Page ThesisMCRazerWeeklyReports

Week 1 (25.1.)

• Q-gram index with open hashing

Week 2 (1.2.)

• Adapt mapRead functions to parallelism
• Overview over parallelization options
• Speed test between normal qgram index and open addressing version

Week 3 (8.2.)

• Slides for the talk at the Charité
• Parallel version compiles without -fopenmp
• hunt for segmentation fault

Week 4 (15.2.)

• Presentation at the Charité
• Herr Kleffe erwartet sich:
  • Einen schnellen read mapper
  • Reduktion der Indexgröße, bzw. weniger Speicherverbrauch
• Exam preparation

Week 5 (22.2.)

• Window version compiles and runs with openmp
• Output for speed analysis (how many base pairs per sec., how many verifications per sec., for each thread and block)

Week 6 (1.3.)

• R script to analyse the 'speed output'
• Documentation for findWindow functions
• Window has the same speed as the normal version.

Week 7 (8.3.)

• reduce the size of the critical region
• analyse higher q's

Week 8 (15.3.)

• script that splits up the reads and runs them with the serial version
  • On small sample (100,000 reads) the parallel version is faster than the parallelizing script
• comparing different window sizes and granularity of the blocks

Week 9 (22.3.)

• speed comparison for different number of cores