Linux Gaming Was Great In 2016, But 2017 Should Be Even Better

[aht0]

Seeing how roughly 4% of Steam users run OSX so it's not like EA is all that conservative when it comes to supporting smaller platforms.

When you are buying yourself Apple hardware, you have to have the monetary means. Equals better probability for paying customer. I cannot really find any stats about Origin usage. It'd be interesting.

In my experience well supported games tend to be pretty low on issues and when someone with the ability to make changes in the original game source code is the one putting out those games there's every possibility for a perfectly fine Linux port even if it's a Windows game in a Wine wrapper.

And are pretty boring and/or old, except of EVE Online and Fallout 3. Nothing else in the list which is remotely fresh. And with EVE you'd have to jump through hoops for disabling DX11. Christ...

Which isn't all that interesting when you could get all the same visual settings, except with better performance under XP.

There were slight but noticeable graphical differences and performance was actually better on Vista and 7. At least playing Bad Company 2. If other game engines did not bother optimizing, it was not Vista/7/DX10 issue but theirs.

Seeing how those games came out 4, 5 and 3 years after Vista I'd call the "migration" to D3D10 anemic at best.

General hardware upgrade cycle tends to be around 5 years for average user. 1-2 years for enthusiast. I'd say the migration proceeded in acceptable pace.

There were plenty of perfectly valid reasons to avoid Vista keep using XP up until 2011 when Windows 7 was out and Microsoft had worked out the launch kinks that are an inevitable in a new OS release. At that point you might as well have started a migration straight to DX11, which for many developers didn't finish until a couple of years ago. As I mentioned, Fallout 4 was not only Bethesda's first D3D11 game, it was also the first Bethesda game to ship with something other than a sole D3D9 render since they moved to one.

Windows 7 RTM was released on August 2009 for MSDN and TechNet clients and for others in October 2009. Not in 2011.

To put it so that even you get it: People using Linux because they're cheapskates is essentially a myth. If you're using a prebuilt you already have a Windows license and if you have the skills to install Linux on a custom built machine you obviously have the skills to pirate Windows.

Even me get it, duh? What an intellectual superiority...
You do not have to pirate Windows to begin with. It's easy enough to get license and use it essentially legally. For very little money if any. Hell of a lot safer as well than trying out cracks. Even you should get it.
Guy willing to have Lubuntu on his ancient PC/laptop because XP EOL'ed (and he is so scared because of media campaign) is not cheapskate? Even if he is utterly unwilling to spend couple hundred on a newer machine. Funny.

Now you're just arguing semantics... The end result is pretty much the same and most wrappers aim to be as transparent as possible. It's pretty common to have games developed in a way where different versions are handled by different studios and all the studios obviously get access to the source code of the game (how else do you port a game to consoles with highly proprietary APIs?). Because of that it's pretty obvious that companies like Feral get source code access when they make a port, even when it's using a semi-transparent wrapper.

No, I am not. Wrappers are bs. You always get extra issues compared to native client port. Each issue is more or less serious annoyance for a gamer. Or even show-stopper to a degree he is switching back to using windows.

Yes, the same way people didn't care that McDonalds makes the Chicken McNuggets out of that pink goop stuff until they found out about it. Never mistake ignorance for approval.

I don't get such references. I don't each garbage. I haven't had even simple hamburger like 4 years.

As I said, we're not talking about just ads in games, we're talking about the OS itself gathering some highly questionable data (which I forgot to mention also includes a summary of sorts of everything you type on your computer) and using that to allow all apps, most importantly non-game ones, to have ad spaces to display ads targeted based on the data collected by the OS.

Topic is about games. I don't even get it why you started about talking about Windows and Android collecting user data. Nor have I any interest in backtracking and checking.
Do you have proof or you are following conspiracy theories? I've watched Win10 telemetry traffic in Wireshark and it's all encrypted. Wan't to stop it, don't whine about it, use firewall, external. You could also "castrate" the Windows enough for neutering it for the most part.

The thing about the internet is that you find people suffering from problems with the products of every hardware vendor under the sun if you look hard enough. It's not like Nvidia has never had issues with the way Microsoft's telemetry data showed that over 30% of all BSODs on early Vista were caused by Nvidia's drivers (AMD's share of the blame was only about half that), how they've had several instances of bad drivers physically breaking cards and the 3.5+0.5 GB thing on the GTX 970.

And AMD cards were at the time much less being used than ones of Nvidia to begin with. 2009 (by the time 7 came out) Nvidia held ~70% market share on discrete graphics card segment. Now extrapolate from there.

[starshipeleven]

You could have found pissed off pre-built pc-sellers who have come public with their own RMA statistics.
One example:Seems to be about R9 lineup.
http://www.pcworld.com/article/20521...eo-cards-.html

Wow, 1% vs 3% failure rate where we don't even know what is "failure" (hardware, software, both). You know what means "too low to matter"?

Also the same one from Puget systems (I know and trust them due to their articles on detecting ECC support and due to their selling stuff that isn't just gaming shit, they also sell ECC workstations, and also use Asrock Rack stuff) states
So yes, AMD does have a higher failure rate, but nothing that puts up such a big red flag that I would want to drop their product

Most recent RX-series AFAIK are sold by OEM’s using pure AMD’s reference design. Can’t accuse OEM design here. Assembly yes, cheap components perhaps yes, but not design.

All ones I see don't look like reference one bit. And I talk of board, not of cooler.

Nvidia's driver issues seem to be mostly last year and Win10-specific.

Learn to use Google, and spare me the rest of your bullshit in that paragraph.

https://answers.microsoft.com/en-us/windows/forum/windows8_1-performance/windows-81-update-1-bluescreen-when-installing/fbf84b40-4e1b-43f5-a00b-dfa6f1b1ec52

https://forums.geforce.com/default/topic/705447/installing-new-nvidia-drivers-causes-blue-screen-of-death/

https://www.reddit.com/r/nvidia/comments/2h418n/windows_81_crashes_with_bsod_after_i_driver_update/

What I claim is „nvidia drivers have better quality on average and you do have plentitude of versions to choose from”.

The "better quality on average" is still bullshit, and you can install any version of AMD drivers too so I don't get what's the point of mentioning that.

Upon running into some sort of issue (MSI Afterburner overlay crashing shit, videostreaming not working properly) I've just tried different versions and found which work. It's never taken more than half-hour.

Hah, so you had issues with NVIDIA drivers too even by using older ones to avoid the normal less-than-stability that most newer drivers have. Then it's AMD that has shitty drivers. Cool story bro.

Anyway, if you need to to the same with AMD, you can too. http://support.amd.com/en-us/downloa...%2010%20-%2064

One particular rushed fuckup from MS does not explain/excuse away issues with AMD cards/drivers throughout the time.

Sure, my point was that you need to be more careful about what you post as proof. If you post a thread where someone is having issues because MS updates, that cause issues across the board and it's well-known, I tend to think you are just a clueless troll.

Intel issue is even totally besides the point. It's an overclocking issue.

Sure, an overclocking issue that happened after a specific Windows update and was fixed after a specific windows update, on a CPU designed for OC in the first place and that was 100% fine with the same settings in OC on Linux. (Yes I had to troubleshoot one of them and I could test Linux on it).
That's an obvious software issue, instability caused by overclocking is the same for every OS as it is hardware instability.

Nice handwaving, btw.

Okay, I shall expand on it:

Irrelevant, your own personal experience is not statistically significant. Same as with hard drives, People swear by one brand or the other or by SSDs but the failure rates are tiny in either case, so it's mostly bad luck and not a pattern.

[aht0]

Wow, 1% vs 3% failure rate where we don't even know what is "failure" (hardware, software, both). You know what means "too low to matter"?
Also the same one from Puget systems (I know and trust them due to their articles on detecting ECC support and due to their selling stuff that isn't just gaming shit, they also sell ECC workstations, and also use Asrock Rack stuff) states
So yes, AMD does have a higher failure rate, but nothing that puts up such a big red flag that I would want to drop their product

1% vs 3% means failure rate which has difference of three times. That during RMA period. What you could possibly conclude from it, is even higher difference AFTER RMA period. Since reliability of electronics goes down in time.

All ones I see don't look like reference one bit. And I talk of board, not of cooler.

Initially there were nothing but reference cards for RX series. Aftermarket designs appeared later. Search better. Youtube is full of PCB breakdowns.

Learn to use Google, and spare me the rest of your bullshit in that paragraph.

https://answers.microsoft.com/en-us/windows/forum/windows8_1-performance/windows-81-update-1-bluescreen-when-installing/fbf84b40-4e1b-43f5-a00b-dfa6f1b1ec52

https://forums.geforce.com/default/topic/705447/installing-new-nvidia-drivers-causes-blue-screen-of-death/

https://www.reddit.com/r/nvidia/comments/2h418n/windows_81_crashes_with_bsod_after_i_driver_update/

The "better quality on average" is still bullshit, and you can install any version of AMD drivers too so I don't get what's the point of mentioning that.

Oh indeed? Mathematically. If you have 30+ driver versions to choose from one manufacturer or handful from the other per particular card, upon meeting an actual driver issue: Which manufacturer is offering you better probability of fixing some particular issue by you trying another driver version? One with umpteen driver releases or one with a few?

Now, developing the same line of thought, one could also extrapolate that manufacturer with bigger amount of driver releases could have bigger total count of errors in them, provided they are of equal quality. What I do not actually think.
Hell, If you want quick, although subjective proof, just check the AMD Linux subforum and compare the thread titles to ones found in Nvidia Linux here in Phoronix.

AMD Linux: (just Page 1)
[AMDGPU-Pro 16.30] CS:GO refuses to launch
AMDGPU-PRO PowerPlay forcing power state
Very strange issue with the latest Mesa OpenGL drivers from Oibaf (picture included)
Amdgpu-pro 16.50, Opensuse Leap 42.2 fails running Steam
Proper driver setup for RX series?
Typing in Eclipse very laggy (Ubuntu 16.04 + Unity + RX480 + AMDGPU PRO 16.30)
Amdgpu-pro 16.30 - What does this backtrace mean?
Multiple amdgpu-pro drivers failing on Radeon 360 + Ubuntu 16.04.
Using AMDGPU with HAWAII on Debian 8

Now, lets compare it with the content of Nvidia Linux subforum page 1: "Problem threads"
Gentoo & Nvidia Optimus issues: MSI GE62 2QD Intel Core i7

Notice the amount of "problem" threads in AMD Linux? I did not include anything from Open-Source AMD subf even. Or I could have added more to AMD's pile.

I did not take "30+" out of air, I went and counted Nvidia's releases from the most current release back to 350.xx and got 36. Conditions applied to the counting were:
-9xx-series supported,
-Win7/8/10
-WHQL drivers.

I also counted 10 WHQL driver releases downward out of curiosity and checked if it still had support for second-newest generation of cards, it had. So, it makes ~50 driver releases for second newest generation. Second newest because majority of the people are not refreshing their GPUs instantly to the latest and baddest but stick for some time to what they already got.

Go count WHQL driver releases for AMD starting from Catalyst 16.x. Applying same conditions. You shall get exactly 4.

Hah, so you had issues with NVIDIA drivers too even by using older ones to avoid the normal less-than-stability that most newer drivers have. Then it's AMD that has shitty drivers. Cool story bro.

Time to time everyone run into issues. My issues with Nvidia card never exceeded the level of
- MSI Afterburner overlay crashing when used simultaneously with OBS Studio.
- FPS is sort of a low, maybe another driver version gives me higher values.
Might be because I mostly use Win7 Pro and when I did use Win10 Pro, I still used older drivers. And had Windows Updates mostly blocked, both in win10 and in firewall. Can't tell if I would have run into BSODs had I acted different.

Anyway, if you need to to the same with AMD, you can too. http://support.amd.com/en-us/downloa...%2010%20-%2064

Can you guess why I picked WHQL drivers for counting, above? Because ones NOT marked WHQL did not pass the Windows Hardware Quality Labs testing process
for 3rd party hardware. Yeah, vendor can sign drivers with it's own certificate and still release them which AMD does. But such particular drivers obviously performed in less than stellar manner in some aspects or they would have passed the testing.

Sure, an overclocking issue that happened after a specific Windows update and was fixed after a specific windows update, on a CPU designed for OC in the first place and that was 100% fine with the same settings in OC on Linux. (Yes I had to troubleshoot one of them and I could test Linux on it).
That's an obvious software issue, instability caused by overclocking is the same for every OS as it is hardware instability.

Does the quality of windows updates matters here at all? Since we were arguing about Nvidia vs AMD so far? It might have had some point but going back in the thread I could not find the logical reasoning to the link. I'll grant you may well be right though in this one.

Irrelevant, your own personal experience is not statistically significant. Same as with hard drives, People swear by one brand or the other or by SSDs but the failure rates are tiny in either case, so it's mostly bad luck and not a pattern.

It is relevant actually, as somebody, either you or somebody else came forth boldly and accused me taking my claims out of thin air, without ever having to deal with ATI/AMD hardware, since I am convinced nVidia-fanboy.. or something along the lines..

[GreekGeek]

Hi yall & ath0,

totally epic thread/debate! Dunno where you get the patience from & good job.

Michael has published so many articles about how pants the ATI/AMD Linux drivers are, not because he is a "fanboY," rather simply as a result of driver issues. Being a functional rocket scientist is not needed, their is a tsunami of data showing the same. Ergo, peeps that claim that is not the case, are simply going to continue on "the trip," they are on & offer "alternative facts." Alas, there are many Donald's out their ....

Ath0, better not engage with such and rather game on the games yall like, hang out with friends and enjoy the many good things this life has to offer. Life is too short.

Perhaps one day the ATI/AMD drivers will be up to speed. The latter would be a good thing, least the situation with Intel being lazy/pricey with CPU/s, happens with GPU's. That would also be pants. For myself, I have a new case, SSD, DDR4 ram chippies & await an Ryzen CPU & AM4 motherboard.

GreekGeek :-)

[starshipeleven]

1% vs 3% means failure rate which has difference of three times. That during RMA period. What you could possibly conclude from it

That it is still too low to matter. And no, tests with hard drives and other electronic devices didn't prove that failure rates flat-out increase with time. They had a relatively large "infant mortality", then a relatively low failure rate, then like 6 years down the line there is a slight rise. For hard drives.

Initially there were nothing but reference cards for RX series. Aftermarket designs appeared later. Search better. Youtube is full of PCB breakdowns.

What I saw is many OEMs using the same pcb design, but that's not reference. Especially for rx460 that has no reference.

Oh indeed? Mathematically. If you have 30+ driver versions to choose from one manufacturer or handful from the other per particular card, upon meeting an actual driver issue: Which manufacturer is offering you better probability of fixing some particular issue by you trying another driver version? One with umpteen driver releases or one with a few?

Lol, if you can't fix your issue after trying what, 20 different drivers, You're better off dumping the card (or the motherboard) as it's either a hardware issue or an issue that isn't gonna be fixed.

That said, there are sites that keep all drivers ever published of both AMD and NVIDIA (guru3d I think... here it is http://www.guru3d.com/files/index.html ), I had to use them a long time ago for a specific NVIDIA 6200 on AGP that worked on Win7 only with driver 77.xx beta for Vista or something like that, but I'm more for blaming the shitty VIA chipset of the mainboard than NVIDIA.

Now, developing the same line of thought, one could also extrapolate that manufacturer with bigger amount of driver releases could have bigger total count of errors in them

baseless extrapolation, what is sure is amount of bugs per code generated. Releases may or may not have much code change in them.

Hell, If you want quick, although subjective proof, just check the AMD Linux subforum and compare the thread titles to ones found in Nvidia Linux here in Phoronix.

Now that you can't claim anything about windows you fall back to beating dead horses on linux? AMD linux dirver has always been meh, and now the driver is immature so it's still very meh and that's a well-known fact.
I thought you were talking about windows, also because all examples you brought up were about windows till now.

Go count WHQL driver releases for AMD starting from Catalyst 16.x. Applying same conditions. You shall get exactly 4.

And this means? They release more often and they pay for WHQL more often. Is that better quality? maybe, maybe not. Meanwhile there is a warning in Starmade (Minecraft in space, a very graphics-intensive game... not), https://starmadedock.net/content/factory-base.2383/
NVIDIA's driver release (378.49) breaks basic OpenGL functions vital to StarMade. This driver update has broken applications that rely on it, this includes similar titles such as Minecraft. We're unable to address the issue ourselves, NVIDIA needs to send a fix. Please read here on how to rollback!

In case you did not know, that driver version is WHQL.

And had Windows Updates mostly blocked, both in win10 and in firewall. Can't tell if I would have run into BSODs had I acted different.

This authorizes me to call you names. Updates must be left enabled, maybe not automatic as you need to make sure they don't break anything so you should wait a week or so and look around on google, but you must run them regularly for the sake of security.

Can you guess why I picked WHQL drivers for counting, above?

Because you don't know what it means?
WHQL is a test that verifies that it is cool for MS, it expensive and takes time, but is not a guarantee that the driver isn't breaking games or other shit (see above).
It's only needed for including the driver in MS's update servers as a MS-approved driver.

Drivers not WHQL are simply not tested by MS at all, they didn't "fail" anything. People use non-WHQL drivers all the time.

Does the quality of windows updates matters here at all?

As long as you are able to tell the difference in issue threads and make sure you are not counting as AMD issues the issues caused by Windows updates, that's fine.

It is relevant actually, as somebody, either you or somebody else came forth boldly and accused me taking my claims out of thin air, without ever having to deal with ATI/AMD hardware, since I am convinced nVidia-fanboy.. or something along the lines..

I said it is irrelevant because personal experience in matters where failure rate is less than 5% IS pretty much irrelevant, unless you personally tested a few hundred thousand cards of both brands you have a too low sample for your findings to be worth using for making any kind of conclusions.

[aht0]

(0)That it is still too low to matter. And no, tests with (1)hard drives and other (2)electronic devices didn't prove that failure rates flat-out increase with time. They had a relatively large "infant mortality", then a relatively low failure rate, then like 6 years down the line there is a slight rise. For hard drives.

(0)Cumulative effect down the road.
(1)First show me hard drive capable for working at up to 100-120 degree Celsius and drawing 100-200+ W. Then let's start comparing such HDDs with discrete GPU's. Even my mechanical 15krpm SAS drives did not draw more than dozen Watts from the controller per unit. Also, as far as I remember mech drives are rated to 50 degrees tops. Components are not comparable and you actually know it. Just throwing bs at me.

(2)Use logic. Any electronics that meets immense thermal stresses in use, is going to degrade. Solders themselves for example. For seconds, electrolytic capacitors on card are not going to last forever. And yes, discrete GPU has'em.

So, you got fairly heavy thermal stress (which faster or slower IS killing solders themselves), certain components that degrade over time and still insist that failures do not flat-out increase over time? Standard electrolytic capacitor is rated for 2000h/85C and "long-life" capacitors have 5000h/85C. Card that runs hotter fails faster, since every 10 degrees up or down either decreases or increases life-expectancy of a capacitor about two times. Once past certain point, "curve of failure" becomes near exponential if I recall graphs I saw ages a go right. Feel free to correct me. Memory may have betrayed me and it was ages a go when I did read the "handbook of the radio amateur".

Following the same train of thought. AMD RX series has cooler working temps (75-90C?) but older Radeon versions were literally ovens (90-100+C under heavy load). My nVidia's have never exceeded 70-75C under heavy load.

You can from here make assumptions about different thermal stresses and stresses on components of PCB on particular cards. And how it may affect longevity and stability of the card vs longevity and stability of the card running hotter. Case closed.

What I saw is many OEMs using the same pcb design, but that's not reference. Especially for rx460 that has no reference.

What do you think, how they got the design? Engineers worked it out separately and all ended up on same design? It was AMD's.

Lol, if you can't fix your issue after trying what, 20 different drivers, You're better off dumping the card (or the motherboard) as it's either a hardware issue or an issue that isn't gonna be fixed.

Did I write anywhere I had to try out 20 drivers? 3-4-5 have not been enough in Radeon's case though...

That said, there are sites that keep all drivers ever published of both AMD and NVIDIA (guru3d I think... here it is http://www.guru3d.com/files/index.html ), I had to use them a long time ago for a specific NVIDIA 6200 on AGP that worked on Win7 only with driver 77.xx beta for Vista or something like that, but I'm more for blaming the shitty VIA chipset of the mainboard than NVIDIA.

Did you think I took my data from Nvidia's site? Which would be showing only 10-12 latest releases per selected generation of cards/OS/arch.

baseless extrapolation, what is sure is amount of bugs per code generated. Releases may or may not have much code change in them.

Now, what is the point of a release if there is no change in a code? Bugfix would be an hotfix. Bot manufacturers release driver hotfixes.

Now that you can't claim anything about windows you fall back to beating dead horses on linux? AMD linux dirver has always been meh, and now the driver is immature so it's still very meh and that's a well-known fact.
I thought you were talking about windows, also because all examples you brought up were about windows till now.

And had I selected some big tech forum you would have had another issue: why this particular tech forum. Sure it's biased somehow against AMD. I took Phoronix because it was "closest" and I particularly did not add Open-Source drivers because these are not really "only" AMD's anymore. So, I compared closed source drivers and ignored the OS itself. Regardless of what I did, you would have had "something" to whine about.

And this means? They release more often and they pay for WHQL more often. Is that better quality? maybe, maybe not. Meanwhile there is a warning in Starmade (Minecraft in space, a very graphics-intensive game... not), https://starmadedock.net/content/factory-base.2383/
NVIDIA's driver release (378.49) breaks basic OpenGL functions vital to StarMade. This driver update has broken applications that rely on it, this includes similar titles such as Minecraft. We're unable to address the issue ourselves, NVIDIA needs to send a fix. Please read here on how to rollback!
In case you did not know, that driver version is WHQL.

For a "graphics intensive game" it looks ugly as sin. Phew. Now I know why OpenGL has never taken off. Other than that. Got an issue, try next driver. You have plenty to choose from.
It's not like only Nvidia has had issues with this this game, I could counter-post you half a dozen links about Starmade and Radeon issues, if you like.

This authorizes me to call you names. Updates must be left enabled, maybe not automatic as you need to make sure they don't break anything so you should wait a week or so and look around on google, but you must run them regularly for the sake of security.

Call me what you like, regardless, it appears to have been justified decision. I have enough preventive measures running in the gateway (squid+clamav/squidguard, autoblocking IDS/IPS and auto-updating (cron task) IP/FQDN block/whitelists used by pf) to go like this in relative peace of mind.

WHQL is a test that verifies that it is cool for MS, it expensive and takes time, but is not a guarantee that the driver isn't breaking games or other shit (see above).

yea-yea, bla-bla, microsoft is evil and is there for only about money.. old song and dance.

[aht0]

Hi yall & ath0,

totally epic thread/debate! Dunno where you get the patience from & good job.

It does not really take much time, is fairly interesting and sometimes I learn new facts.

Perhaps one day the ATI/AMD drivers will be up to speed. The latter would be a good thing, least the situation with Intel being lazy/pricey with CPU/s, happens with GPU's. That would also be pants. For myself, I have a new case, SSD, DDR4 ram chippies & await an Ryzen CPU & AM4 motherboard.
GreekGeek :-)

One sure can hope.