#3659: two-dimensional PArrays in data parallel code
------------------------------+---------------------------------------------
 Reporter:  ams               |          Owner:                  
     Type:  feature request   |         Status:  new             
 Priority:  normal            |      Milestone:                  
Component:  Compiler          |        Version:  6.10.4          
 Severity:  normal            |     Resolution:                  
 Keywords:                    |       Testcase:                  
       Os:  Unknown/Multiple  |   Architecture:  Unknown/Multiple
------------------------------+---------------------------------------------
Changes (by chak):

 * cc: [email protected] (added)

Comment:

 The function `fromList` doesn't currently work on lists of parallel
 arrays.  It arguably should, though.

 From what you are writing, I am not sure whether you want to work
 explicitly with `PArray`s or whether you want to work with `[:t:]` types.
 In the former case, you can use `nestUSegdPA` to construct a nested array.
 In the latter case, you currently probably have to use a combination of
 `singletonP` and `(+:+)` to first create singleton nested arrays from all
 your subarrays and then concatenate them.

 I notice that you are calling the nested array a ''two-dimensional
 array''.  If your aim is to have subarrays of the same size and to operate
 in a regular manner on the nested array (e.g., implementing ''dense''
 matrix operations), you will probably find that the performance isn't
 great and the code is a bit awkward.  The reason is that the current
 system is biased towards nested, irregular arrays.  We are currently
 working on complementing the existing API with explicit support for
 regular, multi-dimensional arrays to address this issue.

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/3659#comment:1>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs

Reply via email to