---
title: "Coder workspace template"
description: "Where the VIRDX Coder workspace template lives and how GPU selection works."
image: "https://docs.virdx.dev/img/virdx-social-card.png"
---

> Documentation Index
> Fetch the complete documentation index at: https://docs.virdx.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Coder workspace template

## Template location

Source template: `infra_k8s/current/coder-support/templates/kubernetes-dev/main.tf`

## GPU parameter

The template exposes `gpu_enabled` as a `coder_parameter` bool (`Attach GPU slice`).

- **Default:** `"true"` — new workspaces start with a GPU
- **Type:** `bool`, mutable
- **Description:** "Request one MIG 1g.24gb GPU slice for this workspace."

When `gpu_enabled` is `true`:

- `runtime_class_name = "nvidia"`
- Resource request/limit: `nvidia.com/mig-1g.24gb = "1"`

When `gpu_enabled` is `false`:

- Both the nvidia runtime class and the MIG resource request/limit are omitted

## Critical deployment gotcha

`infra_k8s/current/argo-cd/coder-support.yaml` excludes `templates/**`:

```yaml
directory:
  exclude: "templates/**"
```

**This means ArgoCD does NOT apply template changes.**

After editing `main.tf`, update the template in Coder manually (UI or CLI) so the new parameter reaches users.

Source: https://docs.virdx.dev/knowledge/wiki/workstreams/infrastructure/sops/coder-workspaces/index.mdx