consider making cpp::reduce more accurate for inexact types

asherliu / thrust

Automatically exported from code.google.com/p/thrust

Apache License 2.0

0 stars 0 forks source link

consider making cpp::reduce more accurate for inexact types #377

Closed GoogleCodeExporter closed 8 years ago

GoogleCodeExporter commented 8 years ago

Some options
 (1) multi-level summation scheme
 (2) compensated summation
 (3) higher precision (float->double)

Option 1 is probably the best approach since it would work for user-defined 
types that included inexact members.

Original issue reported on code.google.com by wnbell on 20 Sep 2011 at 7:14

GoogleCodeExporter commented 8 years ago

I think (1) can be accomplished with a simple divide-and-conquer strategy that 
uses iteration when (last - first) < 100 or something.  We'd need to respect 
the iterator category though.

Original comment by wnbell on 12 Oct 2011 at 2:25

GoogleCodeExporter commented 8 years ago

Forwarded to https://github.com/thrust/thrust/issues/92

Original comment by jaredhoberock on 7 May 2012 at 9:56

Changed state: Forwarded