NVIDIA GeForce GTX 1060 Offers Great Performance On Linux

[SaucyJack]

<----Trollbar complete.

[zboson]

Thank you for the OpenCL, CUDA, and Luxmark tests Michael !!!

I will buy a new laptop this year and I would not be surprised if Nvidia drops the mobile GPUs and laptops come out with the full GPUs. A GTX 1060 laptop would perform better than my GTX 980M laptop and the laptop would probably weight a lot less. I wonder if the GTX 1060 will be VR certified. I don't see why not because it's better than the GTX 970 which is the baseline for VR. Though neither the CV1 nor the Vive headsets support Linux currently :-(

[Passso]

I really hope that someone was indeed being paid to do so

I earned a 960 GTX for that. Please do not judge me.

[bug77]

Guru3D has nice selection of DX12 games benchmarked. But strangely enough they did not test Doom on Vulkan, you'll have to go somewhere else for that.

http://www.guru3d.com/articles_pages..._review,1.html

Edit: Here's the mega list. Some of them barely touch DX12/Vulkan... guess we know who got a check from Nvidia.
http://videocardz.com/62364/nvidia-g...x-1060-reviews

Techspot has a Doom Vulkan benchmark.
http://www.techspot.com/review/1209-...060/page3.html

Techspot isn't using the required driver for Vulkan. Doom needs 368.69 (see: https://community.bethesda.net/thread/54585?tstart=0), Techspot (like HardOCP) is using 368.64. Afaik, that prevents the use of async compute.
That doesn't invalidate the benchmarks, it just means there's a bit more juice to be squeezed out of Nvidia.

[Passso]

It's clear who the winner is with the budget GPUs, and it is very easy to prove. The 480 has a 256 bit memory bus which gives it ~256GB/s memory bandwidth. The 1060 has 192 and @ ~ 190GB/s. For a good gpu, you need fast memory - drivers can't make up for this...
...
<blabla complicated stuff>]

It's clear who the winner is with the budget GPUs, and it is very easy to prove : the more watts, the more power!

[BauerPatrik]

I really think AMDs proprietary driver should be in the comparison. Especially because just picking the open stack means ignoring AMD in the OpenCL benchmarks. I know, most AMD users prefer the open driver stack. But this is a hardware benchmark and it should use the fastest available driver. (I wonder what happened if Nouveau was working with this cards but was slow like rendering in software.)

[bug77]

I know, most AMD users prefer the open driver stack.

They also prefer Vulkan, so let me see them reconcile both of them

[Michael]

I really think AMDs proprietary driver should be in the comparison. Especially because just picking the open stack means ignoring AMD in the OpenCL benchmarks. I know, most AMD users prefer the open driver stack. But this is a hardware benchmark and it should use the fastest available driver. (I wonder what happened if Nouveau was working with this cards but was slow like rendering in software.)

RadeonSI was used since that's what most people prefer... In past NV reviews when I tested with the blob, I get complaints that I should have used the open-source driver... Given that in most benchmarks the RadeonSI vs. AMDGPU-PRO performance is close, I went with the open-source driver to get the least amount of complaints/bitching.

[bug77]

RadeonSI was used since that's what most people prefer... In past NV reviews when I tested with the blob, I get complaints that I should have used the open-source driver... Given that in most benchmarks the RadeonSI vs. AMDGPU-PRO performance is close, I went with the open-source driver to get the least amount of complaints/bitching.

I think people wanted to see AMDGPU-PRO in addition to the open source stack.

[bokal]

I think people wanted to see AMDGPU-PRO in addition to the open source stack.

And nouveau too. Michael doesn't really need to sleep.