Download presentation
Presentation is loading. Please wait.
Published bySullivan Postlewait Modified over 10 years ago
1
Parallelizing Video Transcoding With Load Balancing On Cloud Computing Song Lin, Xinfeng Zhang, Qin Y, Siwei Ma Circuits and Systems, 2013 IEEE
2
Outline Introduction Related work Problem formulation and system architecture Proposed method Experiment Results Conclusion
3
Introduction #1 Parallel programming Share memory Pthread – data dependency Message passing MPI – time delay
4
Introduction #2 Issues Data dependency Cost of data passing Load balance
5
Introduction #3 Cloud computation Data segmentation Computing capacity
6
Introduction #4 GOP-based encoding Independence between GOPs...........
7
Introduction #5 Paralleling in GOP-based Thread1 Thread2 Thread3
8
Related work #1 FCFS - First come first server [2] Easy to implement Load balancing problem is still exist
9
Related work #3 MCT – Minimal complete time [6] Map-Reduce-based
10
Problem formulation and system architecture #1 Load balance problem on cloud computing Executing time Delay time Data passing C is complexity and P is computing capacity
11
Problem formulation and system architecture #2 The overall completion time of set S k is. Goal.
12
Problem formulation and system architecture #3 Optimal solution. n means n task and m means m cores
13
Problem formulation and system architecture #4 Flow chart of the proposed method
14
Problem formulation and system architecture #5 For video coding, if the input sequence has instantaneous decoder refresh (IDR) frame, this video coding task can be divided into sub- tasks.[7]
15
Problem formulation and system architecture #6 For complexity estimation of video transcoding tasks, the existing algorithms [8] [9] can be utilized.
16
Proposed method #1 The framework includes three modules Task pre-analysis Adaptive threshold segmentation Minimal finish time
17
Proposed method #2 The threshold of segmentation
18
Proposed method #3
19
Proposed method #4 The optical finish time The finish time
20
Proposed method #5 Assign all the tasks sequentially in descending complexity order For each unassigned task j, the cores are judged in their descending computing capacity order according to the following criterion: assuming the task j is assigned to core k, if Τ κ T thr, the assignment is verified. Otherwise, we will judge the next core.
21
Proposed method #6 If all the cores are traversed and all the computing time are beyond T thr, the task j will be assigned by MCT algorithm. and T thr is updated to be the new finish time of the received core T k
22
Proposed method #7
23
Experiment results #1
24
Experiment results #2
25
Experiment results #3
26
Conclusion Load balancing problem is a NP-hard problem. The proposed algorithm has strong robustness to the task launching delay.
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.