Skip to content

[Runpod] Require CUDA 12.8+ on the host #3304

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Nov 20, 2025
Merged

[Runpod] Require CUDA 12.8+ on the host #3304

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4f9cfee
[Runpod] Require CUDA 12.8+ on the host #3299
peterschmidt85 Nov 19, 2025
a893390
[Runpod] Require CUDA 12.8+ on the host #3299
peterschmidt85 Nov 20, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion pyproject.toml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -32,7 +32,7 @@ dependencies = [
"python-multipart>=0.0.16",
"filelock",
"psutil",
"gpuhunt==0.1.12",
"gpuhunt==0.1.13",
"argcomplete>=3.5.0",
"ignore-python>=0.2.0",
"orjson",
Expand Down
5 changes: 5 additions & 0 deletions src/dstack/_internal/core/backends/runpod/api_client.py
View file
Open in desktop
Copy link
Collaborator

@jvstme jvstme Nov 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should we also pass the CUDA version when creating clusters? See _generate_create_cluster_mutation

Copy link
Collaborator

@jvstme jvstme Nov 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

If necessary, we can add it later in a separate PR, of course

Copy link
Contributor Author

@peterschmidt85 peterschmidt85 Nov 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I pass it here:

dstack/src/dstack/_internal/core/backends/runpod/api_client.py

Line 479 in 4f9cfee

input_fields.append(f'minCudaVersion: "{RunpodProvider.MIN_CUDA_VERSION}"')

Please confirm

Copy link
Collaborator

@jvstme jvstme Nov 20, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This line is within _generate_pod_deployment_mutation, which is only used for individual pods. My comment is about clusters.

Anyways, you merged the gpuhunt PR without adding a filter for cluster offers, so it's better not to add it in dstack now

Original file line number Diff line number Diff line change
Expand Up @@ -3,6 +3,7 @@
from typing import Any, Dict, List, Optional

import requests
from gpuhunt.providers.runpod import RunpodProvider
from requests import Response

from dstack._internal.core.errors import BackendError, BackendInvalidCredentialsError
Expand Down Expand Up @@ -475,6 +476,8 @@ def _generate_pod_deployment_mutation(
)
input_fields.append(f"allowedCudaVersions: [{allowed_cuda_versions_string}]")

input_fields.append(f'minCudaVersion: "{RunpodProvider.MIN_CUDA_VERSION}"')

pod_deploy = "podFindAndDeployOnDemand" if bid_per_gpu is None else "podRentInterruptable"
# Format input fields
input_string = ", ".join(input_fields)
Expand Down Expand Up @@ -597,6 +600,8 @@ def _generate_create_cluster_mutation(
ports = ports.replace(" ", "")
input_fields.append(f'ports: "{ports}"')

input_fields.append(f'minCudaVersion: "{RunpodProvider.MIN_CUDA_VERSION}"')

# Format input fields
input_string = ", ".join(input_fields)
return f"""
Expand Down