Enum ComputeDeviceType

Namespace

DotCompute.Abstractions.Pipelines.Enums

Assembly

DotCompute.Abstractions.dll

Specifies the type of compute device used for pipeline execution.

public enum ComputeDeviceType

Fields

ARM = 12

ARM-based processor with NEON SIMD capabilities. Common in mobile and embedded devices.

ASIC = 8

Application-Specific Integrated Circuit for specialized computations. Fixed-function hardware optimized for specific algorithms.

Auto = 20

Any available device, letting the system choose automatically. Adaptive selection based on workload characteristics.

CPU = 1

Central Processing Unit with general-purpose computing capabilities. Supports multi-core parallel execution and SIMD vectorization.

CUDA = 2

NVIDIA Graphics Processing Unit with CUDA support. Optimized for massively parallel computations.

Cloud = 18

Cloud-based virtual compute resource. Elastic computing with pay-per-use pricing model.

Cluster = 19

Distributed computing cluster across multiple nodes. Scalable computing for large-scale data processing.

DSP = 9

Digital Signal Processor optimized for signal processing algorithms. Specialized for real-time audio and video processing.

EdgeAI = 15

Edge computing device with specialized AI acceleration. Optimized for low-power, real-time inference.

FPGA = 7

Field-Programmable Gate Array for custom hardware acceleration. Provides reconfigurable computing capabilities.

Hybrid = 17

Hybrid device combining multiple compute types. Heterogeneous computing with automatic work distribution.

IntelGPU = 6

Intel Graphics integrated or discrete GPU with OneAPI support. Intel's unified programming model for heterogeneous computing.

Metal = 4

Apple Metal Performance Shaders on macOS and iOS devices. Native GPU computing framework for Apple platforms.

NPU = 10

Neural Processing Unit specialized for machine learning workloads. Optimized for tensor operations and neural network inference.

OpenCL = 5

OpenCL-compatible device supporting cross-platform parallel computing. Can target CPUs, GPUs, and other accelerators.

QPU = 16

Quantum processing unit for quantum computing algorithms. Experimental platform for quantum advantage applications.

RISCV = 13

RISC-V processor with open-source instruction set architecture. Emerging platform for custom and embedded computing.

ROCm = 3

AMD Graphics Processing Unit with ROCm/HIP support. Alternative GPU platform for parallel computing.

TPU = 11

Tensor Processing Unit designed for machine learning acceleration. Google's custom silicon for AI workloads.

Unknown = 0

Unknown or unspecified device type.

WebAssembly = 14

WebAssembly runtime for portable, sandboxed execution. Cross-platform bytecode format for web and edge computing.