This is the software I developed for Task 2.2 of the "TASTE Maintenance and Evolutions" project of the European Space Agency. It is a port of the IDL implementation of the StrayLight algorithm to C++, and it utilizes OpenMP, Eigen and CUDA to achieve much better speeds than the original IDL code (execution time in gdl: 18 seconds - execution time of my C++ code: 169 ms)
Results for single precision speed - time per frame:
$ ./configure && make && ./src/strayLight -b
Execution time per image : 169 ms
Average and std deviation: 171.23 +/- 1.6 ms
Results for double precision speed - time per frame:
$ ./configure --enable-double && make && ./src/strayLight -b
Execution time per image : 222 ms
Average and std deviation: 224.46 +/- 1.9 ms
I wrote extensive documentation of the optimizations I performed to achieve these speedups inside the deliverable in the doc/ folder.
$ ./src/strayLight -h
strayLight 1.1c (Thu May 30 14:57:08 2013)
Usage: strayLight [OPTIONS]
-h this help
-v increase verbosity
-V show version
-b run benchmark (50 images)
-i filename instead of the ESA test image, process this file
-c channel process this channel
-d N dump log files from computation stages >= N
(default: 13, i.e. the final result is saved,
and never used during benchmarking (no output).
To create a log of the output image generated by the StrayLight algorithm:
$ mkdir -p output
$ ./src/strayLight
This will use the default value of option -d, and generate the stage13 output file. If you wish to generate outputs from all stages, starting with stage 10:
$ ./src/strayLight -d 10
This will generate log files for stages 10, 11, 12 and 13.
To compare the final output (stage13) with the outputs generated by IDL, run this:
$ ./contrib/elementsDiff.py \
./output/stage13_1 \
./output.from.IDL/stage13_1 | ./contrib/stats.py | grep Overall
Overall: 6.61607071392e-06 +/- 375.7%
...which will verify that the difference between the values in the two outputs (from IDL and from C++) is indeed minimal.
Thanassis Tsiodras, Dr.-Ing. [email protected] / [email protected] / [email protected]
straylight's People
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. ๐๐๐
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google โค๏ธ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.