Merge branch 'develop'

Author: xianyi

Changelog.txt

@@ -1,4 +1,15 @@
 OpenBLAS ChangeLog
+====================================================================
+Version 0.2.9
+10-Jun-2014
+common:
+	* Improved the result for LAPACK testing. (#372)
+	* Installed DLL to prefix/bin instead of prefix/lib. (#366)
+	* Build import library on Windows.(#374)
+x86/x86-64:
+	* To improve LAPACK testing, we fallback some kernels. (#372)
+	https://github.com/xianyi/OpenBLAS/wiki/Fixed-optimized-kernels-To-do-List
+
 ====================================================================
 Version 0.2.9.rc2
 06-Mar-2014
@@ -51,7 +62,7 @@ common:
           parallelization model is used by OpenBLAS. (Thank grisuthedragon)
 	* Detect LLVM/Clang compiler. The default compiler is Clang on Mac OS X.
         * Change LIBSUFFIX from .lib to .a on windows.
-	* A walk round for dtrti_U single thread bug. Replace it with LAPACK codes. (#191)
+	* A work-around for dtrti_U single thread bug. Replace it with LAPACK codes. (#191)
 
 x86/x86-64:
 	* Optimize c/zgemm, trsm, dgemv_n, ddot, daxpy, dcopy on 
@@ -284,7 +295,7 @@ x86/x86_64:
 	* Fixed #28 a wrong result of dsdot on x86_64.
 	* Fixed #32 a SEGFAULT bug of zdotc with gcc-4.6.
 	* Fixed #33 ztrmm bug on Nehalem.
-	* Walk round #27 the low performance axpy issue with small imput size & multithreads.
+	* Work-around #27 the low performance axpy issue with small imput size & multithreads.
 
 MIPS64:
 	* Fixed #28 a wrong result of dsdot on Loongson3A/MIPS64. 
@@ -308,7 +319,7 @@ common:
 
 x86/x86_64:
 	* On x86 32bits, fixed a bug in zdot_sse2.S line 191. This would casue 
-	  zdotu & zdotc failures.Instead,Walk around it. (Refs issue #8 #9 on github)
+	  zdotu & zdotc failures. Instead, work-around it. (Refs issue #8 #9 on github)
 	* Modified ?axpy functions to return same netlib BLAS results 
 	  when incx==0 or incy==0 (Refs issue #7 on github)
 	* Modified ?swap functions to return same netlib BLAS results 

Makefile

@@ -57,7 +57,7 @@ endif
 
 ifeq ($(USE_OPENMP), 1)
 	@echo
-	@echo " Use OpenMP in the multithreading. Becasue of ignoring OPENBLAS_NUM_THREADS and GOTO_NUM_THREADS flags, "
+	@echo " Use OpenMP in the multithreading. Because of ignoring OPENBLAS_NUM_THREADS and GOTO_NUM_THREADS flags, "
 	@echo " you should use OMP_NUM_THREADS environment variable to control the number of threads."
 	@echo
 endif
@@ -128,6 +128,11 @@ ifeq ($(CORE), UNKOWN)
 endif
 ifeq ($(NOFORTRAN), 1)
 	$(error OpenBLAS: Detecting fortran compiler failed. Please install fortran compiler, e.g. gfortran, ifort, openf90.)
+endif
+ifeq ($(NO_STATIC), 1)
+ifeq ($(NO_SHARED), 1)
+	$(error OpenBLAS: neither static nor shared are enabled.)
+endif
 endif
 	@-ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
 	@for d in $(SUBDIRS) ; \
@@ -207,6 +212,7 @@ else
 netlib : lapack_prebuild
 ifndef NOFORTRAN
 	@$(MAKE) -C $(NETLIB_LAPACK_DIR) lapacklib
+	@$(MAKE) -C $(NETLIB_LAPACK_DIR) tmglib
 endif
 ifndef NO_LAPACKE
 	@$(MAKE) -C $(NETLIB_LAPACK_DIR) lapackelib
@@ -230,43 +236,19 @@ ifndef NOFORTRAN
 	-@echo "ARCHFLAGS   = -ru" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "RANLIB      = $(RANLIB)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "LAPACKLIB   = ../$(LIBNAME)" >> $(NETLIB_LAPACK_DIR)/make.inc
+	-@echo "TMGLIB      = ../$(LIBNAME)" >> $(NETLIB_LAPACK_DIR)/make.inc
+	-@echo "BLASLIB     = ../../../$(LIBNAME)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "LAPACKELIB  = ../$(LIBNAME)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "LAPACKLIB_P = ../$(LIBNAME_P)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "SUFFIX      = $(SUFFIX)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "PSUFFIX     = $(PSUFFIX)" >> $(NETLIB_LAPACK_DIR)/make.inc
 	-@echo "CEXTRALIB   = $(EXTRALIB)" >> $(NETLIB_LAPACK_DIR)/make.inc
-	-@cat  make.inc >> $(NETLIB_LAPACK_DIR)/make.inc
-endif
-
-lapack-3.4.2 : lapack-3.4.2.tgz
-ifndef NOFORTRAN
-ifndef NO_LAPACK
-	@if test `$(MD5SUM) $< | $(AWK) '{print $$1}'` = 61bf1a8a4469d4bdb7604f5897179478; then \
-		echo $(TAR) zxf $< ;\
-		$(TAR) zxf $< && (cd $(NETLIB_LAPACK_DIR); $(PATCH) -p1 < ../patch.for_lapack-3.4.2) ;\
-		rm -f $(NETLIB_LAPACK_DIR)/lapacke/make.inc ;\
-	else \
-		rm -rf $(NETLIB_LAPACK_DIR) ;\
-		echo "	Cannot download lapack-3.4.2.tgz or the MD5 check sum is wrong (Please use orignal)."; \
-		exit 1; \
-	fi
-endif
-endif
-
-LAPACK_URL=http://www.netlib.org/lapack/lapack-3.4.2.tgz
-
-lapack-3.4.2.tgz :
-ifndef NOFORTRAN
-#http://stackoverflow.com/questions/7656425/makefile-ifeq-logical-or
-ifeq ($(OSNAME), $(filter $(OSNAME),Darwin NetBSD))
-	curl -O $(LAPACK_URL);
-else
-ifeq ($(OSNAME), FreeBSD)
-	fetch $(LAPACK_URL);
+ifeq ($(F_COMPILER), GFORTRAN)
+	-@echo "TIMER       = INT_ETIME" >> $(NETLIB_LAPACK_DIR)/make.inc
 else
-	wget -O $@ $(LAPACK_URL);
-endif
+	-@echo "TIMER       = NONE" >> $(NETLIB_LAPACK_DIR)/make.inc
 endif
+	-@cat  make.inc >> $(NETLIB_LAPACK_DIR)/make.inc
 endif
 
 large.tgz : 
@@ -287,17 +269,15 @@ lapack-timing : large.tgz timing.tgz
 ifndef NOFORTRAN
 	(cd $(NETLIB_LAPACK_DIR); $(TAR) zxf ../timing.tgz TIMING)
 	(cd $(NETLIB_LAPACK_DIR)/TIMING; $(TAR) zxf ../../large.tgz )
-	make -C $(NETLIB_LAPACK_DIR) tmglib
 	make -C $(NETLIB_LAPACK_DIR)/TIMING
 endif
 
 
 lapack-test :
-	$(MAKE) -C $(NETLIB_LAPACK_DIR) tmglib
-	$(MAKE) -C $(NETLIB_LAPACK_DIR)/TESTING xeigtstc xeigtstd xeigtsts xeigtstz xlintstc xlintstd xlintstds xlintsts xlintstz xlintstzc
-	@rm	-f $(NETLIB_LAPACK_DIR)/TESTING/*.out
-	$(MAKE) -j 1 -C $(NETLIB_LAPACK_DIR)/TESTING
-	$(GREP) failed $(NETLIB_LAPACK_DIR)/TESTING/*.out
+	(cd $(NETLIB_LAPACK_DIR)/TESTING && rm -f x* *.out)
+	make -j 1 -C $(NETLIB_LAPACK_DIR)/TESTING xeigtstc  xeigtstd  xeigtsts  xeigtstz  xlintstc  xlintstd  xlintstds  xlintstrfd  xlintstrfz  xlintsts  xlintstz  xlintstzc xlintstrfs xlintstrfc
+	(cd $(NETLIB_LAPACK_DIR); ./lapack_testing.py -r )
+
 
 dummy :
 
@@ -323,4 +303,5 @@ endif
 	@$(MAKE) -C $(NETLIB_LAPACK_DIR) clean
 	@rm -f $(NETLIB_LAPACK_DIR)/make.inc $(NETLIB_LAPACK_DIR)/lapacke/include/lapacke_mangling.h
 	@rm -f *.grd Makefile.conf_last config_last.h
+	@(cd $(NETLIB_LAPACK_DIR)/TESTING && rm -f x* *.out testing_results.txt)
 	@echo Done.

Makefile.arm

@@ -10,3 +10,9 @@ FCOMMON_OPT += -marm -mfpu=vfp -mfloat-abi=hard  -march=armv6
 endif
 
 
+ifeq ($(CORE), ARMV5)
+CCOMMON_OPT += -marm -mfpu=vfp -mfloat-abi=hard  -march=armv6
+FCOMMON_OPT += -marm -mfpu=vfp -mfloat-abi=hard  -march=armv6
+endif
+
+

Makefile.install

@@ -7,6 +7,7 @@ PREFIX ?= /opt/OpenBLAS
 
 OPENBLAS_INCLUDE_DIR := $(PREFIX)/include
 OPENBLAS_LIBRARY_DIR := $(PREFIX)/lib
+OPENBLAS_BINARY_DIR := $(PREFIX)/bin
 OPENBLAS_BUILD_DIR := $(CURDIR)
 
 .PHONY : install
@@ -19,6 +20,7 @@ install : 	lib.grd
 	@-mkdir -p $(DESTDIR)$(PREFIX)
 	@-mkdir -p $(DESTDIR)$(OPENBLAS_INCLUDE_DIR)
 	@-mkdir -p $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
+	@-mkdir -p $(DESTDIR)$(OPENBLAS_BINARY_DIR)
 	@echo Generating openblas_config.h in $(DESTDIR)$(OPENBLAS_INCLUDE_DIR)
 #for inc 
 	@echo \#ifndef OPENBLAS_CONFIG_H > $(DESTDIR)$(OPENBLAS_INCLUDE_DIR)/openblas_config.h
@@ -49,10 +51,12 @@ ifndef NO_LAPACKE
 endif
 
 #for install static library 
+ifndef NO_STATIC
 	@echo Copying the static library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 	@install -pm644 $(LIBNAME) $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
 	@cd $(DESTDIR)$(OPENBLAS_LIBRARY_DIR) ; \
 	ln -fs $(LIBNAME) $(LIBPREFIX).$(LIBSUFFIX)
+endif
 #for install shared library 
 ifndef NO_SHARED
 	@echo Copying the shared library to $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)
@@ -78,10 +82,11 @@ ifeq ($(OSNAME), Darwin)
 	@-ln -fs $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)/$(LIBDYNNAME) $(DESTDIR)$(OPENBLAS_LIBRARY_DIR)/$(LIBPREFIX).dylib
 endif
 ifeq ($(OSNAME), WINNT)
-	@-cp $(LIBDLLNAME) $(OPENBLAS_LIBRARY_DIR)
+	@-cp $(LIBDLLNAME) $(OPENBLAS_BINARY_DIR)
+	@-cp $(LIBPREFIX).lib $(OPENBLAS_LIBRARY_DIR)
 endif
 ifeq ($(OSNAME), CYGWIN_NT)
-	@-cp $(LIBDLLNAME) $(OPENBLAS_LIBRARY_DIR)
+	@-cp $(LIBDLLNAME) $(OPENBLAS_BINARY_DIR)
 endif
 endif
 

Makefile.rule

@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.2.9.rc2
+VERSION = 0.2.9
 
 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library 
@@ -48,6 +48,9 @@ VERSION = 0.2.9.rc2
 # automatically detected by the the script.
 # NUM_THREADS = 24
 
+# if you don't need to install the static library, please comment it in.
+# NO_STATIC = 1
+
 # if you don't need generate the shared library, please comment it in.
 # NO_SHARED = 1
 
@@ -76,10 +79,10 @@ VERSION = 0.2.9.rc2
 # Unfortunately most of kernel won't give us high quality buffer.
 # BLAS tries to find the best region before entering main function,
 # but it will consume time. If you don't like it, you can disable one.
-# NO_WARMUP = 1
+NO_WARMUP = 1
 
 # If you want to disable CPU/Memory affinity on Linux.
-# NO_AFFINITY = 1
+NO_AFFINITY = 1
 
 # Don't use AVX kernel on Sandy Bridge. It is compatible with old compilers 
 # and OS. However, the performance is low.
@@ -129,6 +132,9 @@ VERSION = 0.2.9.rc2
 # The default -O2 is enough.
 # COMMON_OPT = -O2
 
+# gfortran option for LAPACK
+FCOMMON_OPT = -frecursive
+
 # Profiling flags
 COMMON_PROF = -pg
 

Makefile.system

@@ -158,6 +158,7 @@ endif
 
 ifeq ($(OSNAME), Linux)
 EXTRALIB	+= -lm
+NO_EXPRECISION = 1
 endif
 
 ifeq ($(OSNAME), AIX)
@@ -846,19 +847,6 @@ ifeq ($(DEBUG), 1)
 COMMON_OPT += -g
 endif
 
-ifndef COMMON_OPT
-ifeq ($(ARCH), arm)
-COMMON_OPT = -O3
-endif
-endif
-
-ifndef COMMON_OPT
-ifeq ($(ARCH), arm64)
-COMMON_OPT = -O3
-endif
-endif
-
-
 ifndef COMMON_OPT
 COMMON_OPT = -O2
 endif

common.h

@@ -310,10 +310,17 @@ typedef int blasint;
 #define YIELDING	SwitchToThread()
 #endif
 
-#if defined(ARMV7) || defined(ARMV6) || defined(ARMV8)
+#if defined(ARMV7) || defined(ARMV6) || defined(ARMV8) || defined(ARMV5)
 #define YIELDING        asm volatile ("nop;nop;nop;nop;nop;nop;nop;nop; \n");
 #endif
 
+#ifdef BULLDOZER
+#ifndef YIELDING
+#define YIELDING        __asm__ __volatile__ ("nop;nop;nop;nop;nop;nop;nop;nop;\n");
+#endif
+#endif
+
+
 #ifdef PILEDRIVER
 #ifndef YIELDING
 #define YIELDING        __asm__ __volatile__ ("nop;nop;nop;nop;nop;nop;nop;nop;\n");

ctest.c

@@ -125,7 +125,7 @@ ARCH_IA64
 BINARY_64
 #endif
 
-#if defined(__ARM_ARCH) || defined(__ARM_ARCH_7A__)
+#if defined(__ARM_ARCH) || defined(__ARM_ARCH_7A__) || defined(__arm__)
 ARCH_ARM
 #endif
 

driver/others/Makefile

@@ -3,7 +3,7 @@ include ../../Makefile.system
 
 COMMONOBJS	 = memory.$(SUFFIX) xerbla.$(SUFFIX) c_abs.$(SUFFIX) z_abs.$(SUFFIX) openblas_set_num_threads.$(SUFFIX) openblas_get_config.$(SUFFIX) openblas_get_parallel.$(SUFFIX) openblas_error_handle.$(SUFFIX) 
 
-COMMONOBJS	+= slamch.$(SUFFIX) slamc3.$(SUFFIX) dlamch.$(SUFFIX)  dlamc3.$(SUFFIX)
+#COMMONOBJS	+= slamch.$(SUFFIX) slamc3.$(SUFFIX) dlamch.$(SUFFIX)  dlamc3.$(SUFFIX)
 
 ifdef SMP
 COMMONOBJS	+= blas_server.$(SUFFIX) divtable.$(SUFFIX) blasL1thread.$(SUFFIX) 

exports/Makefile

@@ -75,9 +75,6 @@ zip : dll
 	zip $(LIBZIPNAME) $(LIBDLLNAME) $(LIBNAME)
 
 dll  : ../$(LIBDLLNAME)
-#libgoto2.dll
-
-dll2 : libgoto2_shared.dll
 
 # On Windows, we only generate a DLL without a version suffix.  This is because
 # applications which link against the dynamic library reference a fixed DLL name
@@ -86,44 +83,27 @@ dll2 : libgoto2_shared.dll
 # For more details see: https://github.com/xianyi/OpenBLAS/issues/127.
 ../$(LIBDLLNAME) : ../$(LIBNAME) libopenblas.def dllinit.$(SUFFIX)
 	$(RANLIB) ../$(LIBNAME)
-ifeq ($(BINARY32), 1)
-	$(DLLWRAP) -o ../$(LIBDLLNAME) --def libopenblas.def \
-	--entry _dllinit@12 -s dllinit.$(SUFFIX) --dllname $(@F) ../$(LIBNAME) $(EXTRALIB)
-	-lib /machine:i386 /def:libopenblas.def
-else
-	$(DLLWRAP) -o ../$(LIBDLLNAME) --def libopenblas.def \
-	--entry $(FU)dllinit    -s dllinit.$(SUFFIX) --dllname $(@F) ../$(LIBNAME) $(EXTRALIB)
-	-lib /machine:X64 /def:libopenblas.def 
-endif
-
-libgoto2_shared.dll : ../$(LIBNAME) libgoto2_shared.def
-	$(CC) $(CFLAGS) $(LDFLAGS) libgoto2_shared.def -shared -o $(@F) \
-	-Wl,--whole-archive ../$(LIBNAME) -Wl,--no-whole-archive \
-	-Wl,--out-implib,libgoto2_shared.lib $(FEXTRALIB)
+	$(CC) $(CFLAGS) $(LDFLAGS) libopenblas.def dllinit.$(SUFFIX) \
+	-shared -o ../$(LIBDLLNAME) -Wl,--out-implib,../$(LIBPREFIX).lib \
+	-Wl,--whole-archive ../$(LIBNAME) -Wl,--no-whole-archive $(FEXTRALIB)
 
 libopenblas.def : gensymbol
 	perl ./gensymbol win2k    $(ARCH) dummy $(EXPRECISION) $(NO_CBLAS) $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
 
-libgoto2_shared.def : gensymbol
-	perl ./gensymbol win2k    $(ARCH) dummy $(EXPRECISION) $(NO_CBLAS) $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
-
 libgoto_hpl.def : gensymbol
 	perl ./gensymbol win2khpl $(ARCH) dummy $(EXPRECISION) $(NO_CBLAS) $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
 
 $(LIBDYNNAME) : ../$(LIBNAME) osx.def
 	$(FC) $(FFLAGS) -all_load -headerpad_max_install_names -install_name $(CURDIR)/../$(LIBDYNNAME) -dynamiclib -o ../$(LIBDYNNAME) $< -Wl,-exported_symbols_list,osx.def  $(FEXTRALIB)
 
-symbol.$(SUFFIX) : symbol.S
-	$(CC) $(CFLAGS) -c -o $(@F) $^
-
 dllinit.$(SUFFIX) : dllinit.c
 	$(CC) $(CFLAGS) -c -o $(@F) -s $<
 
 ifeq ($(OSNAME), Linux)
 
 so : ../$(LIBSONAME)
 
-../$(LIBSONAME) : ../$(LIBNAME) linux.def linktest.c
+../$(LIBSONAME) : ../$(LIBNAME) linktest.c
 ifneq ($(C_COMPILER), LSB)
 	$(CC) $(CFLAGS) $(LDFLAGS) -shared -o ../$(LIBSONAME) \
 	-Wl,--whole-archive ../$(LIBNAME) -Wl,--no-whole-archive \
@@ -145,7 +125,7 @@ ifeq ($(OSNAME), $(filter $(OSNAME),FreeBSD NetBSD))
 
 so : ../$(LIBSONAME)
 
-../$(LIBSONAME) : ../$(LIBNAME) linux.def linktest.c
+../$(LIBSONAME) : ../$(LIBNAME) linktest.c
 	$(CC) $(CFLAGS) $(LDFLAGS)  -shared -o ../$(LIBSONAME) \
 	-Wl,--whole-archive ../$(LIBNAME) -Wl,--no-whole-archive \
 	$(FEXTRALIB) $(EXTRALIB)
@@ -197,18 +177,12 @@ static : ../$(LIBNAME)
 	$(AR) -cq ../$(LIBNAME) goto.$(SUFFIX)
 	rm -f goto.$(SUFFIX)
 
-linux.def : gensymbol ../Makefile.system ../getarch.c
-	perl ./gensymbol linux $(ARCH) $(BU) $(EXPRECISION) $(NO_CBLAS)  $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
-
 osx.def : gensymbol ../Makefile.system ../getarch.c
 	perl ./gensymbol osx $(ARCH) $(BU) $(EXPRECISION) $(NO_CBLAS)  $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
 
 aix.def : gensymbol ../Makefile.system ../getarch.c
 	perl ./gensymbol aix $(ARCH) $(BU) $(EXPRECISION) $(NO_CBLAS)  $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > $(@F)
 
-symbol.S : gensymbol
-	perl ./gensymbol win2kasm noarch dummy $(EXPRECISION) $(NO_CBLAS)  $(NO_LAPACK) $(NO_LAPACKE) $(NEED2UNDERSCORES) $(ONLY_CBLAS) > symbol.S
-
 test : linktest.c
 	$(CC) $(CFLAGS) $(LDFLAGS) -w -o linktest linktest.c ../$(LIBSONAME) -lm && echo OK.
 	rm -f linktest