Sunday 12-09-01

What to do?

YURY in shame: get the f... scheduling working!

1. Try to have stream_read "get" as many tokens as available (yatish)
2. Improved linked list/queue  (yatish)

3. Get all numbers for latency/bw  (norm)

   a. Graphs:
      1. Bandwidth: local, remote
      2. latency: local, remote

4. Get numbers on granularity of computation per node (norm)
   a. granularity: when does it make sense to scale?

5. Make bigger synthetic apps (norm)

6. Develop the Makespan = .... model and verify that it fits with the
   expected.

7. Rerun our apps with all scheds.

8. How do we interpret all numbers we get, i.e. what statistics to
   plot and make conclusions from? (yurym)

9. What percentage of user time is overhead?  (yurym)

10. Run wavelet with larger dataset? (who???)