Changes between Version 6 and Version 7 of SSE2_Hoisting


Ignore:
Timestamp:
May 27, 2014, 9:05:59 PM (5 years ago)
Author:
cameron
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • SSE2_Hoisting

    v6 v7  
    2222the penalty will be small.
    2323
     24In general, there may be more than one translation strategy for a given SSE2 intrinsic.
     25   a.  One strategy should be general purpose, producing an LLVM IR representation that works for all possible uses of the intrinsic.
     26   b.  Other strategies can be special-purpose, producing shorter LLVM IR sequences that apply in special cases.   
     27
    2428The final project report should include a table listing *all* SSE2
    2529intrinsics and the hoisting transformation used.   If the hoisting
     
    4044
    4145    a.  In each case that an SSE2 intrinsic requires a sequence of LLVM operations (excluding bitcasts and constants), ensure that SSE2 code generation recognizes the transformed sequence to allow the single intrinsic to be produced during code generation.
    42     a.  Modify the code generator for at least one other target to recognize sequences produces by SSE2 hoisting and generate efficient code based on that recognition.
     46    a.  Modify the code generator for at least one other target to recognize sequences produced by SSE2 hoisting and generate efficient code based on that recognition.
    4347
    4448== Project Evaluation ==
     
    6367  a.  The result of the hoisted code complied with non x86 architecture, such as ARM with Neon SIMD instructions.
    6468
     69=== Parabix Transform ===
     70
     71One possible evaluation for this project is to assess against a Parabix application such as regular expression matching.
     72In this case, the pack and shift operations required to implement the ParabixTransform need special attention.
     73
    6574
    6675== I2Result Issue ==