Files
Olares/cli/pkg/amdgpu/prepares.go
eball cf7125aac8 cli, daemon: enhance DGX Spark support and update GPU type handling (#2496)
* feat(gpu): enhance DGX Spark support and update GPU type handling

* feat(amdgpu): refactor AMD GPU detection and support for GB10 chip and APU

* feat(connector): enhance GB10 chip detection with environment variable support

* feat(gpu): enhance DGX Spark support and update GPU type handling

* feat(amdgpu): refactor AMD GPU detection and support for GB10 chip and APU

* feat(connector): enhance GB10 chip detection with environment variable support

* feat: add nvidia device plugin for gb10

* fix(gpu): update pod selector for hami-device-plugin based on GB10 chip detection

fix(deploy): bump app-service image version to 0.4.78

* feat: enable CGO for building on ARM architecture and adjust build constraints for Linux

* feat: enhance multi-architecture support for ARM64 in release workflow

* feat: update multi-arch setup for ARM64 in release workflow

* feat: enhance ARM64 multi-architecture support in release workflow

* feat: streamline ARM64 cross-compilation setup in release workflow

* feat: enhance ARM64 support by adding architecture-specific package installations

* feat: update ARM64 package sources in release workflow for improved compatibility

* feat: amd device plugin and container toolkit install

* refactor: remove GB10 chip type check from GPU info update

* feat(gpu): update hami version to v2.6.10-compatible for spark

* fix: remove gb10 device plugin checking

* fix: update klauspost/cpuid to v2.3.0

* fix: amd gpu check (#2522)

* feat: enhance storage device detection with USB serial properties

* feat: update hami version to v2.6.11-compatible-arm

* feat: add chip type support for AMD and NVIDIA GPUs in node label updates

* feat(gpu): supports auto binding GPU to app

* feat(gpu): remove chip type handling from GPU label updates

* feat(gpu): remove GPU type specification from DaemonSet and values.yaml

* feat(gpu): remove GB10 device plugin installation and related checks

* feat(gpu): update HAMi to v2.6.11

---------

Co-authored-by: dkeven <dkvvven@gmail.com>
Co-authored-by: hys <hysyeah@gmail.com>
2026-02-28 11:44:02 +08:00

57 lines
1.4 KiB
Go

package amdgpu
import (
"github.com/beclab/Olares/cli/pkg/bootstrap/precheck"
"github.com/beclab/Olares/cli/pkg/common"
"github.com/beclab/Olares/cli/pkg/core/connector"
"github.com/beclab/Olares/cli/pkg/core/logger"
)
// RocmInstalled checks if AMD ROCm is installed on the system.
type RocmInstalled struct {
common.KubePrepare
}
func (p *RocmInstalled) PreCheck(runtime connector.Runtime) (bool, error) {
rocmV, err := connector.RocmVersion()
if err != nil {
logger.Debugf("ROCm version check error: %v", err)
return false, nil
}
if rocmV == nil {
return false, nil
}
logger.Infof("Detected ROCm version: %s", rocmV.Original())
return true, nil
}
// RocmNotInstalled checks if AMD ROCm is NOT installed on the system.
type RocmNotInstalled struct {
common.KubePrepare
RocmInstalled
}
func (p *RocmNotInstalled) PreCheck(runtime connector.Runtime) (bool, error) {
installed, err := p.RocmInstalled.PreCheck(runtime)
if err != nil {
return false, err
}
return !installed, nil
}
// ContainerdInstalled checks if containerd is installed on the system.
type ContainerdInstalled struct {
common.KubePrepare
}
func (p *ContainerdInstalled) PreCheck(runtime connector.Runtime) (bool, error) {
containerdCheck := precheck.ConflictingContainerdCheck{}
if err := containerdCheck.Check(runtime); err != nil {
return true, nil
}
logger.Info("containerd is not installed, ignore task")
return false, nil
}