Merge pull request #29 from xianyi/develop

rebase

Author: martin-frbg

.drone.yml

@@ -92,7 +92,7 @@ steps:
     - mkdir build && cd build
     - cmake $CMAKE_FLAGS ..
     - make -j
-    - ctest
+    - ctest -V
 
 ---
 kind: pipeline
@@ -116,7 +116,7 @@ steps:
     - mkdir build && cd build
     - cmake $CMAKE_FLAGS ..
     - make -j
-    - ctest
+    - ctest -V
 
 ---
 kind: pipeline
@@ -140,4 +140,4 @@ steps:
     - mkdir build && cd build
     - cmake $CMAKE_FLAGS ..
     - make -j
-    - ctest
+    - ctest -V

.gitignore

@@ -87,4 +87,5 @@ build.*
 *.swp
 benchmark/*.goto
 benchmark/smallscaling
-
+CMakeCache.txt
+CMakeFiles/*

kernel/power/KERNEL.POWER8

@@ -89,30 +89,52 @@ ZTRSMKERNEL_RT	= ../generic/trsm_kernel_RT.c
 #SMINKERNEL   =  ../arm/min.c
 #DMINKERNEL   = ../arm/min.c
 #
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
+ifneq ($(GCCVERSIONGTEQ9),1)
 ISAMAXKERNEL = isamax_power8.S
 else
 ISAMAXKERNEL = isamax.c
 endif
+else
+ISAMAXKERNEL = isamax.c
+endif
+#
 IDAMAXKERNEL = idamax.c
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+#
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
+ifneq ($(GCCVERSIONGTEQ9),1)
 ICAMAXKERNEL = icamax_power8.S
 else
 ICAMAXKERNEL = icamax.c
 endif
+else
+ICAMAXKERNEL = icamax.c
+endif
+#
 IZAMAXKERNEL = izamax.c
 #
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
+ifneq ($(GCCVERSIONGTEQ9),1)
 ISAMINKERNEL = isamin_power8.S
 else
 ISAMINKERNEL = isamin.c
 endif
+else
+ISAMINKERNEL = isamin.c
+endif
+#
 IDAMINKERNEL = idamin.c
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+#
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
+ifneq ($(GCCVERSIONGTEQ9),1)
 ICAMINKERNEL = icamin_power8.S
 else
 ICAMINKERNEL = icamin.c
 endif
+else
+ICAMINKERNEL = icamin.c
+endif
+#
 IZAMINKERNEL = izamin.c
 #
 #ISMAXKERNEL  = ../arm/imax.c
@@ -128,11 +150,16 @@ ZASUMKERNEL  = zasum.c
 #
 SAXPYKERNEL  = saxpy.c
 DAXPYKERNEL  = daxpy.c
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
+ifneq ($(GCCVERSIONGTEQ9),1)
 CAXPYKERNEL  = caxpy_power8.S
 else
 CAXPYKERNEL  = caxpy.c
 endif
+else
+CAXPYKERNEL  = caxpy.c
+endif
+#
 ZAXPYKERNEL  = zaxpy.c
 #
 SCOPYKERNEL  = scopy.c

kernel/power/KERNEL.PPC440

@@ -15,7 +15,7 @@ ZASUMKERNEL	= zasum_ppc440.S
 
 SAXPYKERNEL	=  axpy_ppc440.S
 DAXPYKERNEL	=  axpy_ppc440.S
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 CAXPYKERNEL	= ../arm/zaxpy.c
 ZAXPYKERNEL	= ../arm/zaxpy.c
 else
@@ -25,7 +25,7 @@ endif
 
 SDOTKERNEL	=  dot_ppc440.S
 DDOTKERNEL	=  dot_ppc440.S
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 CDOTKERNEL	= zdot_ppc440.S
 ZDOTKERNEL	= zdot_ppc440.S
 else
@@ -62,7 +62,7 @@ ZNRM2KERNEL	= znrm2_ppc440.S
 
 SROTKERNEL	=  rot_ppc440.S
 DROTKERNEL	=  rot_ppc440.S
-ifneq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifneq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 CROTKERNEL	= zrot_ppc440.S
 ZROTKERNEL	= zrot_ppc440.S
 else
@@ -132,7 +132,7 @@ ZTRSMKERNEL_LT	=  ztrsm_kernel_ppc440_LT.S
 ZTRSMKERNEL_RN	=  ztrsm_kernel_ppc440_LT.S
 ZTRSMKERNEL_RT	=  ztrsm_kernel_ppc440_RT.S
 
-ifeq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifeq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 SGEMVNKERNEL = ../arm/gemv_n.c
 DGEMVNKERNEL = ../arm/gemv_n.c
 SGEMVTKERNEL = ../arm/gemv_t.c

kernel/power/KERNEL.PPC970

@@ -1,4 +1,4 @@
-ifeq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifeq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 SGEMMKERNEL    =  gemm_kernel.S
 SGEMMINCOPY    =  
 SGEMMITCOPY    =  
@@ -30,7 +30,7 @@ DGEMMITCOPYOBJ =
 DGEMMONCOPYOBJ =  dgemm_oncopy$(TSUFFIX).$(SUFFIX)
 DGEMMOTCOPYOBJ =  dgemm_otcopy$(TSUFFIX).$(SUFFIX)
 
-ifeq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifeq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 CGEMMKERNEL    =  zgemm_kernel.S
 CGEMMINCOPY    =
 CGEMMITCOPY    =
@@ -72,7 +72,7 @@ ZTRSMKERNEL_LT	=  ztrsm_kernel_LT.S
 ZTRSMKERNEL_RN	=  ztrsm_kernel_LT.S
 ZTRSMKERNEL_RT	=  ztrsm_kernel_RT.S
 
-ifeq ($(__BYTE_ORDER__),$(__ORDER_BIG_ENDIAN__))
+ifeq ($(__BYTE_ORDER__),"__ORDER_BIG_ENDIAN__")
 STRSMKERNEL_LN	=  trsm_kernel_LN.S
 STRSMKERNEL_LT	=  trsm_kernel_LT.S
 STRSMKERNEL_RN	=  trsm_kernel_LT.S

kernel/power/caxpy_power8.S

@@ -12,6 +12,13 @@
 
 	PROLOGUE
 
+#if _CALL_ELF ==2
+#ifdef CONJ
+caxpyc_k:
+#else
+caxpy_k:
+#endif
+#endif
 .LCF0:
 0:	addis 2,12,.TOC.-.LCF0@ha
 	addi 2,2,.TOC.-.LCF0@l

kernel/power/icamin_power8.S

@@ -10,7 +10,9 @@
 #include "common.h"
 
 	PROLOGUE
-
+#if _CALL_ELF ==2
+icamin_k:
+#endif
 .LCF0:
 0:	addis 2,12,.TOC.-.LCF0@ha
 	addi 2,2,.TOC.-.LCF0@l

kernel/power/isamax_power8.S

@@ -11,7 +11,10 @@
 #include "common.h"
 
 	PROLOGUE
-
+	
+#if _CALL_ELF == 2	
+isamax_k:
+#endif
 .LCF0:
 0:	addis 2,12,.TOC.-.LCF0@ha
 	addi 2,2,.TOC.-.LCF0@l

kernel/power/isamin_power8.S

@@ -11,6 +11,9 @@
 
 	PROLOGUE
 
+#if _CALL_ELF ==2
+isamin_k:
+#endif
 .LCF0:
 0:	addis 2,12,.TOC.-.LCF0@ha
 	addi 2,2,.TOC.-.LCF0@l

lapack/CMakeLists.txt

@@ -4,7 +4,6 @@ include_directories(${PROJECT_BINARY_DIR})
 
 
 set(LAPACK_SOURCES
-  getrf/getrf_single.c
   potrf/potrf_U_single.c
   potrf/potrf_L_single.c
   lauum/lauum_U_single.c
@@ -45,6 +44,10 @@ GenerateNamedObjects("${LAPACK_MANGLED_SOURCES}" "" "" false "" "" false 3)
 GenerateNamedObjects("laswp/generic/laswp_k_4.c" "" "laswp_plus" false "" ""  false 3)
 GenerateNamedObjects("laswp/generic/laswp_k_4.c" "MINUS" "laswp_minus" false "" ""  false 3)
 
+foreach (float_type ${FLOAT_TYPES})
+GenerateNamedObjects("getrf/getrf_single.c" "UNIT" "getrf_single" false "" "" false ${float_type})
+endforeach ()
+
 # dynamic_arch laswp needs arch specific code ?
 #foreach(TARGET_CORE ${DYNAMIC_CORE})
 #      set(TSUFFIX "_${TARGET_CORE}")
@@ -81,7 +84,7 @@ if (USE_THREAD)
   )
 
   foreach (float_type ${FLOAT_TYPES})
-    GenerateNamedObjects("${GETRF_SRC}" "" "getrf_parallel" false "" "" false ${float_type})  
+    GenerateNamedObjects("${GETRF_SRC}" "UNIT" "getrf_parallel" false "" "" false ${float_type})
   endforeach()
 
   GenerateNamedObjects("${PARALLEL_SOURCES}")