Allow building support for only a subset of variable types

Author: martin-frbg

CMakeLists.txt

@@ -29,8 +29,10 @@ option(NO_AFFINITY "Disable support for CPU affinity masks to avoid binding proc
 else()
 set(NO_AFFINITY 1)
 endif()
-option(CPP_THREAD_SAFETY_TEST "Run a massively parallel DGEMM test to confirm thread safety of the library (requires OpenMP and about 1.3GB of RAM)" OFF)
-option(CPP_THREAD_SAFETY_GEMV "Run a massively parallel DGEMV test to confirm thread safety of the library (requires OpenMP)" OFF)
+option(BUILD_SINGLE "Single precision" OFF)
+option(BUILD_DOUBLE "Double precision" OFF)
+option(BUILD_COMPLEX "Single precision" OFF)
+option(BUILD_COMPLEX16 "Single precision" OFF)
 
 # Add a prefix or suffix to all exported symbol names in the shared library.
 # Avoids conflicts with other BLAS libraries, especially when using
@@ -108,28 +110,33 @@ endif()
 
 set(FLOAT_TYPES "")
 if (BUILD_SINGLE)
-  message(STATUS "Building Single Precision")
-  list(APPEND FLOAT_TYPES "SINGLE") # defines nothing
+  message(STATUS "Building Songle Precision")
+  list(APPEND FLOAT_TYPES "SINGLE") 
+  #    set(CCOMMON_OPT "${CCOMMON_OPT} -DBUILD_SINGLE=1")
 endif ()
 
 if (BUILD_DOUBLE)
   message(STATUS "Building Double Precision")
-  list(APPEND FLOAT_TYPES "DOUBLE") # defines DOUBLE
+  list(APPEND FLOAT_TYPES "DOUBLE")
+  #set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DBUILD_DOUBLE=1")
 endif ()
 
 if (BUILD_COMPLEX)
   message(STATUS "Building Complex Precision")
-  list(APPEND FLOAT_TYPES "COMPLEX") # defines COMPLEX
-endif ()
+  list(APPEND FLOAT_TYPES "COMPLEX")
+  #set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DBUILD_COMPLEX=1")
+endif () 
 
 if (BUILD_COMPLEX16)
   message(STATUS "Building Double Complex Precision")
-  list(APPEND FLOAT_TYPES "ZCOMPLEX") # defines COMPLEX and DOUBLE
+  list(APPEND FLOAT_TYPES "ZCOMPLEX") 
+  #set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DBUILD_COMPLEX16=1")
 endif ()
 
 if (BUILD_HALF)
   message(STATUS "Building Half Precision")
-  list(APPEND FLOAT_TYPES "HALF") # defines nothing
+  list(APPEND FLOAT_TYPES "HALF")
+  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DBUILD_HALF")
 endif ()
 
 if (NOT DEFINED CORE OR "${CORE}" STREQUAL "UNKNOWN")
@@ -236,9 +243,6 @@ if (NOT MSVC AND NOT NOFORTRAN)
     add_subdirectory(ctest)
   endif()
   add_subdirectory(lapack-netlib/TESTING)
-	  if (CPP_THREAD_SAFETY_TEST OR CPP_THREAD_SAFETY_GEMV)
-	  	add_subdirectory(cpp_thread_test)
-    	  endif()
 endif()
 
 set_target_properties(${OpenBLAS_LIBNAME} PROPERTIES

Makefile

@@ -146,9 +146,6 @@ ifneq ($(NO_CBLAS), 1)
 ifeq ($(CPP_THREAD_SAFETY_TEST), 1)
 	$(MAKE) -C cpp_thread_test all
 endif
-ifeq ($(CPP_THREAD_SAFETY_GEMV), 1)
-	$(MAKE) -C cpp_thread_test dgemv_tester
-endif
 endif
 endif
 
@@ -304,6 +301,18 @@ else
 endif
 ifeq ($(BUILD_LAPACK_DEPRECATED), 1)
 	-@echo "BUILD_DEPRECATED      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
+endif
+ifeq ($(BUILD_SINGLE), 1)
+	-@echo "BUILD_SINGLE      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
+endif
+ifeq ($(BUILD_DOUBLE), 1)
+	-@echo "BUILD_DOUBLE      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
+endif
+ifeq ($(BUILD_COMPLEX), 1)
+	-@echo "BUILD_COMPLEX      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
+endif
+ifeq ($(BUILD_COMPLEX16), 1)
+	-@echo "BUILD_COMPLEX16      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
 endif
 	-@echo "LAPACKE_WITH_TMG      = 1" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@cat  make.inc >> $(NETLIB_LAPACK_DIR)/make.inc

Makefile.rule

@@ -272,33 +272,17 @@ COMMON_PROF = -pg
 # work at all.
 #
 # CPP_THREAD_SAFETY_TEST = 1
-#
-# use this to run only the less memory-hungry GEMV test
-# CPP_THREAD_SAFETY_GEMV = 1
 
 
 # If you want to enable the experimental BFLOAT16 support
 # BUILD_HALF = 1
-
-
-# Set the thread number threshold beyond which the job array for the threaded level3 BLAS
-# will be allocated on the heap rather than the stack. (This array alone requires 
-# NUM_THREADS*NUM_THREADS*128 bytes of memory so should not pose a problem at low cpu
-# counts, but obviously it is not the only item that ends up on the stack.
-# The default value of 32 ensures that the overall requirement is compatible
-# with the default 1MB stacksize imposed by having the Java VM loaded without use
-# of its -Xss parameter.
-# The value of 160 formerly used from about version 0.2.7 until 0.3.10 is easily compatible
-# with the common Linux stacksize of 8MB but will cause crashes with unwary use of the java
-# VM e.g. in Octave or with the java-based libhdfs in numpy or scipy code
-# BLAS3_MEM_ALLOC_THRESHOLD = 160
-
-
-
-# the below is not yet configurable, use cmake if you need to build only select types
-BUILD_SINGLE = 1
-BUILD_DOUBLE = 1
-BUILD_COMPLEX = 1
-BUILD_COMPLEX16 = 1
+#
+# Select if you need to build only select types
+# BUILD_SINGLE = 1
+# BUILD_DOUBLE = 1
+# BUILD_COMPLEX = 1
+# BUILD_COMPLEX16 = 1
+#
+#
 #  End of user configuration
 #

Makefile.tail

@@ -11,8 +11,8 @@ COMMONOBJS_P = $(COMMONOBJS:.$(SUFFIX)=.$(PSUFFIX))
 
 HPLOBJS_P   = $(HPLOBJS:.$(SUFFIX)=.$(PSUFFIX))
 
-BLASOBJS    = $(SHEXTOBJS) $(SHBLASOBJS)  $(SBLASOBJS)   $(DBLASOBJS)   $(CBLASOBJS)   $(ZBLASOBJS)
-BLASOBJS_P  = $(SHEXTOBJS_P) $(SHBLASOBJS_P) $(SBLASOBJS_P) $(DBLASOBJS_P) $(CBLASOBJS_P) $(ZBLASOBJS_P)
+BLASOBJS    = $(SHEXTOBJS) $(SHBLASOBJS)  $(SBLASOBJS)   $(DBLASOBJS)   $(CBLASOBJS)   $(ZBLASOBJS) $(CBAUXOBJS)
+BLASOBJS_P  = $(SHEXTOBJS_P) $(SHBLASOBJS_P) $(SBLASOBJS_P) $(DBLASOBJS_P) $(CBLASOBJS_P) $(ZBLASOBJS_P) $(CBAUXOBJS_P)
 
 ifdef EXPRECISION
 BLASOBJS   += $(QBLASOBJS)   $(XBLASOBJS)