Did Valve already get what they wanted from SteamOS? i.e. Win kernel + BigPicture DE

[Rich Oliver]

Unless the container tech can provide performant access to GPU, input devices, and sound devices, it's not going to work very well. It's not like a server chunk that you can plunk into Docker, after all.

Having said that...I don't know if any of the people doing work in this space actually have thought in those terms yet. If they have, then yeah, that's one of those good things to have in hand- it'd allow you to make a single title that plunks onto any of the given consoles as needed with minimal efforts to target several of them at the same time. I just don't think we're there yet.

That's disappointing. Lennart Pottering seemed to me to be hinting at the introduction of containerised apps as part of the systemd project. The other great hope is Wayland for compartmentalising and limiting applications.

I'm just an ordinary desktop user who does a bit of programming. I find the much vaunted stability of Linux to be almost entirely useless. When my X session locks up, I have to reboot. The big advantage of Linux is that its about ten times quicker booting up than Windows. Or at least my opensuse 13.2 is. Kubuntu 14.10 which of course is still using Upstart seems significantly slower than opensuse on boot up. I run three screens off an HD 7790. The AMD driver in the kubuntu 3.16 Kernel seems a significant improvement over the 3.13 Kernel of Opensuse. Things are definitely improving in the Linux desktop space: before about Kernel 3.10 I found the open-source AMD driver essentially unusable for my hardware set up.

[Luke]

All that consumer content has to be produced on something

I walk into most stores now and find nothing but tablets.

Speech to Text is the best ever! Laptops are a waste of money.

Huge PC's are only good for servers.

60'' TV + Chromecast + Tablet + Bluetooth keyboard = Programmer's dream

We're waking up and the companies are dying.

PS4 killed Microsoft with FreeBSD driving the nails home!

They'll have to drop the price another $100 (349, now $249) on the Xbox OnE!

Steam is where it's at. If Valve gets back in bed with Microsoft, it's their own mistake.

All that content (free or paid) that people consume on smartphones and tablets has to be created somewhere. People can and do watch my videos on phones, but I cannot produce them on one. Until a tablet can run Kdenlive or a similar FOSS program and edit 1080p HD video with composited effects (not just cutting files together without "playing" them), and render a video in less then 2-3x realtime, such a device cannot replace my desktop. Hell, my netbook can't replace it either for exactly the same reason. Games and HD video require a lot more CPU power to create than to consume, it's not only the servers they are fetched from that people want Xeons and Bulldozer/Piledriver for but also the production platforms. Just when we get 1080p in hand people now want 4K video, and anyone making movies for a theater actually needs that.

You will never see a smartphone with a pair of Xeons and a Xeon Phi coprocessor, and literally could not build one today for $1 million. To get to that point would require sub 1 nm fab processes-or else smartphones powered by heavy backpack batteries and equipped with external radiators that must be put into water for use over a certain threshold. That supertablet would be one hell of a machine, but it does not exist and might be 20 years away if it ever exists. That same $1M would buy a whole roomfull of workstations with the above specs right now, you might even be able to get them from a Phoronix advertiser if you can catch the right one.

A special purpose video editing tablet is actually possible. There are tablets now that can do some form of editing on their own video, but I do not know if that is multitracking, composition-capable video editing. I suspect it is not, as to render video at a transition means to play both source tracks, combine the results, and encoded the result, requiring at least three times the computing power of encoding one track from the tablet's camera or playing it on the screen. Yes, it could be programmed to run a video editor and simply take however long it takes to render the result, I have done that with 1080p video on Intel Atom in a pinch,

If someone produced a tablet combining hardware accelerated encode and the ability to "play" two videos at once on a multiple unit or multitasking hardware decoder, that would be quite another matter. That could handle video from any source whose codec it included, and do full nonlinear editing. Special software would be needed to do all playback and encoding in the GPU, and keep the data in GPU ram while CPU composition, etc is executed on it. An AMD APU might be able to do this now, a version of Kdenlive written for this setup would blow any multicore CPU with current Kdenlive out of the water. You still won't run Blender on such a video editing tablet and write games, however, and you will need to hook it to a larger monitor if you are doing a final edit for content aimed at movie or TV screens.

[Svartalf]

That's disappointing. Lennart Pottering seemed to me to be hinting at the introduction of containerised apps as part of the systemd project. The other great hope is Wayland for compartmentalising and limiting applications.

I'm just an ordinary desktop user who does a bit of programming. I find the much vaunted stability of Linux to be almost entirely useless. When my X session locks up, I have to reboot. The big advantage of Linux is that its about ten times quicker booting up than Windows. Or at least my opensuse 13.2 is. Kubuntu 14.10 which of course is still using Upstart seems significantly slower than opensuse on boot up. I run three screens off an HD 7790. The AMD driver in the kubuntu 3.16 Kernel seems a significant improvement over the 3.13 Kernel of Opensuse. Things are definitely improving in the Linux desktop space: before about Kernel 3.10 I found the open-source AMD driver essentially unusable for my hardware set up.

Anything Lennart suggests should be...

1. Taken with a 50# block of salt...
2. And viewed with trepidation out of box...

Do you trust the gentleman that got PulseAudio, Avahi, and a few other things all wrong and then effectively walked away from them? I don't. systemd's just like the story with Pulse...it's a solution looking for a problem to solve and it does it poorly with all sorts of feature creep (Containers? Seriously?) that you deeply and honestly don't want, no matter how shiny it might seem to you at the time.