I have seen a lot of people promoting Ubuntu or Linux in general as being more efficient. There are other benchmarks out that test a variety of things and I will link these.
I wanted to test some things I initially was annoyed by these people claiming how amazing Ubuntu/Linux is. I was surprised actually by my results.
All testing was done on my Lenovo laptop.
Specs and information
Windows 8 64bit
AMD Catalyst 3.11
Sun Java 7_45
Minecraft version 1.7.2
Ubuntu 13.10 64bit
AMD Catalyst 3.11
Linux Kernel 3.11.6
OpenJDK Version 1.7.0_25
Sun Java Version 1.7.0_45
OnDemand CPUFreq governor used
Java versions
I tested first OpenJDK and then Sun Java under Ubuntu both have the same performance.
In order to maintain level playing field I ran Sun Java for the test comparison to Windows.
Minecraft settings
The test was done on a 300 block long minecart track running on a Minecraft server. I don't know anyone who plays single player MC so I felt it would be best to try and match how most people run the game.
The game was running in fullscreen on both operating systems.
If you are an open source purist or for some reason think things are inherently better because they are open source you are going to run the FOSSdrivers.
Windows 8
Firstrun: Avg:34.9 Min:23 Max:53
I have noticed with my laptop all games run better when I force them to run only on 1 or 2 cores. Also if I disable Windows from throttling my CPU.
Using CCC I forced the CPU to run at 2300mhz or higher and then forced Mc to run only on core 0 and 1.
Secondrun: Avg:41.684 Min:26 Max:60
Windows 8.1 with affinity set to all 4 cores and the CPU able to downclock and not being forced to run at full speed.
FirstWin8.1run: Avg:48 Min:36 Max:60
AMD did claim they worked with Microsoft to improve frames on their APUs looks like they did.
Would be interesting if you tested on another Linux distro, such as Debian Wheezy.
But from what I know, Minecraft is a YMMV game all the way.
Interesting results though, but given this is java and its universal capabilities, not all out surprising I would say on that part.
Display drivers maybe another interesting thing given some FPS averages.
Debian Wheezy would result in a much lower FPS due to Debian being based around stability they use older Xorg drivers and Linux Kernel.
I might load up fedora and test it.
Even then I was surprised by some results I might do more considering using BFS rather then the CFS.
Also worth mentioning I forgot to in the op I manually enabled DPM for the FOSS drivers in the Kernel to improve battery life and performance.
Doing more tests is a pain as the only decent way to get FPS measurements in most games and log it is to compile an OpenGL tool to do it.
If there was a less time consuming way to measure performance I would probably be more interested in running tests with some different configurations.
Debian Wheezy would result in a much lower FPS due to Debian being based around stability they use older Xorg drivers and Linux Kernel.
I might load up fedora and test it.
Even then I was surprised by some results I might do more considering using BFS rather then the CFS.
Also worth mentioning I forgot to in the op I manually enabled DPM for the FOSS drivers in the Kernel to improve battery life and performance.
Doing more tests is a pain as the only decent way to get FPS measurements in most games and log it is to compile an OpenGL tool to do it.
If there was a less time consuming way to measure performance I would probably be more interested in running tests with some different configurations.
Been so sleep deprived, not bothered to write a response.
Wheezy earned it's name honest, but I remain loyal given its stability and I know my way around in it.
Fedora would be interesting.
Hum, wonder if anything will change without DPM on FOSS. I doubt anything would really, if marginally slightly maybe.
Sadly such is so, though 3 different compares would be interesting.
It is what debian is for a stable platform, that means sometimes they are slower to get features that might improve performance.
I might do fedora not sure, RHEL and Ubuntu/debian distros are the only ones I have ran before.
DPM enables the GPU to use boost states and downclock so it should be a gain in performance. My laptop runs much hotter without it.
Given Debian's history, it has been good they do this.
I have only used Ubuntu and Debian, CentOS rare times because wanted to learn a bit on it.
If the clocks were not dynamic and at their peak as said by the Vbios, I am wondering if the cards otherwise with power management enabled float their clocks even while game is running.
I'd first like to point out that your benchmarks aren't an accurate representation of Linux as a whole, only of Ubuntu Linux. That is, if they really were accurate benchmarks at all - you performed a single test on a single machine only a few times.
Beyond that, your premise is a bit faulty - I don't know anyone who claims that it's the graphics card DRIVERS that are better in Linux. Open source drivers do some tasks better and others worse depending on the card, and proprietary drivers have their own problems as well. The reason people recommend Linux for playing Minecraft is because it performs better on slower computers than Windows does, since it uses less memory for system processes and UI (obviously depending on distro and WM/DE).
I'll give you a "personal benchmark" of my own. My sister's laptop ran Windows XP and couldn't run Minecraft any faster than 4 FPS with Optifine. I installed Arch Linux with LXDE and the frame rate jumped up to 15 without Optifine, 20 with Optifine. I'd never claim that driver superiority is what helped her computer's performance.
How about this - you try doing a real benchmark with several tests on more than one distro on various types of computers, and then we can evaluate your results with more accuracy. As it stands, the data you've provided is insufficient for discussion of graphics drivers (which, again, are generally not touted as Linux's superior area anyway).
I'd just like to point out these tests suck anyway, and are not at all a good benchmark of performance between systems. There's so many variables that could be in play. Not to mention Minecraft might benefit more from running on a particular distro of Linux or even particular drivers more than other games.
In general I find the whole "Linux is better" argument to be almost impossible to reliably test in any case.
The reason people recommend Linux for playing Minecraft is because it performs better on slower computers than Windows does, since it uses less memory for system processes and UI (obviously depending on distro and WM/DE).
Don't think I've seen it recommended around here in quite a while. Also, you're claiming that the memory usage is why Linux runs faster? That's laughable. You've provided no actual proof that any Linux distro does run faster than Windows on any machine ever. Show us some "legitimate" benchmarks to prove it, then we might discuss it. It would also be highly dependent on which distro we're talking about, and what it's being used for.
Memory usage wouldn't make a difference anyway. Minecraft only needs around half a gig of memory generally, and never touches its max of 1 gig. Any system that has that little memory in the first place wouldn't be able to run Minecraft anyways.
I'd first like to point out that your benchmarks aren't an accurate representation of Linux as a whole, only of Ubuntu Linux. That is, if they really were accurate benchmarks at all - you performed a single test on a single machine only a few times.
Beyond that, your premise is a bit faulty - I don't know anyone who claims that it's the graphics card DRIVERS that are better in Linux. Open source drivers do some tasks better and others worse depending on the card, and proprietary drivers have their own problems as well. The reason people recommend Linux for playing Minecraft is because it performs better on slower computers than Windows does, since it uses less memory for system processes and UI (obviously depending on distro and WM/DE).
I'll give you a "personal benchmark" of my own. My sister's laptop ran Windows XP and couldn't run Minecraft any faster than 4 FPS with Optifine. I installed Arch Linux with LXDE and the frame rate jumped up to 15 without Optifine, 20 with Optifine. I'd never claim that driver superiority is what helped her computer's performance.
How about this - you try doing a real benchmark with several tests on more than one distro on various types of computers, and then we can evaluate your results with more accuracy. As it stands, the data you've provided is insufficient for discussion of graphics drivers (which, again, are generally not touted as Linux's superior area anyway).
The differences between most distros with the same mesa version and same driver version with the same CPU scheduler is minimal.
Not to mention I tested the most popular distro on the planet Ubuntu.
This is not standing on its own take a look at the Phoronix benchmarks, this is just adding to the evidence that exists already.
I could have ran more tests might have tested on fedora, however the issue is there is no decent screen recorders. To get this data I had to compile a tool my self to measure the performance. http://www.phoronix....tu13_el64&num=2
Gaming performance is lacking due to drivers this is why it is important. The OpenGL version that is usable by open source drivers is behind DirectX.
My computer has 4GB of ram the 400MB saved by running the lightest DE and WM is not going to improve Minecrafts performance.
The major usage of RAM on a windows install is superfetch, this is a feature and that memory can be freed up if a program needs it.
Fanboism is bad on any level, the general statement that Linux is worse or better is too broad of a statement. When it comes to games we can test and most Linux distros fall behind.
'My clean updated arch install is so much faster then my 8 year old bloated xp install with out of date drivers'
I wanted to test some things I initially was annoyed by these people claiming how amazing Ubuntu/Linux is. I was surprised actually by my results.
All testing was done on my Lenovo laptop.
Specs and information
Windows 8 64bit
AMD Catalyst 3.11
Sun Java 7_45
Minecraft version 1.7.2
Ubuntu 13.10 64bit
AMD Catalyst 3.11
Linux Kernel 3.11.6
OpenJDK Version 1.7.0_25
Sun Java Version 1.7.0_45
OnDemand CPUFreq governor used
Java versions
I tested first OpenJDK and then Sun Java under Ubuntu both have the same performance.
In order to maintain level playing field I ran Sun Java for the test comparison to Windows.
Minecraft settings
The test was done on a 300 block long minecart track running on a Minecraft server. I don't know anyone who plays single player MC so I felt it would be best to try and match how most people run the game.
The game was running in fullscreen on both operating systems.
The Numbers
Ubuntu 13.10 Numbers
FOSSDrivers Avg:34 Min:29 Max:39
AMD Drivers: Avg:45.39 Min:35 Max:60
If you are an open source purist or for some reason think things are inherently better because they are open source you are going to run the FOSSdrivers.
Windows 8
Firstrun: Avg:34.9 Min:23 Max:53
I have noticed with my laptop all games run better when I force them to run only on 1 or 2 cores. Also if I disable Windows from throttling my CPU.
Using CCC I forced the CPU to run at 2300mhz or higher and then forced Mc to run only on core 0 and 1.
Secondrun: Avg:41.684 Min:26 Max:60
Windows 8.1 with affinity set to all 4 cores and the CPU able to downclock and not being forced to run at full speed.
FirstWin8.1run: Avg:48 Min:36 Max:60
AMD did claim they worked with Microsoft to improve frames on their APUs looks like they did.
If you want a look with a lot of difference applications look here.
http://www.phoronix....indows81_ubuntu
Just an interesting look at OS performance and the performance of my laptop in general with Minecraft.
But from what I know, Minecraft is a YMMV game all the way.
Interesting results though, but given this is java and its universal capabilities, not all out surprising I would say on that part.
Display drivers maybe another interesting thing given some FPS averages.
Debian Wheezy would result in a much lower FPS due to Debian being based around stability they use older Xorg drivers and Linux Kernel.
I might load up fedora and test it.
Even then I was surprised by some results I might do more considering using BFS rather then the CFS.
Also worth mentioning I forgot to in the op I manually enabled DPM for the FOSS drivers in the Kernel to improve battery life and performance.
Doing more tests is a pain as the only decent way to get FPS measurements in most games and log it is to compile an OpenGL tool to do it.
If there was a less time consuming way to measure performance I would probably be more interested in running tests with some different configurations.
Been so sleep deprived, not bothered to write a response.
Wheezy earned it's name honest, but I remain loyal given its stability and I know my way around in it.
Fedora would be interesting.
Hum, wonder if anything will change without DPM on FOSS. I doubt anything would really, if marginally slightly maybe.
Sadly such is so, though 3 different compares would be interesting.
It is what debian is for a stable platform, that means sometimes they are slower to get features that might improve performance.
I might do fedora not sure, RHEL and Ubuntu/debian distros are the only ones I have ran before.
DPM enables the GPU to use boost states and downclock so it should be a gain in performance. My laptop runs much hotter without it.
Given Debian's history, it has been good they do this.
I have only used Ubuntu and Debian, CentOS rare times because wanted to learn a bit on it.
If the clocks were not dynamic and at their peak as said by the Vbios, I am wondering if the cards otherwise with power management enabled float their clocks even while game is running.
Beyond that, your premise is a bit faulty - I don't know anyone who claims that it's the graphics card DRIVERS that are better in Linux. Open source drivers do some tasks better and others worse depending on the card, and proprietary drivers have their own problems as well. The reason people recommend Linux for playing Minecraft is because it performs better on slower computers than Windows does, since it uses less memory for system processes and UI (obviously depending on distro and WM/DE).
I'll give you a "personal benchmark" of my own. My sister's laptop ran Windows XP and couldn't run Minecraft any faster than 4 FPS with Optifine. I installed Arch Linux with LXDE and the frame rate jumped up to 15 without Optifine, 20 with Optifine. I'd never claim that driver superiority is what helped her computer's performance.
How about this - you try doing a real benchmark with several tests on more than one distro on various types of computers, and then we can evaluate your results with more accuracy. As it stands, the data you've provided is insufficient for discussion of graphics drivers (which, again, are generally not touted as Linux's superior area anyway).
Sorry, I didn't realize that I was offending you. If you didn't expect to find Linux fans, maybe you should be browsing another thread?
In general I find the whole "Linux is better" argument to be almost impossible to reliably test in any case.
Don't think I've seen it recommended around here in quite a while. Also, you're claiming that the memory usage is why Linux runs faster? That's laughable. You've provided no actual proof that any Linux distro does run faster than Windows on any machine ever. Show us some "legitimate" benchmarks to prove it, then we might discuss it. It would also be highly dependent on which distro we're talking about, and what it's being used for.
Memory usage wouldn't make a difference anyway. Minecraft only needs around half a gig of memory generally, and never touches its max of 1 gig. Any system that has that little memory in the first place wouldn't be able to run Minecraft anyways.
"Programmers never repeat themselves. They loop."
The differences between most distros with the same mesa version and same driver version with the same CPU scheduler is minimal.
Not to mention I tested the most popular distro on the planet Ubuntu.
This is not standing on its own take a look at the Phoronix benchmarks, this is just adding to the evidence that exists already.
I could have ran more tests might have tested on fedora, however the issue is there is no decent screen recorders. To get this data I had to compile a tool my self to measure the performance.
http://www.phoronix....tu13_el64&num=2
Gaming performance is lacking due to drivers this is why it is important. The OpenGL version that is usable by open source drivers is behind DirectX.
My computer has 4GB of ram the 400MB saved by running the lightest DE and WM is not going to improve Minecrafts performance.
The major usage of RAM on a windows install is superfetch, this is a feature and that memory can be freed up if a program needs it.
Fanboism is bad on any level, the general statement that Linux is worse or better is too broad of a statement. When it comes to games we can test and most Linux distros fall behind.
'My clean updated arch install is so much faster then my 8 year old bloated xp install with out of date drivers'