NVIDIA GPU Server Options for Inference and Training | Voters

NVIDIA GPU Server Options for Inference and Training

complete

William Hector

GPU enabled server platform for Workload Optimized configs.

October 8, 2020

Ravinder Braich

marked this post as

complete

Zachary Smith

marked this post as

in progress

Zachary Smith

Merged in a post:

GPU Config for Graphics-Based Workloads

Damian Sieczkowski

GPU enabled server configuration for workloads that are graphics intensive, eg broadcast servers seeking to produce streams at 60FPS at 4K.

May 23, 2022

Zachary Smith

Merged in a post:

Nvidia GPU support

Marlon Ruiz

Ability to create machines with highend GPU for ML/AI applications. Ex: A100

May 23, 2022

Alessandro Baldi Vitali

Hi @Ravinder Braich, thank you for updating this ticket.
Can I ask you if there is any plan to have those servers available on-demand too or if it will always be on contract with at least 1 year reservation?
This is what I hoped for since I opened my Equinix account.
Thank you in advance and have a great day!

Ravinder Braich

Hi Alessandro Baldi Vitali, no plans at this time to move these to on-demand.

Ravinder Braich

marked this post as

complete

See our GPU (NVIDIA A30 & A100) options here as part of our new Workload Optimized configurations:
https://metal.equinix.com/product/servers/a3-large-opt-m3a2/
https://metal.equinix.com/product/servers/a3-large-opt-m3a4/

Zachary Smith

marked this post as

in progress

Zachary Smith

marked this post as

planned / on deck

Rubén perez

hi Hector, we are using different GPUs.
To give some examples.
For our Remote Postproduction project We use GRID for virtualization under VMware. Nvidia M60 GRID or higher. (If I am ambitious M10)
For transcodes (ffmpeg) P2000 / P4000, Nvidia T4
For PlayOut Project. M60
Hope this help in your question.
Regards.

Kiyomichi Arai

Rubén perez: What is the "Remote Postproduction project"?