Home HPC
Accelerated Computing Compute Optimization GPU Acceleration Multi GPU Scalability
VidAi Video Technologies MyClass ☰
Optimization Strategy
Photon Computing Optimization Strategy establishes a methodology, necessary for implementing a successful optimization process.
Our strategy directs the optimization process to start at the earliest level possible, and ends up enabling proper Parallelization and Vectorization techniques.
The Optimization Strategy directs to:
1. Identify the actual Challenges
2. Specify a Target solution approach
3. Tackle each challenge at the Earliestlevel possible
4. Enhance, Modify, or Change the scientific model/approach , at the subject level, in order define an Alternative that aligns best with the specified approach
5. Verify the defined alternative suitability for downstream subjects
6. Validate the proposed solution correctness, and efficiency.
Although, applying Parallelization and Vectorization in order to utilize the processing units cores as well as the wide registers on each core looks like a good approach to perform optimization of a compute intensive system, the truth is, this direct approach might result in a little performance gain.
The actual challenges of a compute intensive system must be defined, it might be the data Volume, data Transfer operations, or the computation.

Case Study
Optimization Strategy for Compute Intensive Systems - The World's Fastest CFD
Photon Computing session at the "High Performance Computing for Mathematics and Applications" conference, Faculty of Science, Suez University.