Skip to content

driver: request gpu when creating container builder #3063

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 1 commit into from
Mar 13, 2025
Merged

driver: request gpu when creating container builder #3063

Show file tree
Hide file tree
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
29 changes: 29 additions & 0 deletions driver/docker-container/driver.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -56,6 +56,7 @@ type Driver struct {
restartPolicy container.RestartPolicy
env []string
defaultLoad bool
gpus []container.DeviceRequest
}

func (d *Driver) IsMobyDriver() bool {
Expand Down Expand Up @@ -158,6 +159,9 @@ func (d *Driver) create(ctx context.Context, l progress.SubLogger) error {
if d.cpusetMems != "" {
hc.Resources.CpusetMems = d.cpusetMems
}
if len(d.gpus) > 0 && d.hasGPUCapability(ctx, cfg.Image, d.gpus) {
hc.Resources.DeviceRequests = d.gpus
}
if info, err := d.DockerAPI.Info(ctx); err == nil {
if info.CgroupDriver == "cgroupfs" {
// Place all buildkit containers inside this cgroup by default so limits can be attached
Expand Down Expand Up @@ -429,6 +433,31 @@ func (d *Driver) HostGatewayIP(ctx context.Context) (net.IP, error) {
return nil, errors.New("host-gateway is not supported by the docker-container driver")
}

// hasGPUCapability checks if docker daemon has GPU capability. We need to run
// a dummy container with GPU device to check if the daemon has this capability
// because there is no API to check it yet.
func (d *Driver) hasGPUCapability(ctx context.Context, image string, gpus []container.DeviceRequest) bool {
Copy link
Member Author

Choose a reason for hiding this comment

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

Time this func takes:

  • ~15ms on MacOS (no GPU)
  • ~230ms on WSL (with GPU)
  • ~200ms on WSL (no GPU)
  • ~200ms on Windows (with GPU)
  • ~160ms on Windows (no GPU)

cfg := &container.Config{
Image: image,
Entrypoint: []string{"/bin/true"},
}
hc := &container.HostConfig{
NetworkMode: container.NetworkMode(container.IPCModeNone),
AutoRemove: true,
Resources: container.Resources{
DeviceRequests: gpus,
},
}
resp, err := d.DockerAPI.ContainerCreate(ctx, cfg, hc, &network.NetworkingConfig{}, nil, "")
if err != nil {
return false
}
if err := d.DockerAPI.ContainerStart(ctx, resp.ID, container.StartOptions{}); err != nil {
return false
}
return true
}

func demuxConn(c net.Conn) net.Conn {
pr, pw := io.Pipe()
// TODO: rewrite parser with Reader() to avoid goroutine switch
Expand Down
6 changes: 6 additions & 0 deletions driver/docker-container/factory.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,6 +51,12 @@ func (f *factory) New(ctx context.Context, cfg driver.InitConfig) (driver.Driver
InitConfig: cfg,
restartPolicy: rp,
}
var gpus dockeropts.GpuOpts
if err := gpus.Set("all"); err == nil {
if v := gpus.Value(); len(v) > 0 {
d.gpus = v
}
}
for k, v := range cfg.DriverOpts {
switch {
case k == "network":
Expand Down