Add `saas-linux-large-amd64-gpu-standard` runner for AI/ML workloads

Release notes

We are aiming to bring the best practices of DevSecOps to data sciences by providing more powerful compute within GitLab runner. Following our first GPU-enabled runners, users have asked for more powerful machines for compute-intensive tasks. Now, we provide a large runner on Linux with 2 standard GPUs enabled.

Overview

Add a new compute machine type, a GCP n1-standard-8 with 2 NVIDIA Tesla T4 GPU to the GitLab SaaS Runners on Linux with tag saas-linux-large-amd64-gpu-standard.

Proposal

  • Create a new shared-runner-manager for GitLab SaaS to autoscale and provision the new machine type. Tag = saas-linux-large-amd64-gpu-standard
  • Configure the runner manager to use a GCP n1-standard-4 (4 vCPUs, 30GB RAM VM) + 1 NVIDIA TESLA T4 GPU configured virtual machine.
  • The Runner that is configured to run workloads on this new machine type will not run un-tagged jobs.

Documentation update

Runner Tag vCPUs Memory Storage GPU GPU Memory
saas-linux-large-amd64-gpu-standard 8 30 GB 100 GB 2 Nvidia Tesla T4 (or similar) 32 GB

Available Tiers

  • Premium
  • Ultimate

Implementation tasks

Reference

This page may contain information related to upcoming products, features and functionality. It is important to note that the information presented is for informational purposes only, so please do not rely on the information for purchasing or planning purposes. Just like with all projects, the items mentioned on the page are subject to change or delay, and the development, release, and timing of any products, features, or functionality remain at the sole discretion of GitLab Inc.

This page may contain information related to upcoming products, features and functionality. It is important to note that the information presented is for informational purposes only, so please do not rely on the information for purchasing or planning purposes. Just like with all projects, the items mentioned on the page are subject to change or delay, and the development, release, and timing of any products, features, or functionality remain at the sole discretion of GitLab Inc.

Edited by Gabriel Engel