Originally posted by Tgui
View Post
Announcement
Collapse
No announcement yet.
AMD A10-7850K Kaveri: Windows 8.1 vs. Ubuntu Linux
Collapse
X
-
Originally posted by gutigen View PostOr maybe stop benchmarking irrelevant for Linux gaming apps and games, but focus on things that matter - games I've mentioned earlier. For maybe 20 irrelevant tests we get 1 that makes sense, kinda stupid if you ask me.
Windows tests always have most popular and most demanding games in them and maybe one artificial benchmark (like 3D Mark). It's like Linux has only open source, Opengl 2.0 (at best) Quake clones and a few benchmarks from a company no one cares about and which are using game engine no one else is using.
Not that I don't appreciate Michael's work, but all those tests do not represent real state of GPUs and drivers under Linux.
There is also huge difference between putting results at main page with and digging through Phoronix depths to find something usefull.
Comment
-
Originally posted by kokoko3k View PostI think the unigine tests are NOT irrelevant at all;
they *DO* "represent real state of GPUs and drivers under Linux" and Windows.
Thanks Michael for doing them.
I wouldn't be suprised if AMD was using Phoronix Test Suite or just Unigine benchmarks to fast test their GPUs and move on to more important stuff like Windows and consoles. It would explain why Unigine benchmarks are the only modern apps which work properly under Linux.Last edited by gutigen; 14 January 2014, 04:12 PM.
Comment
-
Originally posted by gutigen View PostNo they don't, there is that thing called optimization, you know? Often drivers have to be optimizied for specific app to work properly. Testing an app which only Phoronix is using or a Quake Engine Opengl 2.0, 10 years old game with 50 people player base and one basement based developer doesn't represent real state of drivers and GPUs.
I wouldn't be suprised if AMD was using Phoronix Test Suite or just Unigine benchmarks to fast test their GPUs and move on to more important stuff like Windows and consoles. It would explain why Unigine benchmarks are the only modern apps which work properly under Linux.
Trolling...
Comment
-
Originally posted by gutigen View PostOr maybe stop benchmarking irrelevant for Linux gaming apps and games, but focus on things that matter - games I've mentioned earlier. For maybe 20 irrelevant tests we get 1 that makes sense, kinda stupid if you ask me.
Windows tests always have most popular and most demanding games in them and maybe one artificial benchmark (like 3D Mark). It's like Linux has only open source, Opengl 2.0 (at best) Quake clones and a few benchmarks from a company no one cares about and which are using game engine no one else is using.
Not that I don't appreciate Michael's work, but all those tests do not represent real state of GPUs and drivers under Linux.
There is also huge difference between putting results at main page with and digging through Phoronix depths to find something usefull.
But seriously Michael, you *need* to post that link in big bold letters on the first page of every GPU test article you post.
Anyway more on topic of the article, I'm very impressed to see catalyst is so caught up with Windows. I don't ever remember ATI/AMD products on linux having that minimal of a performance gap against windows. I think I'm still going to stick with the open source drivers though.
Comment
-
Originally posted by schmidtbag View PostThe tests are supposed to show progression of how hardware behaves on linux.
It behaves like shit when it comes to games most people actually play.
In other words what you're saying here makes as much sense as those tests, read - none.
Comment
-
Given the age of Linux, the age of catalyst/FOSS and the growing age of decent games that use catalyst,mesa,steam,SteamOS,latest kernal im fairly mind blown someone doesn’t on the entire planet create a decent easy to follow benchmark site for games on Linux.
Forget that there often slower due to lack of individual patching, the way the internet works with regards to GPU's and driver is that gamers bang their chests when the results are slower and then you get a response from the company.. So again forget a totally accurate like for like age old benchmark lets look at what joe blogs is experiencing .
It would go like this
Install Ubuntu , Install FGLRX, get things upto date.. post the versions
Install Steam
Install some free to play games and run them at 1080p with and without AA and Vsync
Do some frame times , take averages, talk about vsync and mouse lag if there is any or loading times etc..
Install win8, install FGLRX get things upto date ... etc do the same
Free to play or extremely cheap games on steam are abundant even on Linux.
TF2
CS:source
DOTA2
Killing Floor
L4D2
Portal
Trine 2
Make the graph simple side by side with one saying Win8 Min/Av/Max FPS = 50/6070 ......... Ubuntu Min/Av/Max FPS = 5/25/40 ......... SteamOS Max FPS = 6/27/38
Yes, it wont be a tried and tested like for like in absolute terms we all get it.. but it will put a spot light on what matters..
Real world performance from modern real world games
Comment
-
Originally posted by gutigen View PostIt behaves great when it comes to irrelevant software.
It behaves like shit when it comes to games most people actually play.
In other words what you're saying here makes as much sense as those tests, read - none.
Comment
Comment