Difference between revisions of "Teleconference 2015-03-25"

From VTKM
Jump to navigation Jump to search
(Created page with "participants: Hank, Dave, Jeremy * Hank: discussion of writing, essentially a VTK-m textbook simultaneously with a grad class in the Fall * we will discuss more later * Jer...")
 
Line 2: Line 2:
  
 
* Hank: discussion of writing, essentially a VTK-m textbook simultaneously with a grad class in the Fall
 
* Hank: discussion of writing, essentially a VTK-m textbook simultaneously with a grad class in the Fall
  * we will discuss more later
+
** we will discuss more later
 
* Jeremy, Dave: quick progress updates, some discussion on these topics:
 
* Jeremy, Dave: quick progress updates, some discussion on these topics:
  * grouped vs ungrouped connectivity:  
+
** grouped vs ungrouped connectivity:  
    * no one has yet come up with a code that natively highly mixes cell types
+
*** no one has yet come up with a code that natively highly mixes cell types
    * hank suggest some algorithms (clipping, material interface, etc.) may generate highly mixed or grouped, though we're not sure that grouped performs better than mixed due to  
+
*** hank suggest some algorithms (clipping, material interface, etc.) may generate highly mixed or grouped, though we're not sure that grouped performs better than mixed due to  
    * there are potentially upsides and downsides to both:
+
*** there are potentially upsides and downsides to both:
      * memory usage favors grouped: minimal memory penalty in the worst case (e.g. alternating tris with quads) and moderate memory benefit in best case
+
**** memory usage favors grouped: minimal memory penalty in the worst case (e.g. alternating tris with quads) and moderate memory benefit in best case
      * grouped makes celltype-specific kernel launches trivially easy
+
**** grouped makes celltype-specific kernel launches trivially easy
      * but may be one extra indirection for grouped (must look up group index and connectivity index from cell index; either two separate cell-length arrays, or look up group index in cell-length array then look up connectivity startindex in group-length array)
+
**** but may be one extra indirection for grouped (must look up group index and connectivity index from cell index; either two separate cell-length arrays, or look up group index in cell-length array then look up connectivity startindex in group-length array)
    * as such, since it is pretty easy to implement both and convert between, we may simply do both
+
*** as such, since it is pretty easy to implement both and convert between, we may simply do both
  * some brief discussion of allowing more than two topology types in a single worklet (e.g. node-and-face-and-cell to cell), decided to pass for now for lack of concrete examples; can always add later
+
** some brief discussion of allowing more than two topology types in a single worklet (e.g. node-and-face-and-cell to cell), decided to pass for now for lack of concrete examples; can always add later

Revision as of 13:48, 25 March 2015

participants: Hank, Dave, Jeremy

  • Hank: discussion of writing, essentially a VTK-m textbook simultaneously with a grad class in the Fall
    • we will discuss more later
  • Jeremy, Dave: quick progress updates, some discussion on these topics:
    • grouped vs ungrouped connectivity:
      • no one has yet come up with a code that natively highly mixes cell types
      • hank suggest some algorithms (clipping, material interface, etc.) may generate highly mixed or grouped, though we're not sure that grouped performs better than mixed due to
      • there are potentially upsides and downsides to both:
        • memory usage favors grouped: minimal memory penalty in the worst case (e.g. alternating tris with quads) and moderate memory benefit in best case
        • grouped makes celltype-specific kernel launches trivially easy
        • but may be one extra indirection for grouped (must look up group index and connectivity index from cell index; either two separate cell-length arrays, or look up group index in cell-length array then look up connectivity startindex in group-length array)
      • as such, since it is pretty easy to implement both and convert between, we may simply do both
    • some brief discussion of allowing more than two topology types in a single worklet (e.g. node-and-face-and-cell to cell), decided to pass for now for lack of concrete examples; can always add later