Creating nekbone repo from https://asc.llnl.gov/coral-2-benchmarks

CHANGES

@@ -0,0 +1,41 @@
+***********************************************************
+*              Changes in 2.0                             *
+***********************************************************
+ -Subroutine gsync() has changed to nekgsync() to avoid
+	possible conflict on certain architectures
+
+ -Executable is renamed 'nekbone' to replace 'nekproxy' and
+	other naming changes.
+
+ -iel0 and ielN  set in data.rea file are now used to control the
+ 	range of tests ran.  Test range in size from iel0
+	to ielN elements per process.  (prevoiusly tests 
+	were ran from 1 to lelt elements per process)  The
+	maximum value of ielN is lelt.
+
+ -nx0 and nxN set in data.rea file are now used to control the
+	range of polynomial orders.   Ranging from nx0 to 
+	nxN, where nxN<=lx1 (which is set in SIZE).  Previously
+	tests only ran with nx1=lx1.  The default is set to
+	reflect this, but nekbone now supports a range of
+	polynomial orders without recompiling the code.
+***********************************************************
+*              Changes in 2.1                             *
+***********************************************************
+ -Fixed nx0 and nxN control of polynomial order.  Default is
+ 	now to use lx1 until further notice.  Variable 
+	nx1 caused memory unstabilities and needs further 
+	development.
+ -Fixed a memory copy bug in the jl/ array transfer code.  
+	sarray_trasfer, used for the tuple transfer, should
+	be fixed now.
+
+***********************************************************
+*              Changes in 2.3                             *
+***********************************************************
+ - added OpenMP parallelism, MPITHREADS preprocessor macro
+   controls if MPI is called from one or multiple threads
+ - added timers controlled by TIMERS preprocessor macro
+ - fixed gather-scatter operation gsop() to always use pairwise
+   method
+ - switched to using system_clock routine in dummy mpi_wtime()

COPYRIGHT

@@ -0,0 +1,34 @@
+				  COPYRIGHT
+
+The following is a notice of limited availability of the code, and disclaimer
+which must be included in the prologue of the code and in all source listings
+of the code.
+
+Copyright Notice
+ + 2012 University of Chicago
+
+Permission is hereby granted to use, reproduce, prepare derivative works, and
+to redistribute to others.  This software was authored by:
+
+P. Fischer: (630) 252-6018; FAX: (630) 252-5986; email: [email protected]
+Mathematics and Computer Science Division
+Argonne National Laboratory, Argonne IL 60439
+
+			      GOVERNMENT LICENSE
+
+Portions of this material resulted from work developed under a U.S.
+Government Contract and are subject to the following license: the Government
+is granted for itself and others acting on its behalf a paid-up, nonexclusive,
+irrevocable worldwide license in this computer software to reproduce, prepare
+derivative works, and perform publicly and display publicly.
+
+				  DISCLAIMER
+
+This computer code material was prepared, in part, as an account of work
+sponsored by an agency of the United States Government.  Neither the United
+States, nor the University of Chicago, nor any of their employees, makes any
+warranty express or implied, or assumes any legal liability or responsibility
+for the accuracy, completeness, or usefulness of any information, apparatus,
+product, or process disclosed, or represents that its use would not infringe
+privately owned rights.
+

README.md

@@ -0,0 +1 @@
+# nekbone_2_3_5

src/DXYZ

@@ -0,0 +1,5 @@
+C
+C     Elemental derivative operators
+C
+      common /dxyz/ dxm1(lx1,lx1),  dxtm1(lx1,lx1)
+

src/INPUT

@@ -0,0 +1,19 @@
+C
+C     Input parameters from preprocessors.
+C
+C     Note that in parallel implementations, we distinguish between
+C     distributed data (LELT) and uniformly distributed data.
+C
+
+      common /input5/ xc(8,lelt),yc(8,lelt),zc(8,lelt)
+     $               ,bc(5,6,lelt,0:ldimt1)
+
+
+      common /input8/ cbc(6,lelt,0:ldimt1),ccurve(12,lelt)
+      character*1     ccurve
+      character*3     cbc
+
+      real mflops
+      integer*8 flop_a, flop_cg
+      common /cflops/ flop_a,flop_cg,mflops
+

src/MASS

@@ -0,0 +1,4 @@
+      common /mass/
+     $       bm1   (lx1,ly1,lz1,lelt)
+     $      ,binvm1(lx1,ly1,lz1,lelt)
+     $      ,volvm1

src/PARALLEL

@@ -0,0 +1,31 @@
+C
+C     Communication information
+C     NOTE: NID is stored in 'SIZE' for greater accessibility
+      common /cube1/ node,pid,np,nullpid,node0
+      integer        node,pid,np,nullpid,node0
+
+
+c     Maximum number of elements (limited to 2**31/12, at least for now)
+      parameter(nelgt_max = 178956970)
+
+      common /hcglb/ nvtot,nelgf(0:ldimt1)
+     $              ,lglel(lelt)
+c    $              ,gllel(lelg)
+c    $              ,gllnid(lelg)
+     $              ,nelgv,nelgt
+
+      integer        lglel
+c     integer        gllel,gllnid
+      integer*8      nvtot
+
+      common /diagl/  ifgprnt
+      logical ifgprnt
+      common/precsn/ wdsize,isize,lsize,csize
+      common/precsl/ ifdblas
+      integer wdsize,isize,lsize,csize
+      logical ifdblas
+C
+C     crystal-router, gather-scatter, and xxt handles (xxt=csr grid solve)
+C
+      common /comm_handles/ cr_h, gsh, gsh_fld(0:ldimt1), xxth(ldimt1)
+      integer               cr_h, gsh, gsh_fld          , xxth

src/README

@@ -0,0 +1,66 @@
+  Nek_comm-1.0
+
+This is the communication testing kernel for the MPI 
+all reduce and point to point communication used within
+the nekbone mini-app and standard NEK5000.  This kernel
+runs a battery of platform timers using MPI standard.
+
+To Run:
+
+  NOTE - Unlike the other nek codes, a data.rea file
+	is not needed, since there is no geometry
+	being set up.
+
+  After untarring nek_comm-1.0.tgz, change working 
+  directory to the nek_comm/test/ directory:
+     cd  ~/nek_comm/test/
+
+  Edit the makenek script to specify the compiler and
+  appropriate compiler flags.
+
+  Compile and link the code using the makenek script:
+      ./makenek n
+  where n is the chosen name of test run.
+
+  A successful compilation will result with:
+
+  #############################################################
+  #                  Compilation successful!                  #
+  #############################################################
+
+  And a nekcomm executable.
+
+  Run the code in parallel using the provided nekpmpi script
+  by specifying the test name (for logfile naming purposes) 
+  and the number of processors.  For example, to run a test
+  called 'n' on 4 processes:
+     ./nekpmpi n 4
+
+  This will produce a logfile, n.log.4.
+
+
+Interpreting results:
+
+  The logfile will have a header describing the parameters the
+  test was ran with and the output of the timing tests.
+
+  All reduce tests are output with the a 'gop' tag:
+		np  nwds time1 time2
+  where,
+	 np - number of processors
+	 nwds - number of words
+	 tmsg - time per message
+ 	 tpwd - time per word
+
+
+  Point to point tests are output with the 'pg' tag:
+		nodeb np nloop nwds tmsg tpwd
+  where,
+	 nodeb - the second processor node 0 is 
+	 testing with
+	 np - number of processors
+	 nloop - number of tests ran with these nodes
+	 nwds - number of words per message
+	 tmsg - time per message
+	 tpwd - time per word  
+

src/TIMER

@@ -0,0 +1,19 @@
+
+      integer tmax
+      parameter (tmax = 1024)
+
+      integer gopi(tmax)
+
+      real  ttemp1, ttemp2, ttemp3, ttemp4
+
+      real trzero(tmax), tcopy(tmax), tsolvem(tmax)
+      real tglsc3a(tmax), tglsc3b(tmax), tglsc3c(tmax), tglsc3d(tmax)
+      real tadd2s1(tmax), tadd2s2a(tmax), tadd2s2b(tmax), tadd2s2c(tmax)
+      real tlocalgrad3(tmax), twrwswt(tmax), tlocalgrad3t(tmax)
+      real tgsop(tmax), tgop(4,tmax)
+
+      real*8 dnekclock
+
+      common /timer/ trzero, tcopy, tsolvem, tglsc3a, tglsc3b, tglsc3c,
+     +tglsc3d, tadd2s1, tadd2s2a, tadd2s2b, tadd2s2c, tlocalgrad3,
+     +twrwswt, tlocalgrad3t, tgsop, tgop, gopi

src/TOTAL

@@ -0,0 +1,5 @@
+      include 'DXYZ'
+      include 'INPUT'
+      include 'MASS'
+      include 'PARALLEL'
+      include 'WZ'

src/WZ

@@ -0,0 +1,7 @@
+
+
+c     Gauss-Labotto and Gauss points
+      common /gauss/ zgm1(lx1,3)
+
+c     Weights
+      common /wxyz/ wxm1(lx1), wym1(ly1), wzm1(lz1), w3m1(lx1,ly1,lz1)

src/bg_aligned3.s

@@ -0,0 +1,41 @@
+.set r0,0; .set r1,1; .set r2,2; .set r3,3; .set r4,4
+.set r5,5; .set r6,6; .set r7,7; .set r8,8; .set r9,9
+.set r10,10; .set r11,11; .set r12,12; .set r13,13; .set r14,14
+.set r15,15; .set r16,16; .set r17,17; .set r18,18; .set r19,19
+.set r20,20; .set r21,21; .set r22,22; .set r23,23; .set r24,24
+.set r25,25; .set r26,26; .set r27,27; .set r28,28; .set r29,29
+.set r30,30; .set r31,31
+.set f0,0; .set f1,1; .set f2,2; .set f3,3; .set f4,4
+.set f5,5; .set f6,6; .set f7,7; .set f8,8; .set f9,9
+.set f10,10; .set f11,11; .set f12,12; .set f13,13; .set f14,14
+.set f15,15; .set f16,16; .set f17,17; .set f18,18; .set f19,19
+.set f20,20; .set f21,21; .set f22,22; .set f23,23; .set f24,24
+.set f25,25; .set f26,26; .set f27,27; .set f28,28; .set f29,29
+.set f30,30; .set f31,31
+
+.file "bg_aligned3.s"
+
+.globl bg_aligned3
+.type  bg_aligned3, @function
+.size  bg_aligned3, 48
+
+.section ".text"
+.align 2
+
+bg_aligned3:
+  andi.    r0,r3,15
+  clrlwi   r9,r4,28
+  cmpwi    cr7,r9,0
+  li       r3,0
+  li       r0,0
+  bne-     .L.das_label.58
+  andi.    r9,r5,15
+  bne-     cr7,.L.das_label.58
+  bne-     .L.das_label.58
+  li       r0,1
+ .L.das_label.58:
+  stw      r0,0(r6)
+  blr     
+
+
+.ident "GCC: (GNU) 3.2"