Talk:CSC 456 Fall 2013/4b cv: Difference between revisions

From Expertiza_Wiki
Jump to navigation Jump to search
No edit summary
 
(One intermediate revision by the same user not shown)
Line 13: Line 13:
* Extending OpenMP for NUMA machines
* Extending OpenMP for NUMA machines
* Experimental comparison of memory management policies for NUMA multiprocessors (though old, from 1991)
* Experimental comparison of memory management policies for NUMA multiprocessors (though old, from 1991)
== Comments on third draft ==
This is a good organization.  Please cite sources that refer to the different fetch, placement, and replacement strategies.  If you do this at the start of class, I can help you during class.  First touch, round-robin, and random would be placement policies.  They should be covered in the appropriate section.
=== Comments on final draft ===
Good start, but I would've liked to see you go into more detail with performance results.  The section on OpenMP is not well integrated into the rest of the discussion, and seems to be an abrupt ending.

Latest revision as of 00:27, 16 December 2013

Comments on second draft

Can't believe I didn't leave any comments before ...

The original first paragraph could be improved in readability. The sentence starting with, "Due to the access ..." is wordy and should be rewritten to progress in a logical order from the known (remote memory) to the unknown (how to handle page allocation).

I've divided your first paragraph into two. In general, shorter paragraphs are more readable than longer paragraphs.

I think I suggested several Google Scholar hits for "NUMA page allocation":

  • Power-aware page allocation
  • Operating system support for improving data locality on CC-NUMA compute servers
  • Extending OpenMP for NUMA machines
  • Experimental comparison of memory management policies for NUMA multiprocessors (though old, from 1991)

Comments on third draft

This is a good organization. Please cite sources that refer to the different fetch, placement, and replacement strategies. If you do this at the start of class, I can help you during class. First touch, round-robin, and random would be placement policies. They should be covered in the appropriate section.

Comments on final draft

Good start, but I would've liked to see you go into more detail with performance results. The section on OpenMP is not well integrated into the rest of the discussion, and seems to be an abrupt ending.