Struct NvmeGpuIngestTelemetry 

pub struct NvmeGpuIngestTelemetry {
    pub submitted_bytes: u64,
    pub completed_bytes: u64,
    pub submitted_reads: u64,
    pub completed_reads: u64,
    pub registered_mapped_read_submissions: u64,
    pub gpudirect_nvme_submissions: u64,
    pub cpu_bounce_bytes: u64,
    pub failed_completions: u64,
}

Runtime telemetry for NVMe/file ingest into GPU-visible memory.

cpu_bounce_bytes is intentionally part of the public snapshot. The NvmeGpuIngestDriver never targets an ordinary userspace bounce buffer, so this counter must remain zero across both registered mapped reads and native GPUDirect NVMe passthrough.

Fields

submitted_bytes: u64

Bytes submitted to io_uring reads.

completed_bytes: u64

Bytes completed and published into the megakernel IO queue.

submitted_reads: u64

Total read submissions accepted by io_uring.

completed_reads: u64

Completed reads published into the megakernel IO queue.

registered_mapped_read_submissions: u64

Submissions using IORING_OP_READ_FIXED into registered GPU-visible memory.

gpudirect_nvme_submissions: u64

Submissions using native IORING_OP_URING_CMD NVMe passthrough into BAR1 memory.

cpu_bounce_bytes: u64

Bytes copied through ordinary userspace bounce buffers.

failed_completions: u64

CQEs that completed with an error or without matching pending metadata.

Implementations

impl NvmeGpuIngestTelemetry

pub fn inflight_reads(self) -> u64

Inflight read count derived from accepted submissions and terminal CQEs.

pub fn path_submissions(self, path: NativeReadPath) -> u64

Read submissions recorded for a specific native ingest path.

pub fn validate_completed_zero_copy( self, path: NativeReadPath, ) -> Result<(), PipelineError>

Validate that the snapshot describes a completed zero-copy ingest run for path.

This method intentionally does not validate a benchmark’s expected byte count; it validates the runtime invariant that every submitted read on the selected path completed without CPU bounce buffers or path mixing.

Errors

Returns PipelineError::Backend with an actionable fix when the snapshot reports a CPU bounce copy, failed/inflight reads, incomplete byte accounting, incomplete read accounting, or submissions on the wrong native ingest path.

Trait Implementations

impl Clone for NvmeGpuIngestTelemetry

fn clone(&self) -> NvmeGpuIngestTelemetry

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for NvmeGpuIngestTelemetry

impl Debug for NvmeGpuIngestTelemetry

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

Formats the value using the given formatter.

impl Default for NvmeGpuIngestTelemetry

fn default() -> NvmeGpuIngestTelemetry

Returns the “default value” for a type.

impl Eq for NvmeGpuIngestTelemetry

impl PartialEq for NvmeGpuIngestTelemetry

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for NvmeGpuIngestTelemetry