NARU API
    Preparing search index...

    Interface RendererResourceStats

    Allocation census of renderer-owned resources, split so that each number has one owner and the inclusion relationships between them are stated rather than inferred. Every buffer figure is exact -- WebGPU allocates what was asked for -- while the attachment figure is an upper bound (see attachmentPairByteLength).

    interface RendererResourceStats {
        attachmentSizePx: readonly [number, number] | null;
        cpuStagingBytes: number;
        gpuAttachmentBytes: number;
        gpuBatchBufferBytes: number;
        gpuBufferBytes: number;
        gpuUniformBytes: number;
        gpuVertexPoolBytes: number;
    }
    Index
    attachmentSizePx cpuStagingBytes gpuAttachmentBytes gpuBatchBufferBytes gpuBufferBytes gpuUniformBytes gpuVertexPoolBytes
    attachmentSizePx: readonly [number, number] | null

    Current depth/pick attachment size in pixels, when a frame has rendered.

    cpuStagingBytes: number

    CPU staging memory retained for per-prototype instance re-packing.

    gpuAttachmentBytes: number

    Upper bound on the depth and object-id attachments at the current size. Zero before the first sizing, and excluded from gpuBufferBytes because a texture is not a buffer allocation.

    gpuBatchBufferBytes: number

    Per-batch index, edge, and instance buffers. Disjoint from the pools.

    gpuBufferBytes: number

    The three figures above, and nothing else.

    gpuUniformBytes: number

    The single camera uniform buffer, which no batch owns.

    gpuVertexPoolBytes: number

    Prototype vertex pools, charged once each however many material groups share one. Summing per batch would count the largest sixty5 prototype's pool 111 times.

    Settings
    Member Visibility

    On This Page
    Properties