Merge branch 'develop'

OpenMathLib · Oct 13, 2014 · 7e4e195 · 7e4e195
2 parents a7126c2 + ac5a7e1
commit 7e4e195
Show file tree

Hide file tree

Showing 134 changed files with 28,183 additions and 5,278 deletions.
diff --git a/Changelog.txt b/Changelog.txt
@@ -1,6 +1,21 @@
 OpenBLAS ChangeLog
 ====================================================================
-Version 0.2.10
+Version 0.2.12
+13-Oct-2014
+common:
+	* Added CBLAS interface for ?omatcopy and ?imatcopy.
+	* Enable ?gemm3m functions.
+	* Added benchmark for ?gemm3m.
+	* Optimized multithreading lower limits.
+	* Disabled SYMM3M and HEMM3M functions 
+	  because of segment violations.
+
+x86/x86-64:
+	* Improved axpy and symv performance on AMD Bulldozer.
+	* Improved gemv performance on modern Intel and AMD CPUs.
+
+====================================================================
+Version 0.2.11
 18-Aug-2014
 common:
 	* Added some benchmark codes.

diff --git a/Makefile.rule b/Makefile.rule
@@ -3,7 +3,7 @@
 #
 
 # This library's version
-VERSION = 0.2.11
+VERSION = 0.2.12
 
 # If you set the suffix, the library name will be libopenblas_$(LIBNAMESUFFIX).a
 # and libopenblas_$(LIBNAMESUFFIX).so. Meanwhile, the soname in shared library

diff --git a/Makefile.system b/Makefile.system
@@ -339,7 +339,7 @@ FCOMMON_OPT	+= -m128bit-long-double
 endif
 ifeq ($(C_COMPILER), CLANG)
 EXPRECISION	= 1
-CCOMMON_OPT	+= -DEXPRECISION
+CCOMMON_OPT	+= -DEXPRECISION 
 FCOMMON_OPT	+= -m128bit-long-double
 endif
 endif
@@ -350,6 +350,7 @@ ifeq ($(C_COMPILER), INTEL)
 CCOMMON_OPT    += -wd981
 endif
 
+
 ifeq ($(USE_OPENMP), 1)
 # ifeq logical or. GCC or LSB
 ifeq ($(C_COMPILER), $(filter $(C_COMPILER),GCC LSB))

diff --git a/README.md b/README.md
@@ -55,16 +55,23 @@ Please read GotoBLAS_01Readme.txt
 
 #### x86/x86-64:
 - **Intel Xeon 56xx (Westmere)**: Used GotoBLAS2 Nehalem codes.
-- **Intel Sandy Bridge**: Optimized Level-3 BLAS with AVX on x86-64.
-- **Intel Haswell**: Optimized Level-3 BLAS with AVX on x86-64 (identical to Sandy Bridge).
+- **Intel Sandy Bridge**: Optimized Level-3 and Level-2 BLAS with AVX on x86-64.
+- **Intel Haswell**: Optimized Level-3 and Level-2 BLAS with AVX2 and FMA  on x86-64.
 - **AMD Bobcat**: Used GotoBLAS2 Barcelona codes.
-- **AMD Bulldozer**: x86-64 S/DGEMM AVX kernels. (Thank Werner Saar)
-- **AMD PILEDRIVER**: Used Bulldozer codes.
+- **AMD Bulldozer**: x86-64 ?GEMM FMA4 kernels. (Thank Werner Saar)
+- **AMD PILEDRIVER**: Uses Bulldozer codes with some optimizations.
 
 #### MIPS64:
 - **ICT Loongson 3A**: Optimized Level-3 BLAS and the part of Level-1,2.
 - **ICT Loongson 3B**: Experimental
 
+#### ARM:
+- **ARMV6**: Optimized BLAS for vfpv2 and vfpv3-d16 ( e.g. BCM2835, Cortex M0+ )
+- **ARMV7**: Optimized BLAS for vfpv3-d32 ( e.g. Cortex A8, A9 and A15 )
+
+#### ARM64:
+- **ARMV8**: Experimental
+
 ### Support OS:
 - **GNU/Linux**
 - **MingWin/Windows**: Please read </~https://github.com/xianyi/OpenBLAS/wiki/How-to-use-OpenBLAS-in-Microsoft-Visual-Studio>.
@@ -116,8 +123,8 @@ Please see Changelog.txt to obtain the differences between GotoBLAS2 1.13 BSD ve
 * Please read [Faq](/~https://github.com/xianyi/OpenBLAS/wiki/Faq) at first.
 * Please use gcc version 4.6 and above to compile Sandy Bridge AVX kernels on Linux/MingW/BSD.
 * Please use Clang version 3.1 and above to compile the library on Sandy Bridge microarchitecture. The Clang 3.0 will generate the wrong AVX binary code.
-* The number of CPUs/Cores should less than or equal to 256.
-* On Linux, OpenBLAS sets the processor affinity by default. This may cause [the conflict with R parallel](https://stat.ethz.ch/pipermail/r-sig-hpc/2012-April/001348.html). You can build the library with NO_AFFINITY=1.
+* The number of CPUs/Cores should less than or equal to 256. On Linux x86_64(amd64), there is experimental support for up to 1024 CPUs/Cores and 128 numa nodes if you build the library with BIGNUMA=1.
+* OpenBLAS does not set processor affinity by default. On Linux, you can enable processor affinity by commenting the line NO_AFFINITY=1 in Makefile.rule. But this may cause [the conflict with R parallel](https://stat.ethz.ch/pipermail/r-sig-hpc/2012-April/001348.html).
 * On Loongson 3A. make test would be failed because of pthread_create error. The error code is EAGAIN. However, it will be OK when you run the same testcase on shell.
 
 ## Contributing

diff --git a/TargetList.txt b/TargetList.txt
@@ -19,6 +19,7 @@ PENRYN
 DUNNINGTON
 NEHALEM
 SANDYBRIDGE
+HASWELL
 ATOM
 
 b)AMD CPU:
@@ -30,6 +31,7 @@ SHANGHAI
 ISTANBUL
 BOBCAT
 BULLDOZER
+PILEDRIVER
 
 c)VIA CPU:
 SSE_GENERIC
@@ -59,3 +61,7 @@ ITANIUM2
 SPARC
 SPARCV7
 
+6.ARM CPU:
+ARMV7
+ARMV6
+ARMV5
diff --git a/benchmark/Makefile b/benchmark/Makefile
@@ -35,7 +35,10 @@ goto :: slinpack.goto dlinpack.goto clinpack.goto zlinpack.goto \
        ssyrk.goto dsyrk.goto csyrk.goto zsyrk.goto \
        ssyr2k.goto dsyr2k.goto csyr2k.goto zsyr2k.goto \
        sger.goto dger.goto \
-       ssymv.goto dsymv.goto \
+       sdot.goto ddot.goto \
+       saxpy.goto daxpy.goto caxpy.goto zaxpy.goto \
+       ssymv.goto dsymv.goto csymv.goto zsymv.goto \
+       chemv.goto zhemv.goto \
        chemm.goto zhemm.goto \
        cherk.goto zherk.goto \
        cher2k.goto zher2k.goto \
@@ -53,7 +56,10 @@ acml :: slinpack.acml dlinpack.acml clinpack.acml zlinpack.acml \
        ssyrk.acml dsyrk.acml csyrk.acml zsyrk.acml \
        ssyr2k.acml dsyr2k.acml csyr2k.acml zsyr2k.acml \
        sger.acml dger.acml \
-       ssymv.acml dsymv.acml \
+       sdot.acml ddot.acml \
+       saxpy.acml daxpy.acml caxpy.acml zaxpy.acml \
+       ssymv.acml dsymv.acml csymv.acml zsymv.acml \
+       chemv.acml zhemv.acml \
        chemm.acml zhemm.acml \
        cherk.acml zherk.acml \
        cher2k.acml zher2k.acml \
@@ -71,7 +77,10 @@ atlas :: slinpack.atlas dlinpack.atlas clinpack.atlas zlinpack.atlas \
        ssyrk.atlas dsyrk.atlas csyrk.atlas zsyrk.atlas \
        ssyr2k.atlas dsyr2k.atlas csyr2k.atlas zsyr2k.atlas \
        sger.atlas dger.atlas \
-       ssymv.atlas dsymv.atlas \
+       sdot.atlas ddot.atlas \
+       saxpy.atlas daxpy.atlas caxpy.atlas zaxpy.atlas \
+       ssymv.atlas dsymv.atlas csymv.atlas zsymv.atlas \
+       chemv.atlas zhemv.atlas \
        chemm.acml zhemm.acml \
        chemm.atlas zhemm.atlas \
        cherk.atlas zherk.atlas \
@@ -90,7 +99,10 @@ mkl :: slinpack.mkl dlinpack.mkl clinpack.mkl zlinpack.mkl \
        ssyrk.mkl dsyrk.mkl csyrk.mkl zsyrk.mkl \
        ssyr2k.mkl dsyr2k.mkl csyr2k.mkl zsyr2k.mkl \
        sger.mkl dger.mkl \
-       ssymv.mkl dsymv.mkl \
+       sdot.mkl ddot.mkl \
+       saxpy.mkl daxpy.mkl caxpy.mkl zaxpy.mkl \
+       ssymv.mkl dsymv.mkl csymv.mkl zsymv.mkl \
+       chemv.mkl zhemv.mkl \
        chemm.mkl zhemm.mkl \
        cherk.mkl zherk.mkl \
        cher2k.mkl zher2k.mkl \
@@ -100,7 +112,12 @@ mkl :: slinpack.mkl dlinpack.mkl clinpack.mkl zlinpack.mkl \
        spotrf.mkl dpotrf.mkl cpotrf.mkl zpotrf.mkl \
        ssymm.mkl dsymm.mkl csymm.mkl zsymm.mkl
 
-all :: goto atlas acml mkl
+
+goto_3m :: cgemm3m.goto zgemm3m.goto
+
+mkl_3m  :: cgemm3m.mkl zgemm3m.mkl
+
+all :: goto mkl atlas acml
 
 ##################################### Slinpack ####################################################
 slinpack.goto : slinpack.$(SUFFIX) ../$(LIBNAME)
@@ -732,6 +749,32 @@ dsymv.atlas : dsymv.$(SUFFIX)
 dsymv.mkl : dsymv.$(SUFFIX)
 	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
 
+##################################### Csymv ####################################################
+csymv.goto : csymv.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+csymv.acml : csymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+csymv.atlas : csymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+csymv.mkl : csymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Dsymv ####################################################
+zsymv.goto : zsymv.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+zsymv.acml : zsymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zsymv.atlas : zsymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zsymv.mkl : zsymv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
 ##################################### Sgeev ####################################################
 sgeev.goto : sgeev.$(SUFFIX) ../$(LIBNAME)
 	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
@@ -896,6 +939,131 @@ zpotrf.atlas : zpotrf.$(SUFFIX)
 zpotrf.mkl : zpotrf.$(SUFFIX)
 	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
 
+##################################### Chemv ####################################################
+
+chemv.goto : chemv.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+chemv.acml : chemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+chemv.atlas : chemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+chemv.mkl : chemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Zhemv ####################################################
+
+zhemv.goto : zhemv.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+zhemv.acml : zhemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zhemv.atlas : zhemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zhemv.mkl : zhemv.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Sdot ####################################################
+sdot.goto : sdot.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+sdot.acml : sdot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+sdot.atlas : sdot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+sdot.mkl : sdot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Ddot ####################################################
+ddot.goto : ddot.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+ddot.acml : ddot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+ddot.atlas : ddot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+ddot.mkl : ddot.$(SUFFIX)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Saxpy ####################################################
+saxpy.goto : saxpy.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+saxpy.acml : saxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+saxpy.atlas : saxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+saxpy.mkl : saxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Daxpy ####################################################
+daxpy.goto : daxpy.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+daxpy.acml : daxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+daxpy.atlas : daxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+daxpy.mkl : daxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Caxpy ####################################################
+
+caxpy.goto : caxpy.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+caxpy.acml : caxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+caxpy.atlas : caxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+caxpy.mkl : caxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Zaxpy ####################################################
+
+zaxpy.goto : zaxpy.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+zaxpy.acml : zaxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBACML) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zaxpy.atlas : zaxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBATLAS) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+zaxpy.mkl : zaxpy.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+
+##################################### Cgemm3m ####################################################
+
+cgemm3m.goto : cgemm3m.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+cgemm3m.mkl : cgemm3m.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
+##################################### Zgemm3m ####################################################
+
+zgemm3m.goto : zgemm3m.$(SUFFIX) ../$(LIBNAME)
+	$(CC) $(CFLAGS) -o $(@F) $^ $(CEXTRALIB) $(EXTRALIB) -lm
+
+zgemm3m.mkl : zgemm3m.$(SUFFIX)
+	-$(CC) $(CFLAGS) -o $(@F) $^ $(LIBMKL) $(CEXTRALIB) $(EXTRALIB) $(FEXTRALIB)
+
 
 ###################################################################################################
 
@@ -1037,6 +1205,12 @@ ssymv.$(SUFFIX) : symv.c
 dsymv.$(SUFFIX) : symv.c
 	$(CC) $(CFLAGS) -c -UCOMPLEX -DDOUBLE -o $(@F) $^
 
+csymv.$(SUFFIX) : symv.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
+
+zsymv.$(SUFFIX) : symv.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
+
 sgeev.$(SUFFIX) : geev.c
 	$(CC) $(CFLAGS) -c -UCOMPLEX -UDOUBLE -o $(@F) $^
 
@@ -1073,8 +1247,35 @@ cpotrf.$(SUFFIX) : potrf.c
 zpotrf.$(SUFFIX) : potrf.c
 	$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
 
+chemv.$(SUFFIX) : hemv.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
+
+zhemv.$(SUFFIX) : hemv.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
+
+sdot.$(SUFFIX) : dot.c
+	$(CC) $(CFLAGS) -c -UCOMPLEX -UDOUBLE -o $(@F) $^
+
+ddot.$(SUFFIX) : dot.c
+	$(CC) $(CFLAGS) -c -UCOMPLEX -DDOUBLE -o $(@F) $^
+
+saxpy.$(SUFFIX) : axpy.c
+	$(CC) $(CFLAGS) -c -UCOMPLEX -UDOUBLE -o $(@F) $^
+
+daxpy.$(SUFFIX) : axpy.c
+	$(CC) $(CFLAGS) -c -UCOMPLEX -DDOUBLE -o $(@F) $^
+
+caxpy.$(SUFFIX) : axpy.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
 
+zaxpy.$(SUFFIX) : axpy.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
 
+cgemm3m.$(SUFFIX) : gemm3m.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -UDOUBLE -o $(@F) $^
+
+zgemm3m.$(SUFFIX) : gemm3m.c
+	$(CC) $(CFLAGS) -c -DCOMPLEX -DDOUBLE -o $(@F) $^
 
 
 clean ::