native cuda support #75
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: Alexandru Costache <[email protected]>
Signed-off-by: Alexandru Costache <[email protected]>
Signed-off-by: Alexandru Costache <[email protected]>
For some reason that needs a bit more digging, the meta-tegra cuda recipe won't get the cuda-repo deb from the devnet. Let's add an append for now, while testing, since distributing cuda libraries already needs special permissions. Signed-off-by: Alexandru Costache <[email protected]>
We used to mask all graphics recipes, since graphics packages go to containers. But now we need some libs and their dependencies in the hostOS, so let's unmask them. Signed-off-by: Alexandru Costache <[email protected]>
Signed-off-by: Alexandru Costache <[email protected]>
Base recipe runtime depends render this package unbuildable. Drop dependencies as we don't need docker-ce, cuda-toolkit or others. Signed-off-by: Alexandru Costache <[email protected]>
includes the 19.03.13 version of balena-engine Signed-off-by: Robert Günzler <[email protected]>
|
Can one of the admins verify this patch? |
|
@resin-jenkins test this please |
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 5, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 5, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 5, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 8, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 9, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 9, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 10, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 11, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
|
@acostach can we run your test device with my supervisor patch applied? would be good to verify that change suffices before it's merged |
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 11, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 11, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
robertgzr
added a commit
to balena-os/balena-supervisor
that referenced
this pull request
Jun 11, 2020
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
|
@robertgzr any updates regarding this PR? |
|
All the changes in this PR have been included in #85 For native cuda support to work there's still necessary for the hostapp extras feature to be implemented in meta-balena. That will allow overlaying the tegra libraries from a container into the rootfs. Those libraries are necessary for the nvidia hook to run. |
pipex
pushed a commit
to balena-io-modules/balena-compose-experiment
that referenced
this pull request
Mar 29, 2021
In the absence of an upstream implementation of the DeviceRequest API introduced as part of Docker API v1.40 we roll our own using a feature label. As per my comment in the code, we fall back to the default behavior of docker cli's `--gpu` and request single device with the `gpu` capabilty. The only implementation at the moment is the NVIDIA driver; here: https://github.com/balena-os/balena-engine/blob/master/daemon/nvidia_linux.go Background on the composefile implementation: compose-spec/compose-spec#74 docker/compose#6691 Change-type: patch Connects-to: balena-os/balena-jetson#75 Signed-off-by: Robert Günzler <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
With balena-engine 19.03.13 merged in meta-balena it's possible to give containers access to the graphics hardware via NVIDIA's container toolkit.
We need:
Connects-to: #57