Struct GpuBackendDeviceStartupRecord 

pub struct GpuBackendDeviceStartupRecord {

    pub device_ordinal: usize,
    pub model_name: Option<String>,
    pub uuid: Option<String>,
    pub physical_total_bytes: Option<u64>,
    pub facts: DeviceFacts,
    pub capacity_bytes: u64,
    pub resource_profile: DeviceResourceProfile,
    pub memory_ownership: GpuMemoryOwnership,
    pub memory_limit: GpuDeviceMemoryLimit,
    pub backend_reserve_bytes: u64,
    pub max_active_attempts: usize,
    pub stream_pool_size: usize,
    pub worker_count: usize,
    pub lane_count: usize,
    pub install_telemetry: bool,
    pub memory_manager_installation: DeviceMemoryManagerInstallationState,
    pub device_health_generation: u64,
    pub backend_health_generation: u64,
}

Immutable per-device record frozen before backend publication.

Each record combines the discovered device facts with the configuration the device was installed under. It is read without any permit, so it stays available after close for diagnostics about a generation that no longer owns the device.

Fields

device_ordinal: usize

CUDA ordinal associated with this immutable installation record.

model_name: Option<String>

NVML product name captured with the immutable device inventory facts.

uuid: Option<String>

NVML UUID captured with the immutable device inventory facts.

physical_total_bytes: Option<u64>

NVML-reported physical framebuffer capacity captured with the immutable device inventory facts.

facts: DeviceFacts

Device facts sampled during backend construction.

capacity_bytes: u64

Managed capacity the ledger commits grants against: the resolved GpuDeviceMemoryLimit minus backend_reserve_bytes.

resource_profile: DeviceResourceProfile

Full ordinal-local resource-controller profile frozen after discovery.

memory_ownership: GpuMemoryOwnership

Deployment ownership promise used to resolve installed capacity.

memory_limit: GpuDeviceMemoryLimit

Admission capacity policy used for this device installation.

backend_reserve_bytes: u64

Bytes excluded from managed query and cache capacity.

max_active_attempts: usize

Maximum simultaneous device-backed admitted attempts.

stream_pool_size: usize

Installed CUDA stream-pool capacity.

worker_count: usize

Installed execution-worker thread count, including lifecycle workers.

lane_count: usize

Installed device execution-lane count.

install_telemetry: bool

Whether diagnostic telemetry was requested for this installation.

memory_manager_installation: DeviceMemoryManagerInstallationState

Native memory-manager installation status recorded during construction.

device_health_generation: u64

Persistent engine health epoch bound to this device installation.

backend_health_generation: u64

Backend-owner health epoch for this published core generation.

Trait Implementations

impl Clone for GpuBackendDeviceStartupRecord

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for GpuBackendDeviceStartupRecord

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Eq for GpuBackendDeviceStartupRecord

impl PartialEq for GpuBackendDeviceStartupRecord

fn eq(&self, other: &Self) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for GpuBackendDeviceStartupRecord