Hey people, got a really nice graph to show off today. On a follow up from my last article, loads more work has been done to Mesa and the Radeon driver, and the speed improvement from 10.0.3 to 10.2 is phenomenal.
Before we begin here are the benchmark conditions:
What the benchmark can't help you with:
What this benchmark can help you with:
So with that out of the way, lets examine the results.
These results are taken from the same demo files and settings with only the kernel version and mesa version being changed. You can also see the addition of Unigine Valley which did not actually run on the 10.0 drivers which is a big step up in itself. The other games ran from 46% increases all the way to 462% which I am pretty sure must be a bug in Portal or something because I did not believe my eyes even after running repeated tests :) .
The majority of the source games ran with 2xAA so there is an increase to be seen there when it used to limit the games quite a bit but now it seems its no longer an issue as although not benchmarked here, I can comfortably run each game at a minimum of 40fps now. From these results, its no wonder that AMD wants to drop its catalyst driver for FOSS principles when you see results like this that catalyst just isn't getting on it's own.
As with all Linux projects, 10.3 is already in full swing and will be benchmarked as soon as the arch repository's update with it. For those of you that want the action now need to make sure your kernel is as up to date as your distro can be while still being stable and some distributions have third party providers to newer mesa versions. Arch and its derivatives can download the latest mesa from its repositories along with the latest 3.15 kernel.
In my own opinion we are accelerating at such a speed with these drivers, especially with the news that valve are now supporting its development.
If you would like to see the results, demo files for your own benchmarks, or the system specifications then visit this Google Docs page. Article taken from GamingOnLinux.com.
Before we begin here are the benchmark conditions:
What the benchmark can't help you with:
- Showing a great detail on how each card has improved.
- Comparing between FGLRX (proprietary) and Radeon (FOSS).
- Performance in extremely high intensive tasks beyond gaming.
What this benchmark can help you with:
- Showing in good detail how performance has been increased in mesa as a whole.
- What driver you should choose on your next update.
- The general trend in progress, and how the acceleration of development has increased.
So with that out of the way, lets examine the results.
These results are taken from the same demo files and settings with only the kernel version and mesa version being changed. You can also see the addition of Unigine Valley which did not actually run on the 10.0 drivers which is a big step up in itself. The other games ran from 46% increases all the way to 462% which I am pretty sure must be a bug in Portal or something because I did not believe my eyes even after running repeated tests :) .
The majority of the source games ran with 2xAA so there is an increase to be seen there when it used to limit the games quite a bit but now it seems its no longer an issue as although not benchmarked here, I can comfortably run each game at a minimum of 40fps now. From these results, its no wonder that AMD wants to drop its catalyst driver for FOSS principles when you see results like this that catalyst just isn't getting on it's own.
As with all Linux projects, 10.3 is already in full swing and will be benchmarked as soon as the arch repository's update with it. For those of you that want the action now need to make sure your kernel is as up to date as your distro can be while still being stable and some distributions have third party providers to newer mesa versions. Arch and its derivatives can download the latest mesa from its repositories along with the latest 3.15 kernel.
In my own opinion we are accelerating at such a speed with these drivers, especially with the news that valve are now supporting its development.
If you would like to see the results, demo files for your own benchmarks, or the system specifications then visit this Google Docs page. Article taken from GamingOnLinux.com.
0 Likes
Some you may have missed, popular articles from the last month:
The comments on this article are closed.
All posts need to follow our rules. For users logged in: please hit the Report Flag icon on any post that breaks the rules or contains illegal / harmful content. Guest readers can email us for any issues.
Patreon
PayPal
Anyway thank you for these graphs :)
Define your "proper drivers". A real argument could be made that the only "proper drivers" on Linux are those that actually rely on the "proper" Linux graphics stack, therefore making all binary blobs painful kludges that weld on a huge chunk of Windows in an effort to try and cut down on "proper" engineering time and in the end results in less "proper" system integration and less "proper" code access. "Proper" can be a difficult adjective.
or did you mean it was released in 2008, and you bought it in 2010?
I never buy graphics cards unless they are used. Retail is too dodgy and expensive in my experience.
I knew which driver he was referring to specifically, I was asking for his definition of what constitutes "proper drivers" as that was the point I was debating. I am not that aloof.
Righty, after a bit of reading the things you will need:
Ubuntu 14.04 as it provides Linux 3.15 which is a must have.
According to this Ubuntu will have mesa 10.2 very soon. For now you will have mesa 10.1.3 but I'm sure within a few weeks you should get these sort of numbers.
Getting mesa should be a simple case of going through this.
I think that loggfreak's point was that the anonymous poster actually said "prop drivers," and, while it is understandable that you would read that as 'proper drivers,' he thinks the poster meant 'proprietary drivers.' Personally, my initial tendency was to read it as 'proper,' but then I thought twice, and it's quite possible that loggfreak's impression is accurate.
You can get reasonable video playback with the Catalyst drivers by using a media player which supports OpenGL output and using that. Otherwise you will tend to get tearing.
Okay, that may be true. Maybe I am being that aloof. :|
Another lesson in the folly of abbreviation.