notesum.ai

Published at November 5

DeepContext: A Context-aware, Cross-platform, and Cross-framework Tool for Performance Profiling and Analysis of Deep Learning Workloads

cs.PF

cs.AI

Released Date: November 5, 2024

Authors: Qidong Zhao¹, Hao Wu², Yuming Hao, Zilingfeng Ye², Jiajia Li¹, Xu Liu¹, Keren Zhou²

Aff.: ¹North Carolina State University; ²George Mason University

Arxiv: http://arxiv.org/abs/2411.02797v1

Refer to caption

Profiling Tools	Python Context	Framework Context	C++ Context	Device Context	Cross GPUs	Cross Frameworks	CPU Profiling
Nsight Systems (nsight_systems, )	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$
RocTracer (roctracer, )	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$
JAX profiler (jax_profiler, )	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$
PyTorch profiler (pytorch_profiler, )	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\times}$	$\boldsymbol{\checkmark}$
DeepContext	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$	$\boldsymbol{\checkmark}$