Equation Solution  
    High Performance by Design
Navigation Tree  

 
neuLoop  
'- - Basic Concept
'- - Limitations & Status
'- - Do Subroutine
'- - Functions
|     '- - nlp$use
|     '- - nlp$done
|     '- - nlp$staysoftcore
|     '- - nlp$loop
|     '- - nlp$syncloop
|     '- - nlp$barrier
|     '- - Parallel Locks
|     '- - nlp$getce
|     '- - User Times
'- - Where To Download
'- - Manual Tool!



Limitations


  1. neuLoop is not allowed to be "nested". Dispatched jobs cannot dispatch other jobs.
  2. neuLoop is not "first-come, first-served". Once a loop or multiple loops are dispatched to soft cores, which loop or which do-index will be executed first or last cannot be predicted.
  3. User must know what he is doing in the do-statements. Whether the do-statements can be computed in parallel is programmer's responsibility. neuLoop is not an auto-parallelizer, but is a tool for programmers to parallelize do-loops.
  4. On x86-based Windows platform, neuLoop must run on Windows 7 or Windows server 2008 or above. The requirement is for neuLoop to support more than 64 cores. Windows 7 (64 bit) and Windows server 2008 support more than 64 cores. 32-bit Windows 7 does not support more than 64 cores, but neuLoop can run on 32-bit Windows 7 or above.
STATUS
  1. Presently, neuLoop is for FORTRAN programming. [neuLoop itself is also programmed in FORTRAN].
  2. neuLoop for Linux is under development.