AMD Radeon Software Crimson Edition Is A Letdown On Linux

Regarding the effectiveness of webcounters, here's the webcounter off one of my own websites, does anyone see the massive number of issues here?

 
The green and grey should be swapped?

Well, first of all I'd like to know just who out there is running Unix! You could argue that OSX is based around FreeBSD which is Unix, but there's a Macintosh section there - So that blows that theory out the window! You could assume that this is supposed to represent Linux, but Linux isn't even a fork of Unix!

Secondly, roughly 30% of that pie graph is 'OS Unknown' - That blows the accuracy of the web counter out the window with the force of a fruit canon! Perhaps that's supposed to represent Linux? Who knows? Perhaps both OS Unknown and Unix represent Linux combined?

Anyone even vaguely attempting to base the popularity of any OS based on web counters may as well just be fishing numbers out of a hat they're so inaccurate. As stated in the past, due to it's distribution methods, we have absolutely no idea what percentage of users out there are running Linux.
 
Last edited:
So why are AMD drivers not as good as Nvidia drivers, my take:
  • AMD can not justify the resources it will take to have better drivers on the level of Nvidia
  • The Linux community are not able to developer open AMD drivers even as good as AMD proprietary drivers with AMD scarce resources
  • Linux development community had many years to figure out and to perfect AMD drivers - they have not
  • My conclusion is the Linux open community development ability is not sufficient or able to develop state of the art AMD drivers for Linux GPU's.

Nvidia has no open drivers, meaning Nvidia needs very little support from the Linux open development community (I am sure they get great support from the super computer folks, rendering farm folks.)

I also believe that Nvidia shows that Linux can be very effective if you put adequate well thought out resources and controls. The idea that open source code over time due to vast access will be better then closed source code is false. It depends on too many factors then just being open source to make it the best code.
 
Last edited:
So why are AMD drivers not as good as Nvidia drivers, my take:
  • AMD can not justify the resources it will take to have better drivers on the level of Nvidia
  • The Linux community are not able to developer open AMD drivers even as good as AMD proprietary drivers with AMD scarce resources
  • Linux development community had many years to figure out and to perfect AMD drivers - they have not
  • My conclusion is the Linux open community development ability is not sufficient or able to develop state of the art AMD drivers for Linux GPU's.

Nvidia has no open drivers, meaning Nvidia needs very little support from the Linux open development community (I am sure they get great support from the super computer folks, rendering farm folks.)

I also believe that Nvidia shows that Linux can be very effective if you put adequate well thought out resources and controls. The idea that open source code over time due to vast access will be better then closed source code is false. It depends on too many factors then just being open source to make it the best code.

No open source driver has ever been ideal for gaming under Linux. That doesn't mean the open source drivers are garbage, they were simply never designed with gaming in mind.
 
So why are AMD drivers not as good as Nvidia drivers, my take:
  • AMD can not justify the resources it will take to have better drivers on the level of Nvidia
  • The Linux community are not able to developer open AMD drivers even as good as AMD proprietary drivers with AMD scarce resources
  • Linux development community had many years to figure out and to perfect AMD drivers - they have not
  • My conclusion is the Linux open community development ability is not sufficient or able to develop state of the art AMD drivers for Linux GPU's.

Nvidia has no open drivers, meaning Nvidia needs very little support from the Linux open development community (I am sure they get great support from the super computer folks, rendering farm folks.)

I also believe that Nvidia shows that Linux can be very effective if you put adequate well thought out resources and controls. The idea that open source code over time due to vast access will be better then closed source code is false. It depends on too many factors then just being open source to make it the best code.

There are open Nvidia drivers for linux. Just not for the newest cards as the firmware is encrypted. Nvidia is extremely open source unfriendly in that regard.

IMHO the difference in drivers comes down to the complexity and design of the hardware. Open source Nvidia developers have stated the hardware is extremely easy to create drivers on. They just can't do it because of a lack of firmware.

Catalyst drivers on linux aren't all that bad compared to windows. in fact earlier in the year they were often faster. The windows catalyst drivers weren't exactly great compared to Nvidia either. That seems to come down more to the architecture than anything else. Look at the current situation we're starting to see with DX12 and likely Vulkan. Both are derived from mantle and a highly asynchronous interface supporting lots of threads. You see AMD get huge gains from the new APIs while Nvidia remains roughly stagnant. Then you have game developers like Studio Wildcard pushing back a DX12 release because of "driver issues" the day prior they didn't know existed. Add in Oxide's spat with Nvidia. It's a little strange to have "driver issues" and devs demoing multi adapter configurations on Vulkan at the same time. I'm guessing the delays have more to do with performance attributed to hardware design than anything else. It's those very design decisions that likely affect driver development. APIs like DX9/10/11/OpenGL map well to Nvidia while DX12/Vulkan map better to AMD.

The current open source push by AMD I think has more to do with the linux development model than anything else. Every time the kernel or display manager makes a major change AMD would be required to release a new supporting driver. Unlike Windows, each distribution could be doing something entirely different. Keeping the drivers open allows the distribution maintainers to recompile the drivers based on their current choices. Those drivers should also become far simpler with Vulkan and easier to maintain.
 
There are open Nvidia drivers for linux. Just not for the newest cards as the firmware is encrypted. Nvidia is extremely open source unfriendly in that regard.

IMHO the difference in drivers comes down to the complexity and design of the hardware. Open source Nvidia developers have stated the hardware is extremely easy to create drivers on. They just can't do it because of a lack of firmware.

Catalyst drivers on linux aren't all that bad compared to windows. in fact earlier in the year they were often faster. The windows catalyst drivers weren't exactly great compared to Nvidia either. That seems to come down more to the architecture than anything else. Look at the current situation we're starting to see with DX12 and likely Vulkan. Both are derived from mantle and a highly asynchronous interface supporting lots of threads. You see AMD get huge gains from the new APIs while Nvidia remains roughly stagnant. Then you have game developers like Studio Wildcard pushing back a DX12 release because of "driver issues" the day prior they didn't know existed. Add in Oxide's spat with Nvidia. It's a little strange to have "driver issues" and devs demoing multi adapter configurations on Vulkan at the same time. I'm guessing the delays have more to do with performance attributed to hardware design than anything else. It's those very design decisions that likely affect driver development. APIs like DX9/10/11/OpenGL map well to Nvidia while DX12/Vulkan map better to AMD.

The current open source push by AMD I think has more to do with the linux development model than anything else. Every time the kernel or display manager makes a major change AMD would be required to release a new supporting driver. Unlike Windows, each distribution could be doing something entirely different. Keeping the drivers open allows the distribution maintainers to recompile the drivers based on their current choices. Those drivers should also become far simpler with Vulkan and easier to maintain.

All very valid points, and I, for one, am all in favour of Vulkan as an API and I hope it gains traction within the industry.

One part of your quote I find confusing however is this one:

Every time the kernel or display manager makes a major change AMD would be required to release a new supporting driver. Unlike Windows, each distribution could be doing something entirely different. Keeping the drivers open allows the distribution maintainers to recompile the drivers based on their current choices.

As Nvidia don't seem to have any issues regarding this situation with their proprietary driver? I think Debian based distributions have pretty much been chosen in favour as gaming distro's anyway, possibly simplifying this issue?
 
As Nvidia don't seem to have any issues regarding this situation with their proprietary driver? I think Debian based distributions have pretty much been chosen in favour as gaming distro's anyway, possibly simplifying this issue?

The situation still exists, Nvidia is just better about updating the driver prior to any release. With linux someone could fork the entire distribution and make significant changes requiring an IHV to release a driver specifically for it. Unlikely, but possible. It's more problematic for the open source development where you don't have "in house" operations. Provided source for the driver, each distribution could compile it to run on their platform.
 
The situation still exists, Nvidia is just better about updating the driver prior to any release. With linux someone could fork the entire distribution and make significant changes requiring an IHV to release a driver specifically for it. Unlikely, but possible. It's more problematic for the open source development where you don't have "in house" operations. Provided source for the driver, each distribution could compile it to run on their platform.

While Nvidia's drivers are propitiatory, which most likely doesn't go down well with the purists, you have to hand it to them, they make a pretty good Linux driver.
 
General reminder time here: If you're talking about the technical merits of your or someone else's statement that's fine. If you're directly discussing another forum member and your personal opinion of them it's probably not.

Sample good post: "I disagree because...."
Sample bad post: "You suck"

Keep it civil please. If you don't agree with someone say why otherwise don't post.
 
I also believe that Nvidia shows that Linux can be very effective if you put adequate well thought out resources and controls. The idea that open source code over time due to vast access will be better then closed source code is false. It depends on too many factors then just being open source to make it the best code.

Well considering driver programming is very specialised performance low level programming there's only so few people who can make that code and most of them are probably extremly well paid specialists so who knows how many of them want to do volunteer work for community.
 
yes that is definitely an issue for open source drivers, the level of expertise needed to create them is very high. AMD can get around this by supporting the community more as this will help people with less experience understand the software building process better and technical information availability increases which will fill in the gaps for the lack of programming experience (of course it will still take longer for less experienced programmers but they will still be able to do the same work).
 
yes that is definitely an issue for open source drivers, the level of expertise needed to create them is very high. AMD can get around this by supporting the community more as this will help people with less experience understand the software building process better and technical information availability increases which will fill in the gaps for the lack of programming experience (of course it will still take longer for less experienced programmers but they will still be able to do the same work).

The problem with open source gaming drivers is the fact that most game code is poorly optimised, Nvidia work closely with developers to try and patch these issues within their proprietary driver set to try and overcome these issues - It's one of the reasons their drivers work so well under Linux. This is going to be difficult, if not impossible on a purely open source driver.
 
The problem with open source gaming drivers is the fact that most game code is poorly optimised, Nvidia work closely with developers to try and patch these issues within their proprietary driver set to try and overcome these issues - It's one of the reasons their drivers work so well under Linux. This is going to be difficult, if not impossible on a purely open source driver.

Which begs the question, "Where are the super Linux well optimized AAA games which are also open sourced where the community just digs on in and improves even better"? How many good titles came out of the open source Doom 3 engine (ID tech 4)? Released in 2011.

I think razor1 is correct, the amount of talent it takes on some tasks can be astronomical compared to a decade or more ago. With the constant hardware updates/progression particularly GPU's it does take some rather specialized talent which is probably hard to find, keep with some rather big money incentives. For AAA games the budgets are 20 million $ plus. Grand Theft Auto was over 100 million $ just for the consoles. Even then there are big issues and problems.

If open source could duplicate even a little of that level of programming and success I would think it would be here already. It is not. The effort, resources, dedication (as in one can solely concentrate on an endeavor) for open source success at that level I see not likely. Just like GPU drivers, it takes a very high level of commitment, financing, experience etc. over and over again to make it even workable. Even then big mistakes are done and need corrected.

I do believe for Linux to really take off, Linux needs to grown more into a commercial type environment. Having almost countless distributions, updating to the next version before you can even get comfortable with the current is not helping matters - very unstable I would think. Many folks have been with Windows 7 for over 5 years, not having to constantly start from scratch with new major bugs, options, old ones going away, hardware/software conflicts. I think Valve, promoting a new platform for games is a good step in that direction. Android for cell phones another good example.
 
Last edited:
Linux doesn't need to rely to FOSS, in other words, the OS is open source, but supporting software can very well be propitiatory - Including software such as Nvidia's excellent drivers.

What we need is an open API such as Vulkan, an open API would make porting between all platforms far easier.

Furthermore, I don't understand where people get the 'countless distributions' idea from? As this just isn't the case. And Debian based distro's such as Ubuntu and Linux Mint as well as SteamOS have pretty much been settled on as the Linux gaming platform of choice.

The open source community just need to maintain the distro, nothing else. Everything else can be propitiatory. You aren't going to run open source drivers in order to play games effectively.

Linux is by no means unstable nor does it suffer the dreaded 'performance rot' of Windows. LTS versions of Linux, such as Linux Mint, have major overhauls/upgrades about as often as Windows - So the idea of not being able to get used to a distro before it's pulled in favour of an upgraded one with a totally different interface is incorrect.
 
Last edited:
The "countless distributions" is just the nature of an open source development model. Any time someone wants to try something different, or thinks the current guy running the show is an ass, they fork the project. In most cases they are still compatible with each other, but there will be differences. It's evolution applied to software development.

For AAA titles, anything that wasn't exclusively DirectX was probably ported. Metros, Bioshocks, Shadow of Mordor, Total Wars, anything through Paradox Interactive, Valve (obviously), and the list goes on. Steam Linux Games

There are some open source games like 0ad, but you need to keep in mind what open source development is doing. In most cases the open source product isn't what is being sold. It's just a component of something else with shared development of everyone using it. The linux kernel is developed by paid employees from almost every major software and hardware company. They do this because linux is running on their servers and and they need to fix a bug or improve features and performance. In other cases you get a grad student writing a thesis. Those changes then benefit everyone. That's why you will find open source tools to build games, including APIs, but not see an actual game.
 
The "countless distributions" is just the nature of an open source development model. Any time someone wants to try something different, or thinks the current guy running the show is an ass, they fork the project. In most cases they are still compatible with each other, but there will be differences. It's evolution applied to software development.

For AAA titles, anything that wasn't exclusively DirectX was probably ported. Metros, Bioshocks, Shadow of Mordor, Total Wars, anything through Paradox Interactive, Valve (obviously), and the list goes on. Steam Linux Games

There are some open source games like 0ad, but you need to keep in mind what open source development is doing. In most cases the open source product isn't what is being sold. It's just a component of something else with shared development of everyone using it. The linux kernel is developed by paid employees from almost every major software and hardware company. They do this because linux is running on their servers and and they need to fix a bug or improve features and performance. In other cases you get a grad student writing a thesis. Those changes then benefit everyone. That's why you will find open source tools to build games, including APIs, but not see an actual game.

Well I hope those ports have made money plus in the future released more in time with the Windows version. I also hope the Linux community supports game development for their platform - as in buying games especially any new titles. Now if Linux had some exclusive titles :D, like Half Life 3 or come out first on Linux before moving onto the Consoles and Windows. I see Valve doing this, releasing like Half Life 3 on Linux Steam Box and a couple months later everything else.
 
I wholeheartedly agree with the HL3 exclusive to Linux marketing! Valve need to do this, even if it's just exclusive to Linux for six months!

Valve games perform way better under Linux using Nvidia hardware and drivers than they do under Windows anyway - This is the kind of kick Linux needs. :)
 
Not going to happen. All that would achieve is hurt Half-Life 3 sales. Maybe you will see some kind of exclusive content for SteamOS, but even that seems unlikely.

Remember how Microsoft originally tried to keep Office off iOS/Android in a bid to improve Windows Phone marketshare. It didn't help at all, but instead resulted in 1+ billion personal computing devices without Microsoft Office.
 
someone found it listed in a steam database, could be just a placeholder however.
 
Not going to happen. All that would achieve is hurt Half-Life 3 sales. Maybe you will see some kind of exclusive content for SteamOS, but even that seems unlikely.

Remember how Microsoft originally tried to keep Office off iOS/Android in a bid to improve Windows Phone marketshare. It didn't help at all, but instead resulted in 1+ billion personal computing devices without Microsoft Office.

SteamOS is a Debian derivative based on Ubuntu. if you release it for Steam OS you release it for Linux, pointless marketing it to an even smaller audience by somehow artificially limiting it to SteamOS only.

And I'm still not running MS Office on my Android device even though I've got a licence, no need to, the alternatives work fine considering it's a phone/tablet. Point being, very few people need MS office on their mobile device, hence it had little leverage as a marketing tool.

I don't know a single person that runs Office on their iOS/Android device. I tried exchange on the wife's Nexus 9 to work with the exchange server at her work (5 PC's, 1 x Windows SBS 2011 server, that's value for money and a nice central point of failure!), it was rubbish! Ended up finding a free alternative that suited her needs perfectly, as the Google Gmail app is useless with Exchange.

In comparison, as far as single player games go, the entire Half Life series is epic - Possibly even better than many single player games available today. I lost myself in Half Life 1, Half Life 2 and spin off's and recently Black Mesa, to this day they're the only single player games I ever played from start to finish without cheating as they maintained my interest and allowed me to loose myself in the game. I think HL3 exclusively for Linux, even if it was just for six months, would hold significant leverage and I'd be surprised if Valve aren't already thinking of it. Even after six months has expired, due to the fact that Valve games perform better under Linux than they do under Windows on Nvidia hardware/drivers the game could still hold significant leverage for Linux.

Oh, I remember half way through playing HL2 I got my first DX9 card, god it looked good compared to DX8!
 
Last edited:
Well I hope those ports have made money plus in the future released more in time with the Windows version. I also hope the Linux community supports game development for their platform - as in buying games especially any new titles.
If you look at pretty much any humble bundle you'll see linux users averaging higher per purchase prices than windows users. Not as many of them sure, but they are worth more.

Ex: https://www.humblebundle.com/
 
Back
Top