AMD's Open-Source Radeon Linux Driver: 2014 Was Incredible

[pingufunkybeat]

so higher fps usually indicates less stutter.

No, they are not related.

Low fps means low fps -- your hardware (or your driver) is too slow and renders each frame slowly.

Stutter means that you have inconsistent fps. Sometimes it's fast, sometimes it's slow. It will go fast for a while, then get stuck, then be fast again.

You can have high fps and stutter, and you can also have low fps, but stutter-free.

[mmstick]

so higher fps usually indicates less stutter.

Rather, higher FPS indicates more stutter. IE: Catalyst may appear to have higher FPS, but it stutters often. Meanwhile, while FOSS drivers have historically demonstrated lower average framerate, they have always given a smoother experience.

[RedRaider]

These test results are STILL inconclusive

I have two problems with this article.

1. To limit this to just one distribution is very short sighted. If Phoronix testers are the gurus they pretend to be, they would be able to use more than just Ubuntu, frankly, what I call kiddy Linux, since practically everyone with the lowest Linux IQ apparently can use it. How about doing testing on more demanding distributions for more advanced users? It ought to be easy for the "experts" at Phoronix right? Me myself am using Slackware 14.1 64 bit multilib, KDE, AMD Phenom II X4 965, AMD Radeon FX 6950 in Crossfire. I am using Linux Native Steam and PlayOnLinux (for stuff you can't play through Steam) as my main game software.

2. These results are inconclusive because it does not include an AMD based system. If you REALLY want to be conclusive, you would run the same tests using Nvidia and AMD cards on both an Intel and AMD system. How do you know that you don't get MORE stutter with Nvidia cards than AMD on an AMD system or that the stuttering will no longer be a factor? YOU WILL NOT. And there is PLENTY of suspicion concerning Intel that there may be a conflict of interest since it can EASILY be surmised that Intel is NOT going to do anything to make their primary competitor, WHICH IS AMD (which they have dealt shadily with AMD in the past), look good and will target their development favoring Nvidia cards.

Not only that, people talk about "micro-stuttering" with AMD Crossfire, and on reviews I have read they are using Intel based systems as their test system (Um, isn't this article mentioning stuttering...I wonder why??? (see below)). Apparently that happens quite frequently yet I use an AMD based system and ALL of my graphics are perfectly smooth. WHY IS THAT? COULD IT BE I'M NOT USING INTEL?!?! So that means these are your choices. Use Intel and AMD and get marginally better, most likely imperceptible, speed bump and have stuttering graphics, OR DO WHAT AMD suggests, use all AMD and keep your graphics smooth and more than likely not see any perceptible performance difference, OR stick with Intel and Nvidia. But I say, personally, the improvements are not prevalent except on paper benchmarks and are not perceptible in real world perception and most likely cost more, which you have every right to do and this is a free country for you to do that but also free for me to question this testing methodology. If you think it makes a perceptible difference in what you do with your computer using Nvidia and Intel, then more power to you, but just sayin' (don't go off getting defensive and calling "fanboy" names, which many people (from my experience, mainly Intel fans) are wont to do, just because I am questioning this).


Until you do what I suggest and what AMD suggests, your results are inconclusive and cannot be taken with more than a grain of salt due to incomplete and superficial (as in not thorough), testing.

[duby229]

Yep, I agree with you. That's why subjective analysis is important. Anybody who authors a performance comparison is entitled to their own opinion. That's why how a game "feels" is soo important.

[Hamish Wilson]

Until you do what I suggest and what AMD suggests, your results are inconclusive and cannot be taken with more than a grain of salt due to incomplete and superficial (as in not thorough), testing.

Even if we do agree with your assertion that Intel is hobbling the AMD performance in these tests, it would not change the main conclusion of the article which is that the free software radeon drivers have improved greatly over the past year.

[darkbasic]

I have two problems with this article.

1. To limit this to just one distribution is very short sighted. If Phoronix testers are the gurus they pretend to be, they would be able to use more than just Ubuntu, frankly, what I call kiddy Linux, since practically everyone with the lowest Linux IQ apparently can use it. How about doing testing on more demanding distributions for more advanced users? It ought to be easy for the "experts" at Phoronix right? Me myself am using Slackware 14.1 64 bit multilib, KDE, AMD Phenom II X4 965, AMD Radeon FX 6950 in Crossfire. I am using Linux Native Steam and PlayOnLinux (for stuff you can't play through Steam) as my main game software.

2. These results are inconclusive because it does not include an AMD based system. If you REALLY want to be conclusive, you would run the same tests using Nvidia and AMD cards on both an Intel and AMD system. How do you know that you don't get MORE stutter with Nvidia cards than AMD on an AMD system or that the stuttering will no longer be a factor? YOU WILL NOT. And there is PLENTY of suspicion concerning Intel that there may be a conflict of interest since it can EASILY be surmised that Intel is NOT going to do anything to make their primary competitor, WHICH IS AMD (which they have dealt shadily with AMD in the past), look good and will target their development favoring Nvidia cards.

Not only that, people talk about "micro-stuttering" with AMD Crossfire, and on reviews I have read they are using Intel based systems as their test system (Um, isn't this article mentioning stuttering...I wonder why??? (see below)). Apparently that happens quite frequently yet I use an AMD based system and ALL of my graphics are perfectly smooth. WHY IS THAT? COULD IT BE I'M NOT USING INTEL?!?! So that means these are your choices. Use Intel and AMD and get marginally better, most likely imperceptible, speed bump and have stuttering graphics, OR DO WHAT AMD suggests, use all AMD and keep your graphics smooth and more than likely not see any perceptible performance difference, OR stick with Intel and Nvidia. But I say, personally, the improvements are not prevalent except on paper benchmarks and are not perceptible in real world perception and most likely cost more, which you have every right to do and this is a free country for you to do that but also free for me to question this testing methodology. If you think it makes a perceptible difference in what you do with your computer using Nvidia and Intel, then more power to you, but just sayin' (don't go off getting defensive and calling "fanboy" names, which many people (from my experience, mainly Intel fans) are wont to do, just because I am questioning this).


Until you do what I suggest and what AMD suggests, your results are inconclusive and cannot be taken with more than a grain of salt due to incomplete and superficial (as in not thorough), testing.

I never read so much bullshit in a single post.

[zoomblab]

2014 Was Incredible shit (as always)

I am froustrated once again with the useless AMD drivers whether closed or open source. I bought this 7870 about 2,5 years ago, because of the promise of the open source drives, but untill today I get nothing but troubles every time. The latest incident. I am updating my kubuntu 14.04 lts, kernel 3.13.0-43 comes and the system stops booting. Great for once more the shitty AMD drivers are the cause of the problem (this time it is the open drivers but they both suck).

Dear AMD developers, when ever you even think about making commits that affects the kernel, better be 200% sure that it won't have negative side effects like making systems unbootable. You are the manufacturer. You have all the hardware at your disposal. Make proper QA before pushing your freaking commits. I am not your beta tester. This will definitely be my last AMD hardware.

[dungeon]

The latest incident. I am updating my kubuntu 14.04 lts, kernel 3.13.0-43 comes and the system stops booting. Great for once more the shitty AMD drivers are the cause of the problem (this time it is the open drivers but they both suck).

That is more of a kernel maintainer problem, both 3.13 and 3.16 kernels mentioned there are not longterm kernels on kernel.org ... so that issue is ubuntu kernel only, not stright an AMD's problem.

[bridgman]

Dear AMD developers, when ever you even think about making commits that affects the kernel, better be 200% sure that it won't have negative side effects like making systems unbootable. You are the manufacturer. You have all the hardware at your disposal. Make proper QA before pushing your freaking commits. I am not your beta tester. This will definitely be my last AMD hardware.

zoomblab, I don't think we push commits directly (or even indirectly) to post-release distro trees (ie updates). We do work with distro developers pre-release to get new features aligned with distro kernel plans, but post-release I don't believe the distro devs usually update the GPU drivers. Not to downplay the problem, but at first glance this looks like a change somewhere else in the kernel tree which caused problems for gfx drivers.

[bridgman]

Stupid 5 minute edit limit...

I just checked the change logs for 3.13.0-43 and found 88 drm/radeon changes, certainly more than I expected. Not sure what the baseline for the changelog is -- I assumed previous 3.13.0-xx but maybe that's not right. That seems higher than I would expect for a post-release kernel update, will look at the changelogs for earlier Ubuntu kernels & see...

Hmm... change log link for previous 3.13.0-xx versions (eg 3.13.0-32 generic) points to the 3.13.0-43 changelog, so not sure how to find what changed between 3.13.0-32 and 3.13.0-43. Stay tuned.