C $Header: /u/gcmpack/MITgcm/eesupp/inc/GLOBAL_SUM.h,v 1.8 2009/06/10 03:45:11 jmc Exp $ C $Name: $ CBOP C !ROUTINE: GLOBAL_SUM.h C !INTERFACE: C include "GLOBAL_SUM.h" C !DESCRIPTION: C *==========================================================* C | GLOBAL\_SUM.h C | o Globals used by Fortran global sum routine. C *==========================================================* C | The global sum shared memory scheme uses global heap data| C | structures (.i.e COMMON blocks ). Each thread writes to | C | an its own element of the shared memory array and then | C | one thread reads all the entries and sums them. The sum | C | result is then read by all threads. | C | Remember - you are working with regions of memory that | C | are being updated concurrently by different threads. | C | What happens, when it happens and who gets to see what | C | happens at what stage depends on the computer systems | C | memory model. Every computer has a different memory model| C | and they are never simple. In all current platforms it is| C | possible for one thread to see events happening in a | C | different order from the order they are written in the | C | code. | C | Unless you understand this it is not a good idea to | C | make modifications te way these header files are setup or| C | the way the global sum routines work. | C *==========================================================* CEOP COMMON / GSUM_COMMON_R8 / phiGSR8, shareBufGSR8 c & , phiVGSR8 Real*8 phiGSR8 (lShare8, 0:MAX_NO_THREADS ) Real*8 shareBufGSR8 ( nSx, nSy ) c Real*8 phivGSR8(MAX_VGS, 0:MAX_NO_THREADS ) COMMON / GSUM_COMMON_R4 / phiGSR4 c & , phiVGSR4 Real*4 phiGSR4 (lShare4, 0:MAX_NO_THREADS ) c Real*4 phivGSR4(MAX_VGS, 0:MAX_NO_THREADS ) COMMON / GSUM_COMMON_I / phiGSI c & , phiVGSI INTEGER phiGSI (lShare4, 0:MAX_NO_THREADS ) c INTEGER phivGSI (MAX_VGS, 0:MAX_NO_THREADS ) CEH3 ;;; Local Variables: *** CEH3 ;;; mode:fortran *** CEH3 ;;; End: ***