Merge 5.1 release branch into amd-fftw

Author: BiplabRaut

CMakeLists.txt

@@ -254,7 +254,7 @@ if (MSVC)
 endif(MSVC)
 
 string(TIMESTAMP TODAY "%Y%m%d")
-add_compile_definitions(AOCL_FFTW_VERSION="AOCL-FFTW 5.0.0 Build ${TODAY}")
+add_compile_definitions(AOCL_FFTW_VERSION="AOCL-FFTW 5.1.0 Build ${TODAY}")
 
 find_library (LIBM_LIBRARY NAMES m)
 if (LIBM_LIBRARY)

COPYRIGHT

@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2003, 2007-14 Matteo Frigo
  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
- * Copyright (C) 2019-2024, Advanced Micro Devices, Inc. All Rights Reserved.
+ * Copyright (C) 2019-2025, Advanced Micro Devices, Inc. All Rights Reserved.
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by

LICENSE

@@ -18385,7 +18385,7 @@ fi
 
 dateVar=`date +"%Y%m%d"`
 
-printf "%s\n" "#define AOCL_FFTW_VERSION \"AOCL-FFTW 5.0.0 Build $dateVar\"" >>confdefs.h
+printf "%s\n" "#define AOCL_FFTW_VERSION \"AOCL-FFTW 5.1.0 Build $dateVar\"" >>confdefs.h
 
 # Check whether --enable-amd-fast-planner was given.
 if test ${enable_amd_fast_planner+y}

NOTICES

@@ -790,7 +790,7 @@ AC_ARG_ENABLE(openmp, [AC_HELP_STRING([--enable-openmp],[use OpenMP directives f
 AC_ARG_ENABLE(threads, [AC_HELP_STRING([--enable-threads],[compile FFTW SMP threads library])], enable_threads=$enableval, enable_threads=no)
 dnl aocl version number of amd-fftw
 dateVar=`date +"%Y%m%d"`
-AC_DEFINE_UNQUOTED(AOCL_FFTW_VERSION,"AOCL-FFTW 5.0.0 Build $dateVar",[AOCL Version of AMD-FFTW])
+AC_DEFINE_UNQUOTED(AOCL_FFTW_VERSION,"AOCL-FFTW 5.1.0 Build $dateVar",[AOCL Version of AMD-FFTW])
 dnl amd optimization switch to enable AMD Fast Planner for AMD cpus --enable-amd-fast-planner
 AC_ARG_ENABLE(amd-fast-planner, [AC_HELP_STRING([--enable-amd-fast-planner],[enable AMD Fast Planner for a faster planning time on AMD cpus])], have_amd_fast_planner=$enableval, have_amd_fast_planner=no)
 dnl amd optimization switch to enable AMD Top N Planner for AMD cpus --enable-amd-top-n-planner

configure

@@ -71,7 +71,7 @@ static void transpose_chunks(int *sched, int n_pes, int my_pe,
            buf[1] = bufs[1];
 #endif
            int pe = sched[0], pe2, j=0;
-           MPI_Status send_status, recv_status;
+           MPI_Request send_request, recv_request;
 
 #ifdef AMD_MPI_TRANSPOSE_LOGS
            printf("TRANSPOSE-PAIRWISE: n_pes[%d], my_pe[%d], first_pe[%d]\n", n_pes, my_pe, pe);
@@ -105,13 +105,13 @@ static void transpose_chunks(int *sched, int n_pes, int my_pe,
 #ifdef AMD_MPI_TRANSPOSE_LOGS
                printf("OVERLAP REGION:(my_pe != pe)=> my_pe[%d], i[%d]:pe[%d], src_offset[%d], dst_offset[%d], size[%d]\n", my_pe, i, pe2, sbo[pe2], rbo[pe2], sbs[pe2]);
 #endif
-                   MPI_Isend(buf[j&0x1], (int) (sbs[pe]), FFTW_MPI_TYPE, pe, (my_pe * n_pes + pe) & 0xffff, comm, &send_status);
-                   MPI_Irecv(O + rbo[pe], (int) (rbs[pe]), FFTW_MPI_TYPE, pe, (pe * n_pes + my_pe) & 0xffff, comm, &recv_status);
+                   MPI_Isend(buf[j&0x1], (int) (sbs[pe]), FFTW_MPI_TYPE, pe, (my_pe * n_pes + pe) & 0xffff, comm, &send_request);
+                   MPI_Irecv(O + rbo[pe], (int) (rbs[pe]), FFTW_MPI_TYPE, pe, (pe * n_pes + my_pe) & 0xffff, comm, &recv_request);
                    memcpy(buf[(j+1)&0x1], O + sbo[pe2], sbs[pe2] * sizeof(R));
                    pe = pe2;
                    //barrier
-                   MPI_Wait(&send_status, MPI_STATUS_IGNORE);
-                   MPI_Wait(&recv_status, MPI_STATUS_IGNORE);
+                   MPI_Wait(&send_request, MPI_STATUS_IGNORE);
+                   MPI_Wait(&recv_request, MPI_STATUS_IGNORE);
                }
                j++;
            }