HIP

E284611

HIP is a C++ runtime and programming model developed by AMD that enables portable GPU-accelerated code across AMD and NVIDIA hardware.

All labels observed (1)

Label Occurrences
HIP canonical 2

How this entity was disambiguated

Statements (48)

Predicate Object
instanceOf C++ runtime
GPU programming model
abstractionLevel low-level GPU programming API
backend CUDA backend for NVIDIA GPUs
ROCm backend for AMD GPUs
category GPGPU framework
parallel programming framework
compatibleWith CUDA programming model
linked to: NVIDIA CUDA
competesWith CUDA
linked to: NVIDIA CUDA

OpenCL
designGoal performance portability across GPU vendors
source-level portability
developer AMD
ecosystem ROCm
enables portable GPU-accelerated code
executionModel SIMT-style GPU execution
fullName Heterogeneous-computing Interface for Portability
hasTool hipify
hipifyPurpose assist CUDA-to-HIP source translation
intendedUsers C++ GPU application developers
interoperatesWith ROCm libraries
license open source
memoryModel explicit device and host memory management
optimizationGoal near-native performance on each GPU backend
partOf AMD ROCm software stack
programmingLanguage C++
provides CUDA-like API
kernel launch syntax similar to CUDA
runtime API for GPU programming
repositoryHost GitHub
supports GPU offloading
asynchronous GPU operations
device kernels
events
heterogeneous computing
memory management on GPU
multi-GPU programming
streams
supportsHardwareVendor AMD
NVIDIA
linked to: NVIDIA Corporation
supportsVendor AMD GPUs
linked to: AMD Radeon GPUs

NVIDIA GPUs
linked to: Nvidia Maxwell GPU
targetPlatform Linux
Windows
useCase data analytics
high-performance computing
machine learning workloads
scientific simulations

How these facts were elicited

Referenced by (2)

Full triples — surface form annotated when it differs from this entity's canonical label.

OpenACC competesWith HIP
CMake supportsLanguage HIP