Ghosh, Abhishek <br/><br/>
PyGraph :   compiler support for efficient and transparent use of CUDA graphs 

 - Bangalore :   Indian Institute of Science.  2024. 
 - vii, 41 p. :   col. ill.   e-Thesis  1.670 Mb 
<br/><br/> Includes bibliographical references. 
<br/><br/>MTech(Res);2025;Computer Science and Automation. 
<br/><br/> CUDA Graphs --- a recent hardware feature introduced for NVIDIA GPUs --- aim to reduce CPU launch overhead by capturing and launching a series of GPU tasks (kernels) as a DAG. However, deploying CUDA Graphs faces several challenges today due to the static structure of a graph. It also incurs performance overhead due to data copy. In fact, we show a counter-intuitive result --- deploying CUDA Graphs hurts performance in many cases. 
<br/><br/><label>Subjects--Topical Terms: </label><br/>GPUs<br/>Hardware-software interface<br/>Performance<br/>GPU software<br/>ML Compilers<br/>CUDA Graphs
<br/><br/><label>Dewey Class. No.: </label>621.39   / BAS