![]() |
OpenVDB 13.1.0
|
Files | |
| Buffer.h | |
| Typed containers for CUDA memory: the owning, resource-aware, stream-ordered cuda::Buffer and the non-owning cuda::BufferView. | |
| DeviceBuffer.h | |
| DualDeviceBuffer has one pinned host buffer and multiple device CUDA buffers. | |
| DeviceMesh.h | |
| nanovdb::cuda::DeviceMesh encapsulates device IDs, CUDA streams, and NCCL communicators in order to facilitate multi-GPU applications. | |
| DeviceResource.h | |
| DeviceStreamMap.h | |
| nanovdb::cuda::DeviceStreamMap maps device IDs to CUDA streams, which is useful for multi-GPU applications. | |
| HandleStorage.h | |
| Allocates the device-resident storage behind a GridHandle or NodeManagerHandle for any buffer family: through the static create() interface for buffers that provide it (the dual-space DeviceBuffer family), and through the buffer's memory resource for a single-space cuda::Buffer. This is the bridge that lets every tool entry point accept either buffer family. | |
| ManagedResource.h | |
| PinnedResource.h | |
| TempPool.h | |
| UnifiedBuffer.h | |
| nanovdb::cuda::DualUnifiedBuffer that uses unified memory management | |