![Page 1: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/1.jpg)
Massively Parallel Astronomic Event Detection
using GPU’s
Matthias Alexander LeeWentworth Institute of Technology
Dr. Pavlos Protopapas
Help and Moral support:
Richard Edgar, Mike Clark and Tom Buckley
![Page 2: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/2.jpg)
What are we detecting?• Trans-Neptunian-Objects• Light Curves– Brightness of a star graphed over time– Massive Data Sets
0 5 10 15 20 25 30 35 40 45 50500
550
600
650
700
750
800
850
900
950
![Page 3: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/3.jpg)
Where’s the Data from?
Whipple• Proposed NASA satellite• Monitors 10,000 stars• 40hz 86,400s/day• Total of 3.456*10^10
![Page 4: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/4.jpg)
Search Method:Sliding Window Cross Correlation
• Generate Templates
0 5 10 15 20 25 30 35 40 450.7
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
![Page 5: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/5.jpg)
Search Method:Sliding Window Cross Correlation
• Generate Templates
0 5 10 15 20 250.7
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
![Page 6: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/6.jpg)
0 5 10 15 20
-5
-4
-3
-2
-1
0
1
2
Search Method:Sliding Window Cross Correlation
• Generate Templates
-3 2 7 12 17 22 27 32 37
-5
-4
-3
-2
-1
0
1
2
Event Detected!
0 2 4 6 8 10 12 14 16 18 200.7
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
![Page 7: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/7.jpg)
The Code: part I
• 2 Main Function.– PreProc()• Reads in Templates• Normalizes Templates• Generates
– Binary: contains all data– Dicionary: contains metadata about templates
![Page 8: Massively Parallel Astronomic Event Detection u sing GPU’s](https://reader036.vdocuments.mx/reader036/viewer/2022062520/56815e9b550346895dcd2eb5/html5/thumbnails/8.jpg)
The Code: part II
• 2 Main Function.– PreProc()– SearchLC()• Reads binary of Templates and Lightcurve
– Stores at Texture• Launches 30 blocks of 64 threads per LC
– Ensures shared memory is kept low enough for 2 blocks to be active on a multiprocessor, therefore hiding latency