start-remote-server

What problem does it solve?

Manually provisioning cloud GPU VMs for emmy compiler workflows requires handling provider-specific configurations, capacity checks, SSH setup, and billing management, which is time-consuming and error-prone. This Skill automates the entire process, handling provider selection, capacity fallbacks, and credential validation to deliver a ready-to-use remote GPU server instantly.

Core Features & Use Cases

• Multi-Provider GPU Provisioning: Automatically provisions VMs on GCP or CloudRift with the exact GPU model and count you request, resolving hardware names and handling capacity fallbacks across zones and instance types.
• SSH & Credential Management: Configures SSH keys, validates VM reachability, and returns connection details so you can immediately run emmy deploy or bench commands against the server.
• Use Case: If you need to test a compiled emmy kernel on an NVIDIA H200 GPU, this Skill provisions the VM, gives you the SSH target, and provides the exact teardown command to avoid unexpected billing when you're done.

Quick Start

Use the start-remote-server skill to provision a 4x NVIDIA A100 GPU VM on GCP for running emmy benchmarks next week.