Statements (33)
Predicate | Object |
---|---|
gptkbp:instanceOf |
GPU programming concept
|
gptkbp:allows |
overlapping of computation and data transfer
|
gptkbp:API |
cudaLaunchKernel
cudaMemcpyAsync cudaStreamCreate cudaStreamDestroy cudaStreamSynchronize cudaStreamWaitEvent |
gptkbp:canBe |
asynchronous
synchronous |
gptkbp:defaultStream |
stream 0
|
gptkbp:enables |
task scheduling
concurrent execution fine-grained control of execution order stream parallelism |
https://www.w3.org/2000/01/rdf-schema#label |
CUDA Streams
|
gptkbp:introduced |
gptkb:NVIDIA
|
gptkbp:relatedTo |
gptkb:CUDA_Graphs
gptkb:CUDA_Events CUDA Context |
gptkbp:usedFor |
asynchronous operations
CUDA events CUDA kernels memory copy operations |
gptkbp:usedIn |
high-performance computing
scientific computing deep learning frameworks NVIDIA CUDA platform |
gptkbp:bfsParent |
gptkb:CUDA_C++
gptkb:CUDA_Graph_API gptkb:CUDA_Graphs gptkb:NVIDIA_CUDA_Toolkit |
gptkbp:bfsLayer |
7
|