Radeon VII Linux review

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
6,312
https://www.phoronix.com/scan.php?page=article&item=radeon-vii-linux&num=1

I know most people that use Linux have already likely read this. But if not.

Impressive Day one Linux performance. Right between 2080 and 2080ti. Not bad at all for day one. Running on a 100% open source driver stack. Seems like if the open source Vulkan driver gets some more tweaking in the next few months at least in regards to Linux the VII is going to creep even closer to 2080ti performance. As the Vulkan tests seem to be the weak spot at the moment. (relatively speaking anyway)

Looking forward to seeing some DXVK / Proton numbers, hopefully we see some testing there soon.
 
Hoping to get my hands on a Navi when they drop. Thanks for sharing this
 
https://www.phoronix.com/scan.php?page=article&item=radeon-vii-mesa&num=1

Only one DXVK title tested here. But look at that the next MESA open source update looks to bump performance at least a bit more. Its look like at least in terms of Linux the VII may well best a 2080 in the majority of titles. Was a bit surprised to see they still found a performance in the OpenGL MESA paths. Just shows there is almost away to squeeze just a bit more out of almost anything.

I know Arkham origins is a little older now and not tested by most windows reviewers. But the 2160p windows 2080ti results I have seen around range from 115-130 FPS depending on the reviewer and the rig. So DXVK Linux Radeon VII pulling 136 FPS at 2160p with the latest MESA open source driver seems pretty damn good to me.
 
Last edited:
Bit wild that it's running better on Linux than Windows- but great to see the envelope pushed nonetheless!
 
Deus Ex 3840x2160 ....
Kernel 4.20 MESA 18.2 36.37 FPS
Kernel 5.0 MESA 19.0 52.60 FPS

That's a pretty big jump. Looks like if anyone does buy one of these they are going to want the latest greatest hot off the GIT stuff for now at least. Now I'll be interested to see if 19.1 in a few months has anymore performance in it. lol
 
Bit wild that it's running better on Linux than Windows- but great to see the envelope pushed nonetheless!

Linux is a faster OS than Windows. I know I should be putting on the flame suit for making such a bold claim, but the simple fact of the matter is 'Linux is faster'. It has a far better scheduler and NUMA implementation than Windows and a far better and faster file system than NTFS, both points that I can substantiate if needed.
 
  • Like
Reactions: GHRTW
like this
Linux is a faster OS than Windows. I know I should be putting on the flame suit for making such a bold claim, but the simple fact of the matter is 'Linux is faster'. It has a far better scheduler and NUMA implementation than Windows and a far better and faster file system than NTFS, both points that I can substantiate if needed.
I like reading material. Whatcha got? (I'm always gonna say linux is faster, regardless of any evidence to the contrary regardless :D)
 
Linux is a faster OS than Windows. I know I should be putting on the flame suit for making such a bold claim, but the simple fact of the matter is 'Linux is faster'. It has a far better scheduler and NUMA implementation than Windows and a far better and faster file system than NTFS, both points that I can substantiate if needed.

I'm inclined to believe that you're right, and the mentioned issues certainly matter to those that use their systems in such a way to expose said performance deltas, but I don't see how either really relate to Linux gaming ;).
 
I'm inclined to believe that you're right, and the mentioned issues certainly matter to those that use their systems in such a way to expose said performance deltas, but I don't see how either really relate to Linux gaming ;).

The OS as a whole is leaner and faster with a better file system and scheduler, gaming is software that relies on all of the above. The simple need for antivirus under Windows sucks up resources, something that's not an issue under Linux.

I like reading material. Whatcha got? (I'm always gonna say linux is faster, regardless of any evidence to the contrary regardless :D)

https://www.phoronix.com/scan.php?page=article&item=2990wx-linux-windows&num=1

NUMA implementation is very important regarding newer multicore processors and IMC's. The scheduler under Windows is overall simply no where near as good as the scheduler under an OS designed to run supercomputer clusters. Furthermore, this seems to be an issue that Microsoft themselves simply cannot address.

https://github.com/Microsoft/WSL/issues/873#issuecomment-425272829

Unresolvable NTFS file system performance issues under Windows that many believe is so integral to the OS that the problem may never be resolved. I can state 100% that when it comes to networked file shares, Ext4 always performs better than NTFS - It may only be a few percent, but Ext4 always performs identically and is generally faster. I can't say the same about NTFS, NTFS performance is all over the place at times.
 
The OS as a whole is leaner and faster with a better file system and scheduler, gaming is software that relies on all of the above. The simple need for antivirus under Windows sucks up resources, something that's not an issue under Linux.

Main issue with this is that once the OS gets out of the way, games just run- it's the hardware itself that should be the differentiator. The biggest issue we've had with Linux was that 'getting out of the way' part, and I'm not convinced that this game is proof that that's happened, as much as it is of AMD's Windows drivers regressing.

NUMA implementation is very important regarding newer multicore processors and IMC's. The scheduler under Windows is overall simply no where near as good as the scheduler under an OS designed to run supercomputer clusters. Furthermore, this seems to be an issue that Microsoft themselves simply cannot address.

This is true on HEDT/workstation/server CPUs that have multiple memory controllers, but it's also not that big of a deal, yet. MS has some time to get NUMA support up to parity.

Unresolvable NTFS file system performance issues under Windows that many believe is so integral to the OS that the problem may never be resolved. I can state 100% that when it comes to networked file shares, Ext4 always performs better than NTFS - It may only be a few percent, but Ext4 always performs identically and is generally faster. I can't say the same about NTFS, NTFS performance is all over the place at times.

I'm sure there are more examples, but generally speaking NTFS is solid for host operating system use. Something performance oriented should be on ReFS, and Microsoft's storage tech while still under active development (as is ZFS and others for Linux/Unix) is quite effective from single NAS implementations up to clusters.
 
Main issue with this is that once the OS gets out of the way, games just run- it's the hardware itself that should be the differentiator. The biggest issue we've had with Linux was that 'getting out of the way' part, and I'm not convinced that this game is proof that that's happened, as much as it is of AMD's Windows drivers regressing.

Where API's are involved, the OS isn't just going to get out of the way isolating hardware as the limiting factor when gaming, it's just not going to happen. the underlying OS doesn't become irrelevant the second a game is started.

The fact is, where a common API such is Vulkan is used it's becoming more apparent that Linux has a performance advantage over Windows on identical hardware.

This is true on HEDT/workstation/server CPUs that have multiple memory controllers, but it's also not that big of a deal, yet. MS has some time to get NUMA support up to parity.

Silicon is at it's limits, clocks aren't ramping like they were in the past and we're beginning to make use of multi core NUMA implementations more than ever before. Vulkan has a far superior multi core implementation than DX11 or OGL, not outright perfect, but without a doubt making use of all cores/threads on my own ageing 12C/24T system and pushing my GPU to it's limits far easier than older mostly single threaded API's. This is something that's going to become a very big deal and the situation appears to be that there's a flaw in Windows itself beyond simple scheduler/kernel upgrades resulting in major performance regressions under certain scenarios. Windows as an OS is beginning to show it's age.

I'm sure there are more examples, but generally speaking NTFS is solid for host operating system use. Something performance oriented should be on ReFS, and Microsoft's storage tech while still under active development (as is ZFS and others for Linux/Unix) is quite effective from single NAS implementations up to clusters.

If you actually read the link provided, the issue goes far deeper than simply changing file systems. Once again, Windows is beginning to show it's age, it's time for a ground up rethink of Windows as an operating system.

The issues surrounding the Windows updater alone that are partially as a result of NTFS are enough to make a man want to kick his Dog.
 
Where API's are involved, the OS isn't just going to get out of the way isolating hardware as the limiting factor when gaming, it's just not going to happen. the underlying OS doesn't become irrelevant the second a game is started.

Not irrelevant, no, but damn near.

The fact is, where a common API such is Vulkan is used it's becoming more apparent that Linux has a performance advantage over Windows on identical hardware.

I don't see this as a trend, but more a few aberrations that need deeper exploration. Generally games run a bit slower on Linux, and the answer as to why is complicated; when they occasionally run faster, it's more likely to indicate that the developers messed something up on Windows. Copy protection and anti-cheat are general suspects...

Silicon is at it's limits, clocks aren't ramping like they were in the past and we're beginning to make use of multi core NUMA implementations more than ever before. Vulkan has a far superior multi core implementation than DX11 or OGL, not outright perfect, but without a doubt making use of all cores/threads on my own ageing 12C/24T system and pushing my GPU to it's limits far easier than older mostly single threaded API's. This is something that's going to become a very big deal and the situation appears to be that there's a flaw in Windows itself beyond simple scheduler/kernel upgrades resulting in major performance regressions under certain scenarios. Windows as an OS is beginning to show it's age.

Something that seems to be missed: just as Linux developers are improving their codebases for their targets, MS is doing the same. I understand that those targets have been different and that both are in different places, but I do expect the proliferation of consumer-bound NUMA implementations to nudge MS into optimizing their own code to take advantage of whatever hardware comes along.

If you actually read the link provided, the issue goes far deeper than simply changing file systems. Once again, Windows is beginning to show it's age, it's time for a ground up rethink of Windows as an operating system.

To do what? Right now, Windows works incredibly well on a wide variety of desktop hardware. It's different from how Linux is set up, of course.

The issues surrounding the Windows updater alone that are partially as a result of NTFS are enough to make a man want to kick his Dog.

The updater itself doesn't really have issues, but MS has certainly configured it aggressively. Which I understand; as much as it infuriates those that demand to have more control over their systems, they're (we're!) in the extreme minority and the majority left to their own devices represent a real threat to each other and everyone else that uses the internet.

And if you compare it to the issues seen with rolling Linux releases... well ;)
 
Last edited:
Not irrelevant, no, but damn near.

I'm not looking for an argument, but the OS doesn't suddenly become anywhere near irrelevant the second you fire up a game.

I don't see this as a trend, but more a few aberrations that need deeper exploration. Generally games run a bit slower on Linux, and the answer as to why is complicated; when they occasionally run faster, it's more likely to indicate that the developers messed something up on Windows. Copy protection and anti-cheat are general suspects...

You create a CPU bound scenario (1080p), effectively taking drivers and the GPU out of the equation, and under a great many instances running Vulkan as it's a common API Linux is in many cases quite a bit faster than Windows. Phoronix benchmarks regarding Dota 2 highlight this perfectly, and these reviews were done in 2018 after Windows saw substantial improvements following Valve's Dota 2 Linux optimizations:

https://www.phoronix.com/scan.php?page=article&item=june-2018-gpus&num=2

Even at 4k The Talos Principle shows substantial gains under Linux vs Windows. Those gains are better than a GPU upgrade in many cases:

https://www.phoronix.com/scan.php?page=news_item&px=Win10-Linux-Vulkan-Early

While it's difficult to track, there's no doubt that even this YouTube video is showing max FPS higher under Linux running Vulkan than Windows:



Something that seems to be missed: just as Linux developers are improving their codebases for their targets, MS is doing the same. I understand that those targets have been different and that both are in different places, but I do expect the proliferation of consumer-bound NUMA implementations to nudge MS into optimizing their own code to take advantage of whatever hardware comes along.

How long have these processors been out? How long does it take for MS to 'optimize code'? The signs are there that this issue under Windows is a little more than simple code optimization.

To do what? Right now, Windows works incredibly well on a wide variety of desktop hardware. It's different from how Linux is set up, of course.

Almost all of the updating issues experienced under Windows, the absolute 'need' to reboot in order to apply updates, are a result of NTFS. I'm not claiming that Windows isn't capable enough to satisfy the average user (provided they don't actually do work with their machine, like encoding 4k video) - I'm stating, quite factually, that simple Ext4 under Linux is vastly better than NTFS and contributes to the overall performance advantages seen under Linux as well as allow for a vastly better updating process.

As stated, Windows, the NT kernel and NTFS are beginning to show their age.

Even Apple's new APFS can't hold a candle to Ext4. HFS+ is downright hopeless where performance is concerned.

The updater itself doesn't really have issues, but MS has certainly configured it aggressively.

The updater under Windows has massive issues. It's inconvenient, it's unreliable, it's slow and it robs the end user of literally all control. That's a point you simply cannot argue. It's not a matter of MS configuring it aggressively, it's a matter of MS shifting more towards the idea of 'Windows as a service' - This is Microsoft claiming back control of their OS.

And if you compare it to the issues seen with rolling Linux releases... well ;)

I've experienced vastly less issues running bleeding edge rolling releases under Linux than the Windows updating process.
 
Last edited:
You create a CPU bound scenario (1080p), effectively taking drivers and the GPU out of the equation, and under a great many instances running Vulkan as it's a common API Linux is in many cases quite a bit faster than Windows. Phoronix benchmarks regarding Dota 2 highlight this perfectly, and these reviews were done in 2018 after Windows saw substantial improvements following Valve's Dota 2 Linux optimizations:

https://www.phoronix.com/scan.php?page=article&item=june-2018-gpus&num=2

The overall point is that while there are definitely some examples of games running faster under Linux, there are many more of games running faster under Windows- or hell, running at all. This does not make a trend, which your link verifies. It makes for points of curiosity.

How long have these processors been out? How long does it take for MS to 'optimize code'? The signs are there that this issue under Windows is a little more than simple code optimization.

How long has Linux been targeted for odd architectures not seen on desktops? How long has Windows? How many NUMA-enabled CPUs are used for desktop work? How many gamers are willing to give up single-core performance for that?

Almost all of the updating issues experienced under Windows, the absolute 'need' to reboot in order to apply updates, are a result of NTFS. I'm not claiming that Windows isn't capable enough to satisfy the average user (provided they don't actually do work with their machine, like encoding 4k video) - I'm stating, quite factually, that simple Ext4 under Linux is vastly better than NTFS and contributes to the overall performance advantages seen under Linux as well as allow for a vastly better updating process.

As stated, Windows, the NT kernel and NTFS are beginning to show their age.

I don't see a problem with the updater, or the need to reboot for updates- you update Linux to the same level, it needs to reboot too. Of course, if you never update your system...

As for NTFS, you can make complaints, but the fact is that it's been working quite well for decades. Even quickly!

And to think, we couldn't possibly encode 4k video without Ext4...

The updater under Windows has massive issues. It's inconvenient, it's unreliable, it's slow and it robs the end user of literally all control. That's a point you simply cannot argue.

It's also been working quite well. I take personal issue with how Microsoft has leveraged it, but I also understand why.

I've had vastly less issues running bleeding edge rolling releases under Linux than the Windows updating process.

I've seen some crazy shit with non-rolling releases. And rolling releases. On a whole, Linux distributions are far less cohesive and that absolutely gets in the way of doing work. You never really know if an update is going to bork the system or an application that the system exists for in the first place. Reality is that can happen on any system and Linux will not be immune to the associated growing pains.


Overall, I use Linux where it makes sense, as a server. I'd like to use it in a workstation context more often, but generally speaking the stuff I do on workstations ranges from harder to do to impossible on Linux. I'm quite happy that Linux gaming is making progress, and I hope that MS continues to develop Windows as well, since I use both!
 
The overall point is that while there are definitely some examples of games running faster under Linux, there are many more of games running faster under Windows- or hell, running at all. This does not make a trend, which your link verifies. It makes for points of curiosity.

Not at all. There's examples of 'bad porting jobs where DX to OGL or even Vulkan wrappers' are concerned, but level up the playing field with a title using a common API such as Vulkan and in most cases Linux is faster.

How long has Linux been targeted for odd architectures not seen on desktops? How long has Windows? How many NUMA-enabled CPUs are used for desktop work? How many gamers are willing to give up single-core performance for that?

Odd architectures? NUMA has been around since the 1990's! The difference is that we're now in a position due to limitations in silicon technology and outright clock speeds where multi threaded applications are becoming more and more important - That includes gaming, especially games running Vulkan. The screenshot below shows Doom 2016 running under Vulkan, while not all cores are being fully utilized, that's a damn near perfect multi threaded implementation and it means that I can max out my GPU using clock speeds lower than 4Ghz. If you like I have another htop screenshot showing the poor multi core utilization running a title under DX11.

dED4mag.png


The CPU's in question have been out for quite some time now and, once again, not a fix in sight from Microsoft. Not looking good when the future is multi threaded application, especially where performance is a necessity.

I don't see a problem with the updater, or the need to reboot for updates- you update Linux to the same level, it needs to reboot too. Of course, if you never update your system...

As for NTFS, you can make complaints, but the fact is that it's been working quite well for decades. Even quickly!

And to think, we couldn't possibly encode 4k video without Ext4...

The fact you don't have a problem with the Windows 10 updater indicates a case of denial as opposed to no real issue with one of the most poorly thought out updating systems known. Even where Linux does need to reboot to apply updates, there is absolutely no noticeable increase in boot times - You definitely can't claim the same for Windows 10! Spinning balls as a progress indicator and all. Enable live updating under Linux and no updates are necessary, try that under Windows.

Denial is not a valid argument, sorry.

I never claimed that Windows, NTFS and the NT kernel hasn't been working well for decades - I claimed that all three are only now beginning to show their age, quite dramatically. NTFS will encode video, slower than Ext4.

I've seen some crazy shit with non-rolling releases. And rolling releases. On a whole, Linux distributions are far less cohesive and that absolutely gets in the way of doing work. You never really know if an update is going to bork the system or an application that the system exists for in the first place. Reality is that can happen on any system and Linux will not be immune to the associated growing pains.

Sounds like another popular operating system I know? Except that OS has issues with non bleeding edge rolling releases!

Overall, I use Linux where it makes sense, as a server. I'd like to use it in a workstation context more often, but generally speaking the stuff I do on workstations ranges from harder to do to impossible on Linux. I'm quite happy that Linux gaming is making progress, and I hope that MS continues to develop Windows as well, since I use both!

For me, the best desktop tool for the job is not the one with a fragmented within the one UI interface, shockingly poor updater, lack of control to the end user suffering from cryptolockers and malware at the rate of 12 million strains a month. I'll steer clear of the overheads regarding antivirus and I'll focus on getting work done - Fast.

Oddly enough, the same OS is also perfect where servers are concerned, and I don't have to pay ridiculous licensing fees.
 
Last edited:
Not at all. There's examples of 'bad porting jobs where DX to OGL or even Vulkan wrappers' are concerned, but level up the playing field with a title using a common API such as Vulkan and in most cases Linux is faster.

We do need to see more of that 'playing field leveling'; fortunately, most major engines are supporting Vulkan with all modern GPU features. And we need to see more of it to see if there are issues. And we still have to ask the questions about what is causing the delta.

Odd architectures? NUMA has been around since the 1990's!

Not in desktops that Windows targets, which is the point- and the assumption that MS won't eliminate the disparity is faulty.

The fact you don't have a problem with the Windows 10 updater indicates a case of denial as opposed to no real issue with one of the most poorly thought out updating systems known.

It's a system designed for users that are not involved in their systems. For that, it works extremely well. Perhaps too well given its goals, which are not terribly considerate of users.

For me, the best desktop tool for the job is not the one with a fragmented within the one UI interface, shockingly poor updater, lack of control to the end user suffering from cryptolockers and malware at the rate of 12 million strains a month. I'll steer clear of the overheads regarding antivirus and I'll focus on getting work done.

The more Linux is used on desktops, the more it will become a target as well...

Oddly enough, the same OS is also perfect where servers are concerned, and I don't have to pay ridiculous licensing fees.

You absolutely would if you were paying for the level of support Microsoft provides.
 
We do need to see more of that 'playing field leveling'; fortunately, most major engines are supporting Vulkan with all modern GPU features. And we need to see more of it to see if there are issues. And we still have to ask the questions about what is causing the delta.

Well the trend is that running Vulkan titles perform better under Linux. So you believe it's a matter of the graphics subsystem and I'll accept certain realities that are evident outside gaming.

Not in desktops that Windows targets, which is the point- and the assumption that MS won't eliminate the disparity is faulty.

We've been using NUMA implementations on the desktop since ~2008, my main workstation is a 2009 model and it's an option in the BIOS. Chances are we've been using NUMA since the end of SMP with the Pentium III.

It's a system designed for users that are not involved in their systems. For that, it works extremely well. Perhaps too well given its goals, which are not terribly considerate of users.

It's a massive inconvenience to the end user and it's intention is to allow Microsoft to regain control of the OS 'as a service'.

The more Linux is used on desktops, the more it will become a target as well...

Globally, Android is the most popular OS around, eclipsing Windows. Yet malware/virus infections are a minuscule fraction compared to the desktop OS people use simply because it's installed on the machine they purchased by default. The popularity argument doesn't hold water based on this very fact alone.

You absolutely would if you were paying for the level of support Microsoft provides.

I just spat my coffee everywhere! You think MS offers good service regarding their server products? Microsoft's level of service is roughly in line with that of Adobe (ie: Garbage).
 
Well the trend is that running Vulkan titles perform better under Linux. So you believe it's a matter of the graphics subsystem and I'll accept certain realities that are evident outside gaming.

And these are few and far between. Not enough to show a trend; more showing how innefficient OpenGL still is for modern gaming.

We've been using NUMA implementations on the desktop since ~2008, my main workstation is a 2009 model and it's an option in the BIOS. Chances are we've been using NUMA since the end of SMP with the Pentium III.

Aside from multi-socket implementations, which again were largely not targeted by Windows. That's the point. Further, AMD's implementation as seen in the Threadripper is new; the bandwidth between dies is actually fast enough to bother trying to sync workloads across blindly. That hasn't been the case with the multi-socket implementations that NUMA is typically used for.

It's a massive inconvenience to the end user and it's intention is to allow Microsoft to regain control of the OS 'as a service'.

Let them be inconvenienced. End users refusing to update their systems are the biggest threat to overall internet health. My only complaint is that MS won't release LTSC versions wider.

Globally, Android is the most popular OS around, eclipsing Windows. Yet malware/virus infections are a minuscule fraction compared to the desktop OS people use simply because it's installed on the machine they purchased by default. The popularity argument doesn't hold water based on this very fact alone.

And Android has a different hardware/software paradigm, just as Apple's iOS, both for far more limited hardware. The Play Store is full of it still. It's a very bad comparison. As for 'popularity', well, the Linux desktop is so close to zero as to be statistically insignificant. It's growing, of course, but that's the point- if the 'year of the Linux desktop' ever arrives, expect plenty of malware to be headed that way.
 
And these are few and far between. Not enough to show a trend; more showing how innefficient OpenGL still is for modern gaming.

They're not 'few and far between' at all! As for OGL, what's highlighted is how inefficient OGL is under Windows. Once again, Linux is the faster OS.

Aside from multi-socket implementations, which again were largely not targeted by Windows. That's the point. Further, AMD's implementation as seen in the Threadripper is new; the bandwidth between dies is actually fast enough to bother trying to sync workloads across blindly. That hasn't been the case with the multi-socket implementations that NUMA is typically used for.

So...You're trying to claim that Microsoft wasn't interested in multi socket NUMA implementations under Windows because bandwidth between dies wasn't fast enough?! The bandwidth between dies on my circa 2009 workstation appears more than ample running Linux. I think you're reaching here.

Let them be inconvenienced. End users refusing to update their systems are the biggest threat to overall internet health. My only complaint is that MS won't release LTSC versions wider.

Actually, forced updates are doing nothing to stop the spread of viruses, trojans and malware under Windows 10 - Internet health is not a problem as the internet runs under Linux. Forcing updates has little to do with 'security', forcing updates was Microsoft's solution to 'Windows as a service'.

And Android has a different hardware/software paradigm, just as Apple's iOS, both for far more limited hardware. The Play Store is full of it still. It's a very bad comparison. As for 'popularity', well, the Linux desktop is so close to zero as to be statistically insignificant. It's growing, of course, but that's the point- if the 'year of the Linux desktop' ever arrives, expect plenty of malware to be headed that way.

Android is globally far more 'popular' an OS than Windows, you were arguing that the popularity of Windows is the main reason for it being targeted by viruses, malware and trojans. The fact that Android suffers a minuscule amount of these issues in direct comparison to Windows highlights that infection issues under Windows have more to do with security issues than outright popularity. It's actually highlighted so effectively that it's literally pointless arguing the topic.
 
They're not 'few and far between' at all! As for OGL, what's highlighted is how inefficient OGL is under Windows. Once again, Linux is the faster OS.

OpenGL is innefficient period- it's pretty outdated with respect to current hardware and needs of software.

So...You're trying to claim that Microsoft wasn't interested in multi socket NUMA implementations under Windows because bandwidth between dies wasn't fast enough?! The bandwidth between dies on my circa 2009 workstation appears more than ample running Linux. I think you're reaching here.

Interested vs. targeted. Linux is targeted towards servers of all types at its core, whereas desktop Windows targets desktops where NUMA CPUs with multiple die per socket were essentially non-existent until the Threadripper line was released.

Actually, forced updates are doing nothing to stop the spread of viruses, trojans and malware under Windows 10 - Internet health is not a problem as the internet runs under Linux. Forcing updates has little to do with 'security', forcing updates was Microsoft's solution to 'Windows as a service'.

We're both right here. Yes, 'Windows as a service' is an apparent goal, but security is also something Microsoft has been tackling rather directly. And yes, forcing security updates helps stop the spread of botnets, which directly threaten internet health and internet users.

Android is globally far more 'popular' an OS than Windows, you were arguing that the popularity of Windows is the main reason for it being targeted by viruses, malware and trojans. The fact that Android suffers a minuscule amount of these issues in direct comparison to Windows highlights that infection issues under Windows have more to do with security issues than outright popularity. It's actually highlighted so effectively that it's literally pointless arguing the topic.

More, I'm arguing that Windows is a more popular desktop operating system, and has been around for a very long time in that role, which it still dominates. Of course it's targeted more. Desktop systems, running Windows or otherwise, also represent a significant resource for botnets, and Windows is a far more versatile operating system than Android, by design. Apple iOS is even less flexible, again by design. They make poor comparisons to desktop operating systems.

This isn't 'Windows good Linux bad', this is Windows and Linux are different, and with respect to the OP, the question that is raised is directed to how that difference is affecting performance and creating a delta and how that can be applied to the Windows version.
 
OpenGL is innefficient period- it's pretty outdated with respect to current hardware and needs of software.

That is untrue. Is DirectX 11 dated for current hardare ?

Open GL 4.6 (the current version) is far newer then DX 11... and it in fact has features MS reserves for DX 12.

https://en.wikipedia.org/wiki/OpenGL
https://en.wikipedia.org/wiki/Feature_levels_in_Direct3D

GL 4.6 introduced SPIRV as and example which was added to DX 12.1

OpenGL / Vulkan .... I would say is like DX 11 / 12.

But that isn't exactly true. Khronos is doing tings the proper way around. Lots of gamers keep saying why are they still updating OpenGL and advocating its use... why not just use vulkan all the time for everything. The truth is they have different uses and there are cases where OpenGL is superior. The advantage of opengl is its a high level shader API. The reason almost no games use DX12 is because its only real advantage is being able to take more low level control, which has dubious advantage in most caes, and is much harder to deal with. Frankly few programers have the chops to deal wiht low level optimizations.

Vulkan still has enough high level hooks that its a bit easier to deal with... and it seems ot be gaining more traction. Still at the end of the day... even for games Vulkan isn't going to be massively faster then well coded openGL.

OpenGL > Dreictx 11
The issue with games... I don't get the feeling that Nvidia or AMD and Intel really spend as much time optimzing OpenGL windows drivers. They focus optimizations more on the professional software that uses it which require absolute precision over speed tweaks. The windows world has falling into that pattern... DX for games, OGL for pro stuff. I guess its made easy on compaies like Nvidia. Speed tweaks go here, quality precision tweaks go here.
Under Linux of course... its a completely different game, and the OpenGL drivers at least of AMD and Intel are out of the MFGs hands.
 
More, I'm arguing that Windows is a more popular desktop operating system, and has been around for a very long time in that role, which it still dominates. Of course it's targeted more. Desktop systems, running Windows or otherwise, also represent a significant resource for botnets, and Windows is a far more versatile operating system than Android, by design. Apple iOS is even less flexible, again by design. They make poor comparisons to desktop operating systems.

This isn't 'Windows good Linux bad', this is Windows and Linux are different, and with respect to the OP, the question that is raised is directed to how that difference is affecting performance and creating a delta and how that can be applied to the Windows version.

Maz is correct. Its used more is untrue. Its targeted more is not.

Android being a simpler device should make an even bigger target for bot net type attacks. I mean IOT things get targeted in that way as well... not because there complicated or simple, but because there unsecure.

Windows is targeted, for the same reason a Lion isn't going to chase down the fastest gazelle. The weak and sickly are targeted first.

Windows has shite user security. Windows doesn't enforce Admin privilege on users... and it operates to many easy to exploit kernel level access processes directly in user space. Bottom line windows is easier to hack. Not because its been around longer, but because its insecure. With Linux a hacker doesn't even have to guess about code hooks or find exploitable memory leaks ect... the code is wide open. Any nefarious hacking genius can find the bugs and exploit them. IF the system allows it... which in most cases it does not. Its not like there aren't more juicy Linux targets... Android for users, and what 90% of the webs servers, the entire cloud, and every big data server everywhere is on Linux or Unix.
 
i am ready to game under linux if performance advantage is enough to make an unplayable title playable.
i doubt MS puts much effort on improving windows performance, as they have their hands full patching security holes.
 
I used to run World of Warcraft on my AMD card using OpenGL instead of DirectX. However, back then, Blizzard cared about OpenGL (most of their games were designed to be cross-platform) and AMD used to care about desktop OpenGL performance.
 
Android being a simpler device should make an even bigger target for bot net type attacks. I mean IOT things get targeted in that way as well... not because there complicated or simple, but because there unsecure.

The devices are simpler and the applications are far more limited. And unlike nearly all IoT devices, Android has actual security updates. Incidentally, Microsoft is leading the charge in securing IoT.

Windows is targeted, for the same reason a Lion isn't going to chase down the fastest gazelle. The weak and sickly are targeted first.

Windows was and is targeted for different reasons over time, most of which don't apply to desktop Linux because the year of desktop Linux has literally never arrived. In the time that Windows has existed and internet access has become the dominant focus of desktop computing, Microsoft has made great strides to secure the operating system and keep up with attackers.

And remember all of those IoT devices that have now become so popular as targets of attacks? They're running Linux.

And that's important, because it shows that the infrastructure around an operating system is extremely important with respect to securty.
 
Back
Top