08.079.590 Parallel Computing Seminar

Instructors: Univ-Prof. Dr. Bertil Schmidt

Event type: Seminar

Displayed in timetable as: 08.079.590

Hours per week: 2

Credits: 4,0

Language of instruction: German

Min. | Max. participants: - | 15

Requirements / organisational issues:
Prerequisites:

Successful partitication in PAA or HPC.

Contents:
Topic Selection:

E-Mail your two preferred topics to  Prof. Dr. Bertil Schmidt by 12th July 2019 latest. 
I will then assign a topic to you.
Note: There is also the otion of suggesting a topic by yourself (needs my approval)

Topics:

1. Accelerating Reduction and Scan Using Tensor Core Units
   
2. diBELLA: Distributed Long Read to Long Read Alignment
   
3. Performance extraction and suitability analysis of multi- and many-core architectures for next generation sequencing secondary analysis
   
4. Adaptive Sparse Matrix-Matrix Multiplication on the GPU
   
5. GPU-Accelerated Large-Scale Genome Assembly
   
6. Hardware-conscious Hash-Joins on GPUs
   
7. Harnessing GPU tensor cores for fast FP16 arithmetic to speed up mixed-precision iterative refinement solvers
   
8. GRASShopPER—An algorithm for de novo assembly based on GPU alignments
   
9. Exascale Deep Learning for Climate Analytics
   
10. Efficient Architecture-Aware Acceleration of BWA-MEM for Multicore Systems
    
11. A Modular Benchmarking Infrastructure for High-Performance and Reproducible Deep Learning
    
12. UPC++: A High-Performance Communication Framework for Asynchronous Computation
    
13. Large-Scale Hierarchical k-means for Heterogeneous Many-Core Supercomputers
    
14. A Framework for the Automatic Vectorization of Parallel Sort on x86-based Processors
    
15. self-suggested topic (needs approval)