Sony Is Working On AMD Ryzen LLVM Compiler Improvements - Possibly For The PlayStation 5

[Michael_S]

As to how Valve could have solved their problem, Starshipeleven and I quite clearly laid it out for you, kid, but here since your reading comprehension is so limited with you being in 4th grade and all, I'll put it in bullet points for you to make it simple to read. Because here's the protip... the Steam Machines didn't need to fail, they failed because Valve is Valve.

• Instead of having people just work on it in their free time when they feel like it since there's no hierarchy at Valve, actually hire an entire division of people to actually work on Steam for Linux and SteamOS as their sole purpose.
• Instead of assuming everyone else was going to do their work for them, actually leverage their vendor relationships to pursue their effort to these ends
• Actually participate in developing efforts like Wayland and actually engage the community in general
• Pay for some actual advertisement off of the steam platform
• Have Gamestop actually put out display units for people to play on
• Have general stores Target, Walmart, etc carry and have display units

I agree with all of your points, but I think even with all of that proper investment it wouldn't have been good enough to be viable by the planned launch date. Google, with its massive mountain of resources, didn't progress Android into "not suck" territory until somewhere in the 4.x release series and they were only supporting applications on APIs they controlled. Mozilla poured many tens of millions of dollars and worker hours into Firefox OS and it was stillborn (though additional external factors were at work there). Canonical put, what, five years into Ubuntu Touch and it still wasn't good enough for anything when they gave up.

But no, no kid... obviously the correct answer for Valve was to sit on their hands, and focus on the steam controller and VR, and work on Vulkan but almost completely ignore the SteamOS part of the equation until they finally really just abandoned the Steam Machine project altogether.

Also kid, if you think Valve didn't think it was ready or didn't think it was worth putting in the effort... riddle me this... why did they launch actual hardware at all? They could have waited till it was actually ready or just silently killed the hardware if they didn't want the investment. Instead kid, it was all quarter assing all the way for one actual reason: it isn't sexy to work on. Valve developers basically have free choice in what they want to work on... if your choice is between OS development, Client Development, VR development, Controller Development, graphics API development (Vulkan), or games... guess which 2 are going to largely go ignored. That is why it failed.

It's not "Oh they need a decade to work on this" it's "They sat on their hands, they launched early, and then they didn't even follow through and then forgot about it"

Again, agreed. But I think even with proper investment they would have needed until 2017 or even now to have a genuinely excellent product instead of being another five years away from it.

I hadn't really thought of Valve's free for all development model from the angle you presented. It makes sense. Most engineers have a few favorite areas to work on, and even though we generally recognize that all aspects of the product are critical it takes a little thumb-twisting from leadership to get us to shift focus to other less-interesting-but-equally-essential aspects of work.

[Vistaus]

Interesting to say the least but isnt Zen version one old hat. This especially if the next playstation is 2 years away? Maybe they have something else planned.

The next PS could be released this year:

https://twitter.com/mochi_wsj/status/773880640953475073

https://www.resetera.com/threads/semiacccurate-gets-some-playstation-5-next-details-1000-subscription-to-read-summary-in-op.33810/

[Duve]

Zen+ (2000-series) is literally just a node change. They didn't even fundamentally change layout.

True, but they never said that it was going to be anything major.

Zen 2 is an architecture update. If they hope to compete with Intel on single-thread performance, it's very much needed.

Yeah, as I said Zen 2 should be something more fundamental when it comes to architecture and features than what Zen + offered, but I would not say that they would be rebuilding the architecture from the ground up at this point. Zen has made it's mark similar to the old Athlon line, I would not expect a shift from it so swiftly and I don't think that it is needed.
Funny thing about the single-core performance, that is one of the bigger changes with Zen + currently. They are about equal in most aspects compared to 8-gen Intel's, the thing is that from what I saw in the numbers produced by several reviews... gaming will not be the beneficiary of that. Kind of a shame but that is not to be unexpected, It is a general purpose chip, gaming should not be the sole beneficiary to improvements.

[Citan]

I'd like to believe, but deep down in my heart I know they won't. For one reason and one reason only. They're Valve, and so if they do launch hardware again they're going to quarter ass the launch again because they didn't even go so far as half assing it. Which is the main reason the previous batch of steam machines went nowhere.

Valve gets a 30% cut of every game sold on Steam and literally has the most leverage of any company in the game industry to get others to do what they want. You really want to take that stance when what they could and should have done is hired an entire division similar to Sony or Microsoft to do OS development which would have easily made all those problems go away well before launch. Then at launch they could have actually spent the time and money to actually market the thing to the general public, and set up stores to have proper display units rather than just shoving them into gamestops to have a wall of before promptly seeming to forget the whole project even existed.

This comment is, sorry I have no other words, lacking any serious hindsight to support it.
You're judging Valve for a failure that was not imputable to them only but on hardware designers for a good part.
In fact Valve played smart for their own interest, although they did sell a dream short.
They threw a big stone in the pond to see how and which fish would come up.

When they announced Steam OS, it was mainly to stress the risk for them, as well as for any content-provider, to be locked out (or in, depending on how you see it) of Windows Store.
They absolutely didn't have the required know-how to build machines, so of course they wouldn't try to make one themselves (especially since the upstream investment is of another magnitude as for pure software).
They also probably didn't expect many problems they ran into when trying to build a SteamOS, and realized there were many areas which were still unsuitable for consumer use, notably drivers.

So they waited to see what would happen... What happened is that assemblers made a mess, completely missing the mark on market. That was a bad thing. What happened too is that...
- Game designers and producers began showing a real attention to Linux gaming, even if most of the big publishers manage it as an afterthought. Yet, the light Valve put on this area made an impact, confer the number of Linux-available games over the years: there is a before and an after.
- Linux developers announced (or continued) developing new hardware-related libraries and subsystems, like the infamous Systemd and Wayland (there was also some big work that began on audio but I don't remember well so I won't speak of it ).
- AMD started pushing a true open-source strategy, thus tackling, at least as far as its products go, the dreaded driver problem.
- Other "gaming publishers", like Killer, started taking Linux into account, even if they have a big lag in providing drivers (count 2 years minimum after product release, yeeee).
- As stupid and anti-efficient as may be for true infrastructure management, the all-in-one package systems such as Flatpack are rising rapidly.
- A new low-level, open-source, cross-platform API has been pushed (thanks to AMD by the way) and has gained enough popularity to be accepted by everyone (included nVidia who now contributes to it), which eases the cross-platform development as a consequence..

Thing is, Linux has always been rock-solid as a system. But developing a game for it meant mastering many different subsystems, each with their own logic, perks (some of which useless for games) and flaws (some of which blockers for games), thinking notably of windowing (X11) and packaging (deb/rpm). Which is logical, Linux had always been thought as a work platform. ^^
Plus the drivers problem which was a big deal, especially in the gamer crowd for which buying 2-year old hardware is kinda counter-intuitive.

Situation in 2018 is much much brighter than it was in 2010, even if there is still a long road, because most of the real dealbreakers for industrial gaming (and other fields, but that's beyond the point) are being tackled one by one.
As a user, you can expect something stable enough though for the 2020 horizon as far as graphics, windowing and audio are concerned.

Short story, they didn't make things half-assed. They pushed to the most of their ability, considering their primary field of expertise and commerce, which is NOT developing a full operating system NEITHER assembling hardware.

What you are speaking of, hiring as needed to resolve all problems, would have been a suicidal managerial decision no matter how you look at it. So obviously they didn't do it. They tried to the best of their ability, see it wouldn't enough, so put the project on hold, coming back on SteamOS sometimes to see what is better and what still is a problem. They understood that the main strength of open source is that it (usually) improves naturally over time, without you asking for it.

So as Valve said, they are still commited to Linux. They just consider (probably, I'm not working inside it ^^) that it would be a waste of efforts to push SteamOS further while so many things under the hood are under heavy rework. Because it would require to much for them to hire enough people to become Linux-system experts by themselves.

You can legitimately criticize this mindset which we could call "vulture" mindset (profit from other's work, contribute as minimum yourself), but this is sadly, from an entreprenarial point of view, a sound decision. At least for current standards.
Few people really get how contributing without a revolver on the head is actually beneficial in the long run, but that's another problem.

You can also legitimately regret that they didn't decide to push the whole idea by themselves, but again, that's your own wet dream. They are not open source believers, everything they do is to develop their own company. And investing everything needed to end with a viable Steam Machine & OS all by themselves was not, has never been in their interest: the market was (and still is tbh) far too small for that.

In summary, they made an attempt with limited means (also a way to study market potential), decided it was not worth creating an entire division dedicated just to it, and they are waiting until they feel the base to build on is good enough.
Confer what Starshipeleven said: they are probably waiting for Wayland, audio, drivers and Flatpack to be all stable and largely used to try again building a proper Steam client / Steam OS.

You can safely expect a big announcement in the two years to come is my bet.
And indeed, considering how an "open-ended project" worked last time, even if the context is much different today, they'll probably team up with one constructor to provide one unique model of Steam Machine.

As for Steam controller / Steam Link: this is something totally different: both products were much easier to develop (conception 100% @home, much more manageable functional perimeter) and had a much clearer goal, more suited to Valve's main activity (publishing games, being an entertainment content provider): the first one was a way to bridge games originally unsuited to the "couch gamer", the second was a way to expand the potential public towards couch gamer but also improve Steam as a multimedia provider (animes/movies streaming). Neither need SteamOS to bring value to the consumer, neither to Valve. That's the big difference.

[DMJC]

SteamVR is NOT ready yet, all this talk of Steam Machines is a joke. I have the X.org 1.20 server built for debian (compiled myself) with the latest NVIDIA Drivers and I got the PSVR to work in Direct Mode with SteamVR. SteamVR is currently very unstable and has crashes on Miles Audio, as well as missing gamepad support.

From my logs:
Mon May 21 2018 19:05:17.730551 - Unable to load driver gamepad from /home/james/.steam/steam/steamapps/common/SteamVR/drivers/gamepad/bin/linux64/driver_gamepad.so.
Mon May 21 2018 19:05:17.730563 - Unable to load driver gamepad because of error VRInitError_Init_FileNotFound(103). Skipping.

SteamVR is still in Beta and is a long way from being production ready.

[chithanh]

Google, with its massive mountain of resources, didn't progress Android into "not suck" territory until somewhere in the 4.x release series and they were only supporting applications on APIs they controlled.

I disagree with this part.
I think the version of Android that made the OS usable for the general public was 2.3 (Gingerbread).
With 4.0 (Ice Cream Sandwich), it was roughly on par vs. iOS.
Since 4.4 (Kitkat) no other mobile OS could compete anymore.

When Google announced 1 billion Android device activations in September 2013, around half of Android smartphones were still running Android 4.0 or older.

[garwynn]

The next PS could be released this year:

https://twitter.com/mochi_wsj/status/773880640953475073

https://www.resetera.com/threads/sem...y-in-op.33810/

Actually, there's an update to that. And it falls right in line with what we're seeing here.

https://twitter.com/mochi_wsj/status/998787436606976002

https://twitter.com/mochi_wsj/status/998785954620977152

This was widely perceived as PS4 is in sunset mode, internal h/w & s/w dev will largely shift to PS5 development with a expected market release in 2021. Gives Sony 2 years to start working with AMD to hone Zen for a semi-custom SoC order probably placed next year. So might be working with Zen/Zen+ as prep for Zen 2 base. Graphics likely to be Vega, don't think Navi ES have been produced yet for them to even try.

In the meantime PS4 (original) will probably be EOL in next 12-18 months, PS4 Pro will be "the" console then for 2 years, full lifetime of PS4 Pro about 4.5-5 years.

Yep, all of that sounds right for a console life cycle.

[coder]

...

Well, you certainly win my tl;dr award.

[chithanh]

In the meantime PS4 (original) will probably be EOL in next 12-18 months, PS4 Pro will be "the" console then for 2 years, full lifetime of PS4 Pro about 4.5-5 years.

Are you sure about that? I think PS4 Pro will go EOL as soon as PS5 launches, while PS4 will be made as cheap as possible (maybe even dropping the optical drive and making it an optional external accessory) with Sony trying to replicate the PS2 success in its later years.

When PS3 came out, PS2 price got reduced to $129, and Sony proceeded to sell another 40M or so. I don't think this will be possible with PS4 Pro. Plus, Sony always maintained that PS4 Pro was just another variant of the PS4 and will not outlive the PS4.

[Novan Leon]

Despite the challenges that Valve faces trying to enter into the console market, there's one more obvious one that not many people seem to be mentioning: the fact that the console space is currently oriented around a living-room controller-based experience while the majority of the Steam library is mouse+keyboard based. This may sound like a small thing, but mainstream markets can very very finicky about stuff like this, and if your everyday John Doe can't sit down play the latest version of Madden, NBA Live or Call of Duty from the comfort of his couch with minimal muss-and-fuss, there's little reason for him to buy a "SteamBox" over one of the established consoles.