Expand description
GPU acceleration module for scirs2-core
This module provides hardware acceleration support across different GPU backends.
Modules§
- async_
execution - Asynchronous execution and event-based synchronization for GPU operations
- auto_
tuning - Automatic kernel tuning and optimization for GPU operations
- backends
- GPU backend implementations and detection utilities
- benchmarks
- GPU vs CPU performance benchmarking suite
- heterogeneous
- Heterogeneous computing support for CPU-GPU hybrid operations
- kernels
- GPU kernel library for common scientific computing operations
- tensor_
cores - Tensor core acceleration support for modern GPUs
Structs§
- GpuBuffer
- GPU buffer
- GpuChannel
- GPU communication channel for multi-GPU operations
- GpuCompiler
- GPU compiler for dynamic kernels
- GpuContext
- GPU context for managing GPU resources and operations
- GpuDevice
- GPU device abstraction
- GpuKernel
- GPU kernel abstraction
- GpuKernel
Handle - GPU kernel handle
- GpuPtr
- GPU memory pointer abstraction
Enums§
- Dynamic
Kernel Arg - Non-generic kernel argument for mixed-type kernel calls
- GpuBackend
- GPU backend type
- GpuError
- Error type for GPU operations
- Kernel
Arg - Kernel argument types for GPU kernel execution
Traits§
- GpuData
Type - Trait for types that can be used with GPU operations