discuss things):

• PowerPC/MIPS
  Close tie here, I like them both a lot and have worked with them. Both come with years, no - decades, of 64 bit experience under their belt. There should be no unwanted suprises or A20 gates around.
• SPARC
  Same pro as above, the programming is a bit of the "dude, wt..." kind untill you get your head around register windows. Cores available as VHDL (three implementations acording to wikipedia).
• ARM
  While this is a nice concept, I am not sure about the scaling when it comes to server grade systems. No proven 64 bit experience to my knowledge. Maybe covered by royalties and patents. Since there is some money in the attempt to push this into the servers, an independent attempt (like ours) may attract land sharks (a.k.a. lawyers. See "patent troll").

Form factor would be most likely ATX.

So, and to throw more than my hat in the ring, I'll do some things now to maybe get the ball rolling. I'll try to get some hardware designer to check his desk drawers for spare design parts for donation. In that case they (his company) would most likely do the manufacture,so they can also have their profit from it.

Ok, now let's pool ideas, requirements and ideas. Please keep in mind that there is a road map to be behind this and that there is a difference between "want" and "need". What would be a start point, what is the end goal? And yes, I also flunked that with the ECC.

Now let's hear what you need, want and are willing to do.

There are many system on module products at least for the ARM architecture CPUs. For example here and here.

Something like the Google Chromebook based on such module may be interesting - put the module with a SSD in the box, fill rest of the space with a battery packs, get let say 20 hours uptime? Or create board with several dozens sockets and an ethernet switch, put it in a 2 U box, boot the modules from the network and have about 100 cores available and so on.

Something based on this perhaps? https://www.olimex.com/Products/OLinuXi ... Xino-LIME/

I've been meaning to get one, mainly for use in the bedroom (XBMC) but I also want to try FreeBSD on it.

These ARM boards are all nice and well - but I for one would like to do a "make buildworld" in minutes, not days. But I think that that is also not the goal of these boards.

Edit: One reason against ARM I have, but it took me some time to realize what was nagging me. The lack of available, reliable 64 bit support will make it hard for ZFS. And that is one thing I think is important, more than the easily available SoCs (as nice as they are).

I suppose that one of the problems here is the different uses that people want to put these systems to. I had a look at the board SirDice recommended and I thought it looked excellent value for the money: they do a more powerful one for EUR 65 that would be great for a boat computer project that I have in mind. However, I can see that something along those lines is not, for most people, going to be an acceptable replacement for a more traditional desktop computer of some kind: you can't even buy a case for the more powerful Olimex board so it means either something home made or you have a very geeky-looking item sitting in your study.

In the thread referred to by OP, I was one of the people suggesting that it would be nice to have an alternative to the traditional Wintel platform, particularly with the issue of the new secure bootloader in mind. I was really thinking of an off-the-shelf type of solution: I suggested something like this http://www.tekmote.nl/epages/61504599.sf/en_GB/?ObjectPath=/Shops/61504599/Products/CFL-006 which costs about five times as much as the high-spec Olimex board, but includes such refinements as a hard disk and a case: it is actually quite an attractive looking little box. It certainly does not cost any more than an average PC. It is undoubtedly lower spec, but it seems likely that it will catch up. It is not powered by an ARM chip, it is Chinese and open source. My thoughts in suggesting this as a platform for FreeBSD were that it might make an acceptable home computer that would be as easy to set up and use as any other PC box. It obviously would have no relevance to the server market, and it would not suit people needing a decent amount of cpu grunt at home.

I think the idea of a machine that specifically supports FreeBSD is a nice one, but I would guess that the financial implications would pretty much rule it out before it even got under way. Maybe I'm being unduly pessimistic, but it can be expensive to produce anything in small numbers, production would have to be financed up front, and whoever foots that particular bill might just end up with a lot of worthless stock they cannot sell. The price would have to be at least reasonably competitive with a basic PC, or hardly anyone would buy it.

If this were a poll, my vote would be to try to improve support for some commercially available, not too expensive, alternative platform for home use. I realise this means spreading developer resources more thinly, but maybe it would have a benefit in terms of increased use and thus financial support for FreeBSD.

From a slightly different angle, it seems to me that there is a huge increase in the number of people putting together bespoke devices to do specific tasks (I know of lots of people who are trying to build things like navigation systems for boats using Raspberry Pis) and there is a whole new market for small computers running as home entertainment systems and it would be really nice if FreeBSD would run on at least some of these.

You see, ARM sells designs. One would need to invest into an ARM64 design. Now, if someone was to pay and have FreeBSD- along with OpenBSD and NetBSD- on the board and llvm/gcc/pcc support added, it would be nice.

I would love to have a play with an ARM system but when you look at operating systems to download and you get a page like: http://download.cyanogenmod.org

It puts me right off. Seriously? A bespoke build for every different kind of ARM device? It is ridiculous that there is no standard yet. Until I can boot a generic FreeBSD cd using a usb CDROM drive, ARM devices can GTFO

It is a shame because I am really interested in replacing my archaic home server (Thinkpad T23) with a low power and quiet ARM alternative. But until then, I cannot understand why Intel is having a hard time from ARM but I would be very careful if I was them about rocking the boat too much with UEFI and Secureboot.

For example, for the Intel Microsoft Surfaces, and the Intel Macs, much of the hardware options is set by the OS rather than what the BIOS used to have a menu for, giving the OS far too much power. In the hands of Microsoft and Apple, this probably will mean greater lock in. In the case of Surface, Secureboot seems to only be able to be disabled from within Windows 8 making the hardware useless if you do not agree with the Windows license.

I had a good look at the lemote some time ago, and I would have liked to get the new laptop from them - but I guess that it was too close to the macbook to be possible. That is sad. But I have toyed with the idea of getting one of these mini-'PC's and check them out in more detail. Lemote has some motherboards on their home page which I would like to examine in more detail, but I would need some time for that. The systems seem to use a standard uBoot loader so it should be not too hard to get FreeBSD/MIPS booting. Making it stable, that is another story.

And yes, untill ARM gets it's act together (as would others do) with regard to hardware setup/detection during boot, the ARM camp is going to be balkanized from an OS point of view.