Statements (35)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:REST_API
|
| gptkbp:category |
GPU programming
|
| gptkbp:developedBy |
gptkb:NVIDIA
|
| gptkbp:documentation |
https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#cuda-graphs
|
| gptkbp:enables |
capture of GPU work as graphs
replay of GPU work graphs |
| gptkbp:feature |
event handling
dependency management asynchronous execution kernel launch batching memory operations batching node-based execution model |
| https://www.w3.org/2000/01/rdf-schema#label |
CUDA Graph API
|
| gptkbp:introducedIn |
gptkb:CUDA_10.0
|
| gptkbp:language |
gptkb:C++
C |
| gptkbp:license |
Proprietary
|
| gptkbp:partOf |
gptkb:CUDA_Toolkit
|
| gptkbp:platform |
gptkb:Windows
gptkb:macOS gptkb:Linux |
| gptkbp:purpose |
GPU workload optimization
|
| gptkbp:reduces |
CPU overhead
|
| gptkbp:relatedTo |
gptkb:cuGraph
gptkb:CUDA_Events gptkb:CUDA_Kernels gptkb:CUDA_Streams |
| gptkbp:supports |
graph execution
graph instantiation stream capture |
| gptkbp:usedIn |
high-performance computing
scientific computing deep learning frameworks |
| gptkbp:bfsParent |
gptkb:CUDA
|
| gptkbp:bfsLayer |
6
|