The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas...
Transcript of The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas...
![Page 1: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/1.jpg)
®
Lukas Breitwieser
Early-Career Researchers in Medical ApplicationsShort Talks on Computing and Simulation
The Anticipated Challenges ofRunning Biomedical Simulations
in the Cloud
1
![Page 2: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/2.jpg)
Understanding DevelopmentalDiseases
2
![Page 3: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/3.jpg)
Executive Summary
Motivation: Run tightly-coupled HPC workloads in the cloud
Widely accessible, cost-effective
Problem: Data exchange between servers will be a bottleneck
Key Idea: Exploit inherent simulation characteristics to reduce
data volume3
![Page 4: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/4.jpg)
Outline
1. Biological Simulation Basics
2. Distributed Runtime
3. HPC on Cloud
4. Data Movement Minimization
4
![Page 5: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/5.jpg)
From Atoms to Organisms...
Atom
Molecule
Macromolecule
Cells
Tissue
Organ
Organism
5
![Page 6: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/6.jpg)
Demo: Tumor Growth
6
![Page 7: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/7.jpg)
Agent-based simulations
Simulation object = Agent
Collision
Local region
7
![Page 8: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/8.jpg)
Outline
1. Biological Simulation Basics
2. Distributed Runtime
3. HPC on Cloud
4. Data Movement Minimization
8
![Page 9: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/9.jpg)
Domain-Decomposition
Hauri, Andreas. Self-construction in the context of cortical growth. Diss. 2013. 9
![Page 10: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/10.jpg)
Distributed Runtime
10
Frontend
master
worker
![Page 11: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/11.jpg)
Border Region
11
![Page 12: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/12.jpg)
Outline
1. Biological Simulation Basics
2. Distributed Runtime
3. HPC on Cloud
4. Data Movement Minimization
12
![Page 13: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/13.jpg)
NAMD Scaling
A. Gupta et al., “Evaluating and Improving the Performance and Scheduling of HPC Applications in Cloud,”
core count
time
in s
econ
ds
13
![Page 14: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/14.jpg)
Performance Issues in the Cloud
Poor network performance compared tosupercomputersVirtualizationResource contention with other tenants
14
![Page 15: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/15.jpg)
Outline
1. Biological Simulation Basics
2. Distributed Runtime
3. HPC on Cloud
4. Data Movement Minimization
15
![Page 16: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/16.jpg)
Key Observations
Some regions and data members are staticChanges are incrementalValues might be predictedCommunication can be replaced withadditional computation
16
![Page 17: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/17.jpg)
Static Regions
Example: growth of the cerebral cortex
https://www.youtube.com/watch?v=9InvFfnAkus
Definition: simulation objects whose values do notchange along the time dimension
17
![Page 18: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/18.jpg)
Static Regions
S. Liu, X. Huang, Y. Ni, H. Fu, and G. Yang, “A High Performance Compression Method for Climate Data,” in 2014 IEEE International
Climate Simulation: continental data does not changeduring simulation of the ocean surface
18
![Page 19: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/19.jpg)
Static Data Members
class NeuriteSegment { std::array<double, 3> position_; NeuriteSegment* parent_; ...};
Neuron
Soma Neurite segment
19
![Page 20: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/20.jpg)
Incremental ChangesObservation: Change between two time steps mightbe insignificantIdea: Communicate only significant updates Example: GAIA: Geo-distributed machine learningapproaching LAN speeds
K. Hsieh et al., “Gaia: Geo-Distributed Machine Learning Approaching LAN Speeds,” in 14th USENIX Symposium on NetworkedSystems Design and Implementation (NSDI 17).
20
![Page 21: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/21.jpg)
Value PredictionIdea: Predict values for safe-to-approximate variablesInspiration from computer architecture:"Rollback free value prediction"
A. Yazdanbakhsh, G. Pekhimenko, B. Thwaites, H. Esmaeilzadeh, O. Mutlu, and T. C. Mowry, “RFVP: Rollback-free value predictionwith safe-to-approximate loads,” ACM Transactions on Architecture and Code Optimization (TACO), vol. 12, no. 4, p. 62, 2016.
class NeuriteSegment { std::array<double, 3> position_; NeuriteSegment* parent_; ...};
21
![Page 22: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/22.jpg)
Computation vs Communication
Idea: Recompute certain events on the destinationserver instead of transferring the resultsExamples:
Cell devisionNeurite
extensionbifurcation...
Send event descriptor instead of whole new simulationobject
22
![Page 23: The Anticipated Challenges of Running Biomedical Simulations in … · 2018-04-03 · Lukas Breitwieser Early-Career Researchers in Medical Applications Short Talks on Computing and](https://reader033.fdocuments.in/reader033/viewer/2022042116/5e934fffa5b3e36e3b5fe1e5/html5/thumbnails/23.jpg)
Next Steps
Develop the distributed runtimeVerify that network properties are a bottleneck inthe cloudProvide detailed analysis for differentapproachesDevelop new ideas
23