SC24 Home
Close

Session

Paper
: Communication Optimization for ML
Session Chair
Nikoli Dryden
Lawrence Livermore National Laboratory (LLNL)
Event Type
Paper
TimeThursday, 21 November 20241:30pm - 3pm EST
LocationB308
Tags
Artificial Intelligence/Machine Learning
Distributed Computing
Heterogeneous Computing
Performance Optimization
Registration Categories
TP
Similar Sessions
Machine Learning Applications
GPU Optimizations for ML
Efficient Transformers
Presentations
1:30pm - 2:00pm ESTOptimizing Distributed ML Communication with Fused Computation-Collective Operations
Authors
Kishore Punniyamurthy
Khaled Hamidouche
Bradford Beckmann
Artificial Intelligence/Machine Learning
Distributed Computing
Heterogeneous Computing
Performance Optimization
2:00pm - 2:30pm ESTAccelerating Communication in Deep Learning Recommendation Model Training with Dual-Level Adaptive Lossy Compression
Authors
Hao Feng
Boyuan Zhang
Fanjiang Ye
Min Si
Ching-Hsiang Chu
Jiannan Tian
Chunxing Yin
Zhaoxia (Summer) Deng
Yuchen Hao
Pavan Balaji
Tong Geng
Dingwen Tao
Artificial Intelligence/Machine Learning
Distributed Computing
Heterogeneous Computing
Performance Optimization
2:30pm - 3:00pm ESTAPTMoE: Affinity-aware Pipeline Tuning for MoE Models on Bandwidth-constrained GPU Nodes
Authors
Yuanxin Wei
Jiangsu Du
Jiazhi Jiang
Xiao Shi
Xianwei Zhang
Dan Huang
Nong Xiao
Yutong Lu
Artificial Intelligence/Machine Learning
Distributed Computing
Heterogeneous Computing
Performance Optimization