Merge 3.2 release branch into amd-fftw

Change-Id: If976175f9a64305ff59bb973fb44568c01802ae2
amd · Jul 8, 2022 · e04beee · e04beee
2 parents 2b0bbb5 + 7f36218
commit e04beee
Show file tree

Hide file tree

Showing 97 changed files with 2,321 additions and 658 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -122,6 +122,10 @@ if (ENABLE_AMD_APP_OPT)
   endif ()
 endif ()
 
+if(ENABLE_AMD_DYNAMIC_DISPATCHER)
+	message (FATAL_ERROR "AMD Dynamic Dispatcher feature is not available on windows.")
+endif()
+
 option (DISABLE_FORTRAN "Disable Fortran wrapper routines" OFF)
 
 if (CMAKE_C_COMPILER_ID MATCHES MSVC OR CMAKE_C_COMPILER_ID MATCHES Clang)
@@ -243,7 +247,7 @@ if (MSVC)
   add_definitions(-D_CRT_SECURE_NO_WARNINGS)
 endif(MSVC)
 
-add_compile_definitions(AOCL_FFTW_VERSION="AOCL FFTW 3.1")
+add_compile_definitions(AOCL_FFTW_VERSION="AOCL-FFTW-3.2")
 
 find_library (LIBM_LIBRARY NAMES m)
 if (LIBM_LIBRARY)
@@ -338,6 +342,7 @@ else()
   if (ENABLE_SSE)
     foreach (FLAG "-msse" "/arch:SSE")
       unset (HAVE_SSE CACHE)
+      unset (HAVE_SSE)
       check_c_compiler_flag (${FLAG} HAVE_SSE)
       if (HAVE_SSE)
         set (SSE_FLAG ${FLAG})
@@ -349,6 +354,7 @@ else()
   if (ENABLE_SSE2)
     foreach (FLAG "-msse2" "/arch:SSE2")
       unset (HAVE_SSE2 CACHE)
+      unset (HAVE_SSE2)
       check_c_compiler_flag (${FLAG} HAVE_SSE2)
       if (HAVE_SSE2)
         set (SSE2_FLAG ${FLAG})
@@ -360,6 +366,7 @@ else()
   if (ENABLE_AVX)
     foreach (FLAG "-mavx" "/arch:AVX")
       unset (HAVE_AVX CACHE)
+      unset (HAVE_AVX)
       check_c_compiler_flag (${FLAG} HAVE_AVX)
       if (HAVE_AVX)
         set (AVX_FLAG ${FLAG})
@@ -371,6 +378,7 @@ else()
   if (ENABLE_AVX2)
     foreach (FLAG "-mavx2" "/arch:AVX2")
       unset (HAVE_AVX2 CACHE)
+      unset (HAVE_AVX2)
       check_c_compiler_flag (${FLAG} HAVE_AVX2)
       if (HAVE_AVX2)
         set (AVX2_FLAG ${FLAG})
@@ -383,6 +391,7 @@ else()
   if (ENABLE_AVX2)
     foreach (FLAG "-mfma" "/arch:FMA")
       unset (HAVE_FMA CACHE)
+      unset (HAVE_FMA)
       check_c_compiler_flag (${FLAG} HAVE_FMA)
       if (HAVE_FMA)
         set (FMA_FLAG ${FLAG})
@@ -486,11 +495,7 @@ if (HAVE_AVX2)
   list (APPEND SOURCEFILES ${fftw_dft_simd_avx2_SOURCE} ${fftw_rdft_simd_avx2_SOURCE})
 endif ()
 
-if (HAVE_MPI)
-  list (APPEND SOURCEFILES ${fftw_mpi_SOURCE})
-endif ()
-
-set (FFTW_VERSION 3.3.8)
+set (FFTW_VERSION 3.3.10)
 
 set (PREC_SUFFIX)
 if (ENABLE_FLOAT)
@@ -539,7 +544,7 @@ if (ENABLE_AMD_TOP_N_planner)
 
   endif ()
 
-set (fftw3_lib fftw3${PREC_SUFFIX})
+set (fftw3_lib libfftw3${PREC_SUFFIX})
 
 configure_file (cmake.config.h.in config.h @ONLY)
 include_directories (${CMAKE_CURRENT_BINARY_DIR})
@@ -549,6 +554,10 @@ if (BUILD_SHARED_LIBS)
   set(CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS ON)
 endif ()
 
+set(lib_name ${fftw3_lib})
+add_library (${lib_name} ${SOURCEFILES})
+target_include_directories (${lib_name} INTERFACE $<INSTALL_INTERFACE:include>)
+
 if (Threads_FOUND)
   if (WITH_COMBINED_THREADS)
     set (lib_name ${fftw3_lib})
@@ -561,18 +570,26 @@ if (Threads_FOUND)
     target_include_directories (${lib_name} INTERFACE $<INSTALL_INTERFACE:include>)
     target_link_libraries (${lib_name} ${CMAKE_THREAD_LIBS_INIT})
   endif ()
-elseif (OPENMP_FOUND)
+endif ()
+
+if (OPENMP_FOUND)
   set (lib_name ${fftw3_lib}_omp)
   add_library (${lib_name} ${fftw_omp_SOURCE} ${SOURCEFILES})
   target_include_directories (${lib_name} INTERFACE $<INSTALL_INTERFACE:include>)
   target_link_libraries (${lib_name} ${CMAKE_THREAD_LIBS_INIT})
-  target_compile_options (${lib_name} PRIVATE ${OpenMP_C_FLAGS})
-else ()
-  set(lib_name ${fftw3_lib})
-  add_library (${lib_name} ${SOURCEFILES})
+  target_compile_options (${lib_name} PRIVATE ${OpenMP_C_FLAGS})  
+endif ()
+
+if (HAVE_MPI)
+  list (APPEND SOURCEFILES ${fftw_mpi_SOURCE})
+  set (lib_name ${fftw3_lib}_mpi)
+  add_library (${lib_name} ${fftw_omp_SOURCE} ${SOURCEFILES})
   target_include_directories (${lib_name} INTERFACE $<INSTALL_INTERFACE:include>)
+  target_link_libraries (${lib_name} ${CMAKE_THREAD_LIBS_INIT})
+  target_compile_options (${lib_name} PRIVATE ${OpenMP_C_FLAGS}) 
 endif ()
 
+
 target_include_directories(${lib_name} PRIVATE ${CMAKE_SOURCE_DIR}/api)
 
 if (CMAKE_C_COMPILER_ID MATCHES MSVC)
@@ -616,18 +633,58 @@ install(TARGETS ${lib_name}
           ARCHIVE DESTINATION ${CMAKE_INSTALL_LIBDIR})
 
 install (FILES api/fftw3.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
-if (EXISTS ${CMAKE_SOURCE_DIR}/api/fftw3.f)
+
+file(COPY api/fftw3.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+if (EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/api/fftw3.f)
   install (FILES api/fftw3.f api/fftw3l.f03 api/fftw3q.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+  file(COPY api/fftw3.f api/fftw3l.f03 api/fftw3q.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 endif ()
-if (EXISTS ${CMAKE_SOURCE_DIR}/api/fftw3.f03.in)
+if (EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/api/fftw3.f03.in)
   file (READ api/fftw3.f03.in FFTW3_F03_IN OFFSET 42)
   file (WRITE ${CMAKE_CURRENT_BINARY_DIR}/fftw3.f03 "! Generated automatically.  DO NOT EDIT!\n\n")
   file (APPEND ${CMAKE_CURRENT_BINARY_DIR}/fftw3.f03 "  integer, parameter :: C_FFTW_R2R_KIND = ${C_FFTW_R2R_KIND}\n\n")
   file (APPEND ${CMAKE_CURRENT_BINARY_DIR}/fftw3.f03 "${FFTW3_F03_IN}")
   install (FILES ${CMAKE_CURRENT_BINARY_DIR}/fftw3.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+  file(COPY ${CMAKE_CURRENT_BINARY_DIR}/fftw3.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+endif ()
+
+
+if (HAVE_MPI)
+	install (FILES mpi/fftw3-mpi.h DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+	if (EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/mpi/fftw3-mpi.f03.in)
+	  file (READ mpi/fftw3-mpi.f03.in FFTW3-MPI_F03_IN OFFSET 42)
+	  file (WRITE ${CMAKE_CURRENT_BINARY_DIR}/fftw3-mpi.f03 "! Generated automatically.  DO NOT EDIT!\n\n")
+	  file (APPEND ${CMAKE_CURRENT_BINARY_DIR}/fftw3-mpi.f03 "${FFTW3-MPI_F03_IN}")
+	  install (FILES ${CMAKE_CURRENT_BINARY_DIR}/fftw3-mpi.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+	endif ()
+	if (EXISTS ${CMAKE_CURRENT_SOURCE_DIR}/mpi/fftw3l-mpi.f03.in)
+	  file (READ mpi/fftw3l-mpi.f03.in FFTW3L-MPI_F03_IN OFFSET 42)
+	  file (WRITE ${CMAKE_CURRENT_BINARY_DIR}/fftw3l-mpi.f03 "! Generated automatically.  DO NOT EDIT!\n\n")
+	  file (APPEND ${CMAKE_CURRENT_BINARY_DIR}/fftw3l-mpi.f03 "${FFTW3L-MPI_F03_IN}")
+	  install (FILES ${CMAKE_CURRENT_BINARY_DIR}/fftw3l-mpi.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
+	endif ()
+	file(COPY mpi/fftw3-mpi.h ${CMAKE_CURRENT_BINARY_DIR}/fftw3-mpi.f03 ${CMAKE_CURRENT_BINARY_DIR}/fftw3l-mpi.f03 DESTINATION ${CMAKE_INSTALL_INCLUDEDIR})
 endif ()
 
 add_library (libbench2 STATIC ${fftw_libbench2_SOURCE})
+add_library (libtestbench STATIC tests/bench.c tests/hook.c tests/fftw-bench.c tests/fftw-bench.h)
+
+if(Threads_FOUND)
+        set(CMAKE_C_COMPILER_CFLAGS ${PTHREAD_CFLAGS})
+        if(NOT COMBINED_THREADS)
+                set(LIBFFTWTHREADS = libfftw3${PREC_SUFFIX}_threads)
+        endif()
+elseif(OPENMP_FOUND)
+        set(CMAKE_C_COMPILER_CFLAGS ${OpenMP_C_FLAGS})
+        set(LIBFFTWTHREADS libfftw3${PREC_SUFFIX}_omp)
+endif()
+
+set(fftw${PREC_SUFFIX}-wisdom_src tools/fftw-wisdom.c)
+
+add_executable(fftw${PREC_SUFFIX}-wisdom ${fftw${PREC_SUFFIX}-wisdom_src})
+target_link_libraries(fftw${PREC_SUFFIX}-wisdom libbench2  ${LIBFFTWTHREADS} ${lib_name} libtestbench)
+install(TARGETS fftw${PREC_SUFFIX}-wisdom ${INSTALL_TARGETS_DEFAULT_ARGS} DESTINATION ${CMAKE_INSTALL_BINDIR})
+
 
 if (BUILD_TESTS)
   add_executable (bench tests/bench.c tests/hook.c tests/fftw-bench.c)
@@ -661,9 +718,9 @@ set (exec_prefix ${CMAKE_INSTALL_PREFIX})
 set (libdir ${CMAKE_INSTALL_FULL_LIBDIR})
 set (includedir ${CMAKE_INSTALL_FULL_INCLUDEDIR})
 set (VERSION ${FFTW_VERSION})
-configure_file (fftw.pc.in fftw${PREC_SUFFIX}.pc @ONLY)
+configure_file (fftw.pc.in fftw3${PREC_SUFFIX}.pc @ONLY)
 install (FILES
-          ${CMAKE_CURRENT_BINARY_DIR}/fftw${PREC_SUFFIX}.pc
+          ${CMAKE_CURRENT_BINARY_DIR}/fftw3${PREC_SUFFIX}.pc
          DESTINATION ${CMAKE_INSTALL_LIBDIR}/pkgconfig
          COMPONENT Development)
 

diff --git a/COPYRIGHT b/COPYRIGHT
@@ -1,7 +1,7 @@
 /*
  * Copyright (c) 2003, 2007-14 Matteo Frigo
  * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology
- * Copyright (C) 2019-2021, Advanced Micro Devices, Inc. All Rights Reserved.
+ * Copyright (C) 2019-2022, 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

diff --git a/NEWS b/NEWS
@@ -1,3 +1,47 @@
+FFTW 3.3.10:
+
+* Fix bug that would cause 2-way SIMD (notably SSE2 in double precision)
+  to attempt unaligned accesses in certain obscure cases, causing
+  segfaults.
+
+  The following test triggers the bug (SSE2, double precision):
+
+    ./tests/bench -oexhaustive r4*2:5:3
+
+  This test computes a pair of length-4 real->complex transforms where
+  the second input is 5 real numbers away from the first input.  That
+  is, there is a gap of one real number between the first and second
+  input array.  The -oexhaustive level allow FFTW to attempt to
+  compute this transform by reducing it to a pair of complex
+  transforms of length 2, but now the second input is not aligned to a
+  complex-number boundary.  The fact that 5 is odd is the problem.
+
+  The bug cannot occur in complex->complex transforms because the
+  complex interface accepts strides in units of complex numbers, so
+  strides are aligned by construction.
+
+  This bug has been around at least since fftw-3.1.2 (July 2006), and
+  probably since fftw-3.0 (2003).
+
+FFTW 3.3.9:
+
+* New API fftw_planner_nthreads() returns the number of threads
+  currently being used by the planner.
+
+* Fix incorrect math in 128-bit generic SIMD
+
+* Fix wisdom for avx512.
+
+  The avx512 alignment requirement was set to 64 bytes, but this is
+  wrong.  Alignment requirements are a property of the platform (e.g.,
+  x86) and not of the instruction set (e.g., AVX).  Among other
+  things, this broke wisdom with avx512.
+
+  Note that avx512 support is still experimental because the FFTW
+  authors have no avx512 hardware available for testing.
+
+* fftw_threads_set_callback function to change the threading backend at runtime.
+
 FFTW 3.3.8:
 
 * Fixed AVX, AVX2 for gcc-8.
@@ -33,7 +77,7 @@ FFTW 3.3.6-pl1:
 
 * Bugfix: FFTW 3.3.6 had the wrong libtool version number, and generated
   shared libraries of the form libfftw3.so.2.6.6 instead of
-  libfftw3.so.3.*.  
+  libfftw3.so.3.*.
 
 FFTW 3.3.6:
 
@@ -58,7 +102,7 @@ FFTW 3.3.5:
     This code is expected to work but the FFTW maintainers do not have
     hardware to test it.
   - generic SIMD support using gcc vector intrinsics
-* Add fftw_make_planner_thread_safe() API    
+* Add fftw_make_planner_thread_safe() API
 * fix #18 (disable float128 for CUDACC)
 * fix #19: missing Fortran interface for fftwq_alloc_real
 * fix #21 (don't use float128 on Portland compilers, which pretend to be gcc)
@@ -107,7 +151,7 @@ FFTW 3.3.2
 
 * Added stack-alignment hack necessary for gcc on Windows/i386.  We
   will regret this in ten years (see previous change).
-  
+
 * Fix incompatibility with Intel icc which pretends to be gcc
   but does not support quad precision.
 
@@ -118,7 +162,7 @@ FFTW 3.3.2
 FFTW 3.3.1
 
 * Changes since 3.3.1-beta1:
-  
+
   - Reduced planning time in estimate mode for sizes with large
     prime factors.
 
@@ -190,7 +234,7 @@ FFTW 3.3
   fftw_import_wisdom_from_filename that export/import wisdom
   to a file, which don't require you to open/close the file yourself.
 
-* New function fftw_cost to return FFTW's internal cost metric for 
+* New function fftw_cost to return FFTW's internal cost metric for
   a given plan; thanks to Rhys Ulerich and Nathanael Schaeffer for the
   suggestion.
 
@@ -248,7 +292,7 @@ FFTW 3.2.1
 
 * FAQ was accidentally omitted from 3.2 tarball.
 
-* Remove some extraneous (harmless) files accidentally included in 
+* Remove some extraneous (harmless) files accidentally included in
   a subdirectory of the 3.2 tarball.
 
 FFTW 3.2
@@ -417,7 +461,7 @@ FFTW 3.1
 
 * Various documentation clarifications.
 
-* 64-bit clean.  (Fixes a bug affecting the split guru planner on 
+* 64-bit clean.  (Fixes a bug affecting the split guru planner on
   64-bit machines, reported by David Necas.)
 
 * Fixed Debian bug #259612: inadvertent use of SSE instructions on

diff --git a/README b/README
@@ -0,0 +1,21 @@
+*DO NOT CHECK OUT THESE FILES FROM GITHUB UNLESS YOU KNOW WHAT YOU ARE
+DOING.*  (See below.)
+
+This is the git repository for the FFTW library for computing Fourier
+transforms (version 3.x), maintained by the FFTW authors.
+
+Unlike most other programs, most of the FFTW source code (in C) is
+generated automatically.  This repository contains the *generator* and
+it does not contain the *generated code*.  *YOU WILL BE UNABLE TO
+COMPILE CODE FROM THIS REPOSITORY* unless you have special tools and
+know what you are doing.   In particular, do not expect things to
+work by simply executing `configure; make` or `cmake`.
+
+Most users should ignore this repository, and should instead download
+official tarballs from http://fftw.org/, which contain the generated
+code, do not require any special tools or knowledge, and can be
+compiled on any system with a C compiler.
+
+Advanced users and FFTW maintainers may obtain code from github and
+run the generation process themselves.  See [README](README) for
+details.