Steam Machines?

SEANMCAD · May 2016

So I was terribly wrong about Steam Machines. It appears as of now to be a rather large scale failure namely because of performance issues.

That said, If Vulkan proves to be better than DX12 could that bring Steam Machines back from the dead? I am thinking no but maybe I am wrong again?

thoughts on a Steam Machine revival because of Vulkan?

Ridelynn · May 2016

I don't think Steam Machine's success hinges on Vulkan at all. The success of Steam Machine rides entirely on the success of SteamOS. Right now, people are wondering - why buy this machine when I'm just going to have to go install Windows on it anyway...

If SteamOS starts to see some adoption, vendors will start to see Steam Machine sales move. For SteamOS to start to see adoption, it needs a few key items:

Very regular development and active release schedule. Every time there is a major patch, you get some press, and if you are fixing more things than you can break, even better. Since it's introduction in 2013, we've seen 2 official releases.... there should be something "major, new, and great" every few months, not every few years.

Drivers Drivers Drivers. This has always been the Achilles Heel of Linux gaming.Valve has enough muscle and resources to be able to push back on this a lot more than a few open source purists sitting in their cubicle can. This may mean that you support a very limited hardware base - and that's not such a bad problem. Just get hardware vendors to slap "Ready for Steam" on the boxes of hardware you do support, and it's a marketing win. Just make damned sure that what hardware you do support, works very well. People expect that you plug in a controller - it works. You plug in your speakers, you get sound, etc.

An easy adoption method. Make it easy to coexist with Windows. Right now, the latest image nukes any previous installations of anything. But if it's painless to install and as easy as flipping a switch to boot back and forth, you may see some people try it out of sheer curiosity. Without that - you just have the geeks.

SteamOS already has deep Steam integration, so there is already an ecosystem ready to go with it. I think game support for Linux is better than it's ever been - that's not holding SteamOS back. Right now, I really think Valve is the only thing holding back SteamOS.

You don't necessarily need to benchmark faster than Windows, but you should be close. You do need to be more convenient than Windows - Valve should be jumping all over the Windows 10 issues right now, showing how SteamOS is different and better than Windows, and hey, it's free.

They really should be billing this, in my opinion, as "Turn your PC into a Console" and bring out all the latest console features -- one button live streaming, integrated party management and voice chat, etc. Most of that stuff already exists in Steam as it is, they just need to do a better job of getting it out in front. Make sure it's as prominent and easy to use as it is on a console. Make it as easy to load and play games as a console - Big Screen Mode is pretty darn close to that already. This should be the OS by gamers, for gamers, with everything optimized for gaming in mind.

Right now, it just kinda exists. But at this point, I don't think Valve is very serious about the effort. It's like they are half heartedly supporting their project, and just hoping it picks up steam (hah) magically on it's own on the fact that it's not Microsoft. Which I think dooms it to obscurity.

I tried 1.0, it looked ok on the surface, but at the time the video drivers didn't support my AMD, so it didn't really run anything. I haven't tried it since I've got an nVidia or the 2.0 release though, and the only reason is that I don't feel like manually going in and juggling my hard drives around.

zaberfangx · May 2016

SEANMCAD said:

So I was terribly wrong about Steam Machines. It appears as of now to be a rather large scale failure namely because of performance issues.

That said, If Vulkan proves to be better than DX12 could that bring Steam Machines back from the dead? I am thinking no but maybe I am wrong again?

thoughts on a Steam Machine revival because of Vulkan?

Steam Machines not far from the dead is not as big it should be with many factor in play. As it still can grow but with Vulkan will help the performance side of thing. But people use there pc more then just gaming will still a hard catch for some people.

Back to Vulkan even if is the same performance or little lower then DX12 with alot big things over DX12 as only runs with windows only of things. With Vulkan got a upper hand on that when it runs a lot more OS then just windows.

It will take Valve sometime for SteamOS to grow with games getting ported over from windows to SteamOS a lot more, but the idea of steam machine in general is not dead just harder to game with Xbox controller when there tons of games don't work for it. Steam Controller help with some even if it's hated by people who are so use to 1 controller.

In my thinking what will help SteamOS/Steam Machines more in the future is that making Steam Link 2.0 with better cpu/gpu to take on higher better games with out digger to much in to people pockets over time then just being a stream box. The more people get use to the OS the better.

Cleffy · May 2016

The functionality of Steam Link is most important for me. I really don't care too much to have a dedicated Steam OS since most of the functionality on my computer relies on Windows. Steam Link is an appealing offer for people who don't want to game next to their loud desktop that has an obscene amount of cooling. Its the most cost effective and seamless way to stream a game across the home. I just wish it could act more like a remote desktop so I can put my machine far away.
Whoops, just found out you can... Only problem is I have 2-4k displays.

SomethingUnusual · May 2016

Ridelynn said:

Drivers Drivers Drivers. This has always been the Achilles Heel of Linux gaming.Valve has enough muscle and resources to be able to push back on this a lot more than a few open source purists sitting in their cubicle can. This may mean that you support a very limited hardware base - and that's not such a bad problem. Just get hardware vendors to slap "Ready for Steam" on the boxes of hardware you do support, and it's a marketing win. Just make damned sure that what hardware you do support, works very well. People expect that you plug in a controller - it works. You plug in your speakers, you get sound, etc.

-Snipped for length/pointers

True, drivers for certain hardware prove to be a big problem with Linux gaming, but there is a considerable bit more.

Not all of Linux is the same, sure the kernel, which even then can be different (Modular kernel configuration, Monolithic kernel configuration) not to mention the init system such as systemd and openRC. Which makes the operating systems extremely different from one another. SteamOS does give an opportunity and part of its core goal to centralize this by restricting to a single standard of libraries, init system, and kernel.

The issue for developers will also be the tangle and vastness of different system configurations, there is no single standard, unlike that of Windows, or MacOS which are always the same. Drivers can be bypassed, having to write in a million symlinks cannot. In order for a developer to properly make games for a Linux system they would need to accommodate to every single standard, or at least as many as possible. This is a huge roadblock just considering the number of X Window systems and variants. X11? Wayland? Mir? Then comes the managers which have their own sets of variables... Gnome? XFCE? LXDE? KDE? OpenBox? The lists go on and on and on.

Find a way around these issues, such as SteamOS aiming for a direct set of standards, (one window system, one manager, one init system, one library configuration, one kernel) and the situation becomes significantly more viable.

Nevertheless, I think the issue that most seem to look at is "marketshare" kind of hard to judge something that is free in this department, the user circle is pretty huge for linux but there isn't much money floating around other than those (usually businesses) paying for tech support. Marketshare isn't a fair judgment, and neither would Steam Machine sales -- you can build your own pc and install the operating system [SteamOS] yourself, even dual boot your current windows machine if you know a little about configuring disks and a bootloader (GRUB, GRUB2, LILO etc).

SEANMCAD · May 2016

To be clear the reason I say Vulkan is that currently all games run slower on a Steam Machine compared to a Windows machine of the same spec.

However, Valve has sunk in tons of money, attention, and work into solving this problem via Vulkan (from my understanding).

Ridelynn · May 2016

Cleffy said:

The functionality of Steam Link is most important for me. I really don't care too much to have a dedicated Steam OS since most of the functionality on my computer relies on Windows. Steam Link is an appealing offer for people who don't want to game next to their loud desktop that has an obscene amount of cooling. Its the most cost effective and seamless way to stream a game across the home. I just wish it could act more like a remote desktop so I can put my machine far away.
Whoops, just found out you can... Only problem is I have 2-4k displays.

I really really like Steam In-Home Streaming. The best part about it, it doesn't ~require~ Steam Link hardware - just two machines with Steam installed, and that can be on any OS.

Valve may have a niche there just providing SteamOS for older hardware which essentially turns it into a Steam Link machine. I run Steam OS X on a 2010 Mac Mini with horrible nVidia 320M graphics, and it's able to push 1080 streaming without any issues at all. But try to run anything locally, and ... good luck.

Fortunately, Apple is pretty good about OS upgrades on older hardware, but had that been a PC stuck with something like XP, that makes a "free" OS that is maintaining security and driver updates a whole lot more appealing.

Ridelynn · May 2016

SEANMCAD said:

To be clear the reason I say Vulkan is that currently all games run slower on a Steam Machine compared to a Windows machine of the same spec.

However, Valve has sunk in tons of money, attention, and work into solving this problem via Vulkan (from my understanding).

Vulkan certainly won't hurt, but it isn't the panacea for anything.

SteamOS doesn't have to outperform or even equal Windows. It just needs to be "close enough" for performance. Sure, the purists will bicker over 98 vs 92 FPS, but most people won't care, so long as it brings convenience features that Windows or OS X can't offer, or offer poorly.

SEANMCAD · May 2016

Ridelynn said:

SEANMCAD said:

To be clear the reason I say Vulkan is that currently all games run slower on a Steam Machine compared to a Windows machine of the same spec.

However, Valve has sunk in tons of money, attention, and work into solving this problem via Vulkan (from my understanding).

Vulkan certainly won't hurt, but it isn't the panacea for anything.

SteamOS doesn't have to outperform or even equal Windows. It just needs to be "close enough" for performance. Sure, the purists will bicker over 98 vs 92 FPS, but most people won't care, so long as it brings convenience features that Windows or OS X can't offer, or offer poorly.

I think the bottom line for Linux and for that matter Steam Machine is the performance problem.

is there any other thing that is trying to address that on Linux? as far as I know the only one is Vulkan.

If the performance issue was not of top concern I think by now (some 25 years later) we would have seen more games in Linux before Valve steped in. In fact I think I have a speech given by Gabe Newel that said the performance problem was the most important issue regarding Linux in gaming (I can look it up if you like and if I have time()

SomethingUnusual · May 2016

Vulkan is a decent way to go, it's a new graphics, input, and audio api that takes OpenGL 6.2 a step further.
DirectX has had it's day with Microsoft platforms, Essentially all Vulkan is, is an API like DirectX for all platforms (Windows, Linux, MacOS, BSD).
Prior to Vulkan we only had OpenGL which is only a graphics API, for other tasks (input, audio, etc) other API's had to be used, such as freeGLUT, OpenAL, SDL, SDL2 etc.

I suspect Vulkan will be the end all beat all solution for future Linux, BSD, and Mac gaming. Cut's out all the middle men.

I have my doubts though that any games have been ported to the Vulkan API, likely the performance difference you are seeing is the port from DirectX to OpenGL/SDL2

SomethingUnusual · May 2016

Sorry to double post, but back to the drivers thing, in the Linux world, NVidia is horrible and doesn't play nice. AMD is the way to go as FGLRX driver (official Linux drivers for ati/amd hardware) is extremely stable, x11 drivers play nice with them as well (generic drivers) Nvidia, not so much. They have an official but it's poorly supported. X11 drivers also work but aren't as stable as the counterparts.

gervaise1 · May 2016

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

SEANMCAD · May 2016

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

Ridelynn · May 2016

Vulkan - developed and maintained by a non-profit committee ( Khronos Group )
DirectX - developed and maintained by a for-profit company

When has anything lead by committee beaten out a free market solution driven by profit. Ayn Rand would approve of DirectX.

You could also replace Vulkan/DirectX with Linux/Windows and it is pretty much the same thing, unfortunately.

SEANMCAD · May 2016

Ridelynn said:

Vulkan - developed and maintained by a non-profit committee ( Khronos Group )
DirectX - developed and maintained by a for-profit company

When has anything lead by committee beaten out a free market solution driven by profit. Ayn Rand would approve of DirectX.

You could also replace Vulkan/DirectX with Linux/Windows and it is pretty much the same thing, unfortunately.

and that has what to do with small team vs 'going it alone'?
oh...nothing

I dont mind people thinking that crowd working is all a pile of horseshit but I do mind people calling that 'going it alone'

gervaise1 · May 2016

SEANMCAD said:

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

AMD's aim / hope was that the Mantle API would - as a result of offering various benefits - become a "go to" API, followed by an SDK for other hardware etc. Big win for AMD if they pulled it off obviously.

About a year ago however AMD changed tack. It was clear that DX12 would also have these advantages - spurred on by Mantle. They abandoned the SDK, changed development plans and advised developers accordingly. They essentially joined the DX12 "consortium" - when I said team I meant "group of companies" rather than a group of people. That said Mantle was deemed a "win" for AMD.

The other thing coming out of AMD's change of direction was Vulkan, basically a derivative or evolved Mantle.

So - based on AMD's actions and what they advised developers to do - DX12 was "as good" as Mantle. (There were a few Mantle features that AMD suggested that developers do but basically it was DX12.)

Hence the answer to your question will Vulkan give Steam Machines a big edge? Well AMD didn't think. If they did there would have been no Vulkan they would have pushed ahead with a Mantle SDK.

SEANMCAD · May 2016

gervaise1 said:

SEANMCAD said:

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

AMD's aim / hope was that the Mantle API would - as a result of offering various benefits - become a "go to" API, followed by an SDK for other hardware etc. Big win for AMD if they pulled it off obviously.

About a year ago however AMD changed tack. It was clear that DX12 would also have these advantages - spurred on by Mantle. They abandoned the SDK, changed development plans and advised developers accordingly. They essentially joined the DX12 "consortium" - when I said team I meant "group of companies" rather than a group of people. That said Mantle was deemed a "win" for AMD.

The other thing coming out of AMD's change of direction was Vulkan, basically a derivative or evolved Mantle.

So - based on AMD's actions and what they advised developers to do - DX12 was "as good" as Mantle. (There were a few Mantle features that AMD suggested that developers do but basically it was DX12.)

Hence the answer to your question will Vulkan give Steam Machines a big edge? Well AMD didn't think. If they did there would have been no Vulkan they would have pushed ahead with a Mantle SDK.

sorry to sound confused but can you break it down in one sentence what 'go out on their own' means

Ridelynn · May 2016

I think AMD's selection as the CPU/GPU in XBox One had a lot to do with that as well.

I don't know that AMD has a lot of input into DX12. Mantle obviously was a big influence, but that's different than actually being at the table making decisions.

gervaise1 · May 2016

SEANMCAD said:

gervaise1 said:

SEANMCAD said:

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

AMD's aim / hope was that the Mantle API would - as a result of offering various benefits - become a "go to" API, followed by an SDK for other hardware etc. Big win for AMD if they pulled it off obviously.

About a year ago however AMD changed tack. It was clear that DX12 would also have these advantages - spurred on by Mantle. They abandoned the SDK, changed development plans and advised developers accordingly. They essentially joined the DX12 "consortium" - when I said team I meant "group of companies" rather than a group of people. That said Mantle was deemed a "win" for AMD.

The other thing coming out of AMD's change of direction was Vulkan, basically a derivative or evolved Mantle.

So - based on AMD's actions and what they advised developers to do - DX12 was "as good" as Mantle. (There were a few Mantle features that AMD suggested that developers do but basically it was DX12.)

Hence the answer to your question will Vulkan give Steam Machines a big edge? Well AMD didn't think. If they did there would have been no Vulkan they would have pushed ahead with a Mantle SDK.

sorry to sound confused but can you break it down in one sentence what 'go out on their own' means

AMD wanted to rule the world with Mantle.

SEANMCAD · May 2016

gervaise1 said:

SEANMCAD said:

gervaise1 said:

SEANMCAD said:

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

AMD's aim / hope was that the Mantle API would - as a result of offering various benefits - become a "go to" API, followed by an SDK for other hardware etc. Big win for AMD if they pulled it off obviously.

About a year ago however AMD changed tack. It was clear that DX12 would also have these advantages - spurred on by Mantle. They abandoned the SDK, changed development plans and advised developers accordingly. They essentially joined the DX12 "consortium" - when I said team I meant "group of companies" rather than a group of people. That said Mantle was deemed a "win" for AMD.

The other thing coming out of AMD's change of direction was Vulkan, basically a derivative or evolved Mantle.

So - based on AMD's actions and what they advised developers to do - DX12 was "as good" as Mantle. (There were a few Mantle features that AMD suggested that developers do but basically it was DX12.)

Hence the answer to your question will Vulkan give Steam Machines a big edge? Well AMD didn't think. If they did there would have been no Vulkan they would have pushed ahead with a Mantle SDK.

sorry to sound confused but can you break it down in one sentence what 'go out on their own' means

AMD wanted to rule the world with Mantle.

much better .
I think however its more:
AMD wanted to take the rulership of the world away from those who currently ruled it and rule it themselves with mantel. Its not like the world is not already ruled.

having said that, they were FAR from 'alone' in that

gervaise1 · May 2016

Ridelynn said:

I think AMD's selection as the CPU/GPU in XBox One had a lot to do with that as well.

I don't know that AMD has a lot of input into DX12. Mantle obviously was a big influence, but that's different than actually being at the table making decisions.

Don't disagree with this. However AMD do have a seat at the DX table! Basically they were playing two horses. Whether there were "Chinese walls" involved - no idea.

And they scored a win (early last year) when it was announced that stuff developed using Mantle could "easily" be ported to DX12 - which at the time wasn't fully rolled out. And as Mantle is embedded in AMD's drivers games would be more likely to run better on AMD cards - so better reviews, higher scores, higher sales etc.

Cleffy · May 2016

Actually, a lot of reason why AMD has such a huge performance advantage in DX12 is because they implement all the features Microsoft tells them to. What will be in DX12 has been known to nVidia and AMD for 5 years or so. Easily enough time to implement all its specifications. Yet once again only AMD implemented all the specifications as they did in DX10 and DX11. NVidia felt it more important to have the lead the last 3 years. So as a result nVidia cards simply don't support DX12 to the fullest until they release a new architecture as more DX12 titles are released.

zaberfangx · May 2016

Cleffy said:

Actually, a lot of reason why AMD has such a huge performance advantage in DX12 is because they implement all the features Microsoft tells them to. What will be in DX12 has been known to nVidia and AMD for 5 years or so. Easily enough time to implement all its specifications. Yet once again only AMD implemented all the specifications as they did in DX10 and DX11. NVidia felt it more important to have the lead the last 3 years. So as a result nVidia cards simply don't support DX12 to the fullest until they release a new architecture as more DX12 titles are released.

Was async compute why AMD was head of Nvidia with DX12 games. Nvidia did some work around with the cards that are now out now. And now with Nvidia new cards that are coming out have better support for async compute. Just AMD was a head of the game took the risk it work out for them a bit.

When it comes down to Vulkan Nvidia been on the ball updating there drivers for it, when AMD lags behind a bit.

Malabooga · May 2016

gervaise1 said:

SEANMCAD said:

gervaise1 said:

SEANMCAD said:

gervaise1 said:

Doubt it.

Vulkan has been created out of the Mantle - which AMD made available. (Some business elements they kept in-house.)

DX12 has also absorbed Mantle - AMD having decided it was better to stick with team DX rather than go it alone. (Developers, as mentioned, like conformity).

Steam Machines issues are much more serious imo.

wait...what do you mean by 'going it alone'?

the Vulkan team is in essence larger than the DX team

AMD's aim / hope was that the Mantle API would - as a result of offering various benefits - become a "go to" API, followed by an SDK for other hardware etc. Big win for AMD if they pulled it off obviously.

About a year ago however AMD changed tack. It was clear that DX12 would also have these advantages - spurred on by Mantle. They abandoned the SDK, changed development plans and advised developers accordingly. They essentially joined the DX12 "consortium" - when I said team I meant "group of companies" rather than a group of people. That said Mantle was deemed a "win" for AMD.

The other thing coming out of AMD's change of direction was Vulkan, basically a derivative or evolved Mantle.

So - based on AMD's actions and what they advised developers to do - DX12 was "as good" as Mantle. (There were a few Mantle features that AMD suggested that developers do but basically it was DX12.)

Hence the answer to your question will Vulkan give Steam Machines a big edge? Well AMD didn't think. If they did there would have been no Vulkan they would have pushed ahead with a Mantle SDK.

sorry to sound confused but can you break it down in one sentence what 'go out on their own' means

AMD wanted to rule the world with Mantle.

You got it wrong. When they started they offered NVidia so they both could develop API that uses maximum out of hardware (similar to consoles) and NVidia refused. Couple of years later NVidia hopped on Vulkan train wholeheartedly (and its just evolved Mantle)

And NVidia boasts with "we have worked with MS on DX12 for years"

The way Mantle ended up is huge win for AMD because that is what they wanted in the first place - and nobody would listen. If they did we would have much superior API much sooner.

Howdy, Stranger!

Steam Machines?

Comments

Howdy, Stranger!

Quick Links

Steam Machines?

Comments