NVIDIA GeForce: Windows 10 vs. Ubuntu 15.04 Linux OpenGL Benchmarks

[philips]

The Linux results are too low, on my (inferior system) I get actually better results for Linux:
(i7-4790K Nvidia GTX 780)

[johnc]

Well why is there such a shortfall when the resolution increases or advanced effects are turned on. Kernel Developers need to work out why they are losing points on large block transfers vs Windows.

But that was only on this Xonotic benchmark (which looks flawed on both OSes) and SoM. We don't see that everywhere.

[e8hffff]

I don't think we've seen that at all.

Well why is there such a shortfall when the resolution increases or advanced effects are turned on. Kernel Developers need to work out why they are losing points on large block transfers vs Windows.
484

[johnc]

It's clear from various test of history when comparing Linux, is that Microsoft has incorporated kernel profiling to better handle large blocks of data suited for high-res gaming. As soon as the resolution is high or effects are demanding, Linux falls over itself to compete with Windows.

I don't think we've seen that at all.

[e8hffff]

124
It's clear from various test of history when comparing Linux, is that Microsoft has incorporated kernel profiling to better handle large blocks of data suited for high-res gaming. As soon as the resolution is high or effects are demanding, Linux falls over itself to compete with Windows.

[gamerk2]

The benchmarks we typically see here though seem to indicate that games that are CPU-limited show a performance advantage on Linux. And games that are GPU-limited are about identical. SoM is almost out on its own it seems. (At least in terms of games benched here.) If it was a scheduling issue, wouldn't it show in other games as well?

Not necessarily. SoM in particular uses a lot of render worker threads, where other games (and I'll throw Borderlands 2 out there as an example) typically doesn't. So on a quad core system, even if the render thread got bumped, it would almost certainly get rescheduled in short order. But if you had 8 worker threads to compete with...

Just saying, would be an interesting experiment to see how SoM, which is very thread heavy, performs on various schedulers.

[blackout23]

And beneath that: "The Time Has Come: Powerful Profiling and Debugging Tools Arrive for Linux and OpenGL With NVIDIA’s Linux Graphics Debugger"

Hopefully they can profile SoM. lol.

I am surprised they didn't just join Valve with vogl. Might be because they at least temporarily moved on to develop the Vulkan debugger with LunarG. I hope NVDIAS OpenGL Profiler and Debugger for Linux can also help with other vendors.

[Michael]

WTF Michael? You're so fast the use EON games when benching AMD...

Huh? These are the tests that I have working via PTS on Win/Linux. The full collection of PTS tests aren't currently setup/verified/ported to Windows test profiles.

In no Windows vs. Linux AMD articles do I have EON game tests...

[nightmarex]

WTF Michael? You're so fast the use EON games when benching AMD...

[johnc]

EDIT: Even better there is an NVIDIA Talk "Vulkan on NVIDIA GPUs" on August 9th.
https://www.khronos.org/news/events/2015-siggraph

And beneath that: "The Time Has Come: Powerful Profiling and Debugging Tools Arrive for Linux and OpenGL With NVIDIA?s Linux Graphics Debugger"

Hopefully they can profile SoM. lol.