peermem

Documentation

Overview

Package peermem contains the implementation of the peermem query for NVIDIA GPUs.

Index

• Constants
• func HasLsmodInfinibandPeerMem(lsmodOutput string) bool
• type LsmodPeermemModuleOutput
  • func CheckLsmodPeermemModule(ctx context.Context) (*LsmodPeermemModuleOutput, error)

Constants

const RegexInvalidContext = `.*ERROR detected invalid context, skipping further processing`

repeated messages may indicate more persistent issue on the inter-GPU communication e.g., [Thu Sep 19 02:29:46 2024] nvidia-peermem nv_get_p2p_free_callback:127 ERROR detected invalid context, skipping further processing [Thu Sep 19 02:29:46 2024] nvidia-peermem nv_get_p2p_free_callback:127 ERROR detected invalid context, skipping further processing [Thu Sep 19 02:29:46 2024] nvidia-peermem nv_get_p2p_free_callback:127 ERROR detected invalid context, skipping further processing

Functions

func HasLsmodInfinibandPeerMem

func HasLsmodInfinibandPeerMem(lsmodOutput string) bool

Returns true if infiniband (ib_core module) is using nvidia_peermem.

Types

type LsmodPeermemModuleOutput

type LsmodPeermemModuleOutput struct {
	IbstatExists             bool   `json:"ibstat_exists"`
	InfinibandClassExists    bool   `json:"infiniband_class_exists"`
	Raw                      string `json:"raw"`
	IbcoreUsingPeermemModule bool   `json:"ibcore_using_peermem_module"`
}

func CheckLsmodPeermemModule

func CheckLsmodPeermemModule(ctx context.Context) (*LsmodPeermemModuleOutput, error)