can a tesla k80 be used to help processing and ram server side

[Daized Joker]

Is there a way to use the cuda kit to just offload the processing and ram into the gpu on the server side. I'm hoping to reduce the amount of ram and processor usage on my server if this is at all possible. Running windows server 2019 on a dell poweredge r820 with 4 E5-4650 8core @2.7ghz, 512gb ddr3 ram, 4 500gb SSD in raid 10, 2x 2tb ssd in raid 0, and tesla k80 gpu

 

[schizrade]

Process what, cause it all depends on the what.

 

[Daized Joker]

Process what, cause it all depends on the what.

I'm running game servers some are ran in VM and some are ran right on the windows server 2019 os

 

[schizrade]

I'm running game servers some are ran in VM and some are ran right on the windows server 2019 os

Then no. Its a HPC GPGPU unit. Its designed for massive parallel number crunching, not gaming or general server ops offloading.

 

[Daized Joker]

Then no. Its a HPC GPGPU unit. Its designed for massive parallel number crunching, not gaming or general server ops offloading.

Yep I'm aware of that I just wasn't sure if I could implement a way for the server to offload some of the processing. These are ran headless so all it is processing no graphic output.

 

[Kardonxt]

No. You can use it to boost performance of programs that support GPU acceleration.(Such as these https://www.nvidia.com/en-us/gpu-accelerated-applications/ .) I'm not aware of any game servers that support that as system requirements are typically pretty low.

 

[Daized Joker]

No. You can use it to boost performance of programs that support GPU acceleration.(Such as these https://www.nvidia.com/en-us/gpu-accelerated-applications/ .) I'm not aware of any game servers that support that as system requirements are typically pretty low.

Thanks I'm looking into it actually seems to take over what pe hosting the steam version Servers of the game because they can be ran directly on the server. The other version needs to be hosted in a VM.