Skip to content
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

Jump to bottom

nixos/nvidia: move TOPOLOGY_FILE_PATH and DATABASE_PATH into nvidia-fabricmanager service definition #320830

Merged
merged 1 commit into from
Jun 27, 2024
+8 −5
Merged

nixos/nvidia: move TOPOLOGY_FILE_PATH and DATABASE_PATH into nvidia-fabricmanager service definition #320830

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
13 changes: 8 additions & 5 deletions nixos/modules/hardware/video/nvidia.nix
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -46,8 +46,6 @@ in
TRUNK_LINK_FAILURE_MODE = 0;
NVSWITCH_FAILURE_MODE = 0;
ABORT_CUDA_JOBS_ON_FM_EXIT = 1;
TOPOLOGY_FILE_PATH = "${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
DATABASE_PATH = "${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
Comment on lines -49 to -50
Copy link
Contributor

Choose a reason for hiding this comment

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

Side-note: I find these passthru packages in nvidia_x11 rather annoying; we recently found that e.g. libXNctrl could be moved out of nvidia_x11 entirely (#318092); do you know if is fabricmanager driver-locked?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

https://docs.nvidia.com/datacenter/tesla/pdf/fabric-manager-user-guide.pdf

That diagram suggests that it is, sadly, but I don't know for sure.

};
defaultText = lib.literalExpression ''
{
Expand All @@ -69,8 +67,6 @@ in
TRUNK_LINK_FAILURE_MODE=0;
NVSWITCH_FAILURE_MODE=0;
ABORT_CUDA_JOBS_ON_FM_EXIT=1;
TOPOLOGY_FILE_PATH="''${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
DATABASE_PATH="''${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
}
'';
description = ''
Expand Down Expand Up @@ -628,7 +624,14 @@ in
TimeoutStartSec = 240;
ExecStart =
let
nv-fab-conf = settingsFormat.generate "fabricmanager.conf" cfg.datacenter.settings;
# Since these rely on the `nvidia_x11.fabricmanager` derivation, they're
# unsuitable to be mentioned in the configuration defaults, but they _can_
# be overridden in `cfg.datacenter.settings` if needed.
fabricManagerConfDefaults = {
TOPOLOGY_FILE_PATH = "${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
Copy link
Contributor

Choose a reason for hiding this comment

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

Sanity-check: here nvidia_x11 != null because it's guarded by cfg.enable?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Correct.

DATABASE_PATH = "${nvidia_x11.fabricmanager}/share/nvidia-fabricmanager/nvidia/nvswitch";
};
nv-fab-conf = settingsFormat.generate "fabricmanager.conf" (fabricManagerConfDefaults // cfg.datacenter.settings);
in
"${lib.getExe nvidia_x11.fabricmanager} -c ${nv-fab-conf}";
LimitCORE = "infinity";
Expand Down