There is more to a performance comparison than one system or configuration or user. I know for a fact that there are hardware combinations that will show Windows always faster than Linux.
Even just back in 2.79 in some 'ad-hoc' online testing demonstrated Blender was faster in Windows than Linux.
The only reason I even decided to comment, is that users should do their own testing differences, and beware of going all-in and committing to Linux based on one YouTube test. As CG Geek himself suggests, do a dual or second installation of Linux and test the difference on your own system.
From code I have been looking through, with 2.8 there were a lot of huge changes that just have not yet gotten much attention or time for Windows optimizations. This can be fairly significant in cross platform development, especially when its focus in on Linux and originates on Linux.
One simple example of this, is code that is doing things at the Blender/Application level because Linux's model relies on applications to do this work. However, when this code is used on Windows, it is redundant, as Windows is already doing the work, and with Blender doing the work again, it is going to run 2x slower on average.
As code is optimized for Windows properly, this type of code would be removed from the Windows version, as Blender can just ask Windows to perform the task, which will also be faster as the method used by Windows is specifically designed to be highly efficient on Windows and often works significantly different than the code in Blender which is less efficient, in addition to be redundant.
I am generalizing the concepts at play here, so that hopefully they will be easily understandable to everyone and not just programmers or engineers.
3 Comments
There is more to a performance comparison than one system or configuration or user. I know for a fact that there are hardware combinations that will show Windows always faster than Linux.
Even just back in 2.79 in some 'ad-hoc' online testing demonstrated Blender was faster in Windows than Linux.
The only reason I even decided to comment, is that users should do their own testing differences, and beware of going all-in and committing to Linux based on one YouTube test. As CG Geek himself suggests, do a dual or second installation of Linux and test the difference on your own system.
From code I have been looking through, with 2.8 there were a lot of huge changes that just have not yet gotten much attention or time for Windows optimizations. This can be fairly significant in cross platform development, especially when its focus in on Linux and originates on Linux.
One simple example of this, is code that is doing things at the Blender/Application level because Linux's model relies on applications to do this work. However, when this code is used on Windows, it is redundant, as Windows is already doing the work, and with Blender doing the work again, it is going to run 2x slower on average.
As code is optimized for Windows properly, this type of code would be removed from the Windows version, as Blender can just ask Windows to perform the task, which will also be faster as the method used by Windows is specifically designed to be highly efficient on Windows and often works significantly different than the code in Blender which is less efficient, in addition to be redundant.
I am generalizing the concepts at play here, so that hopefully they will be easily understandable to everyone and not just programmers or engineers.
* @Anthony * Thanks for the reply! ☺
got my Ubuntu USB installation boot ready :D