Merge pull request #4960 from martin-frbg/gemmtr
Provide copies of GEMMT under its new name GEMMTR adopted by the Reference BLAS
This commit is contained in:
commit
c634114c8a
|
@ -21,7 +21,7 @@ blasobjsc="
|
|||
chbmv chemm chemv cher2 cher2k cher cherk scabs1 scamax
|
||||
chpmv chpr2 chpr crotg cscal csrot csscal cswap scamin scasum scnrm2
|
||||
csymm csyr2k csyrk ctbmv ctbsv ctpmv ctpsv ctrmm ctrmv ctrsm
|
||||
ctrsv icamax icamin cimatcopy comatcopy cgeadd scsum cgemmt"
|
||||
ctrsv icamax icamin cimatcopy comatcopy cgeadd scsum cgemmt cgemmtr"
|
||||
|
||||
blasobjsd="
|
||||
damax damin dasum daxpy daxpby dcabs1 dcopy ddot dgbmv dgemm
|
||||
|
@ -29,7 +29,7 @@ blasobjsd="
|
|||
dscal dsdot dspmv dspr2 dimatcopy domatcopy
|
||||
dspr dswap dsymm dsymv dsyr2 dsyr2k dsyr dsyrk dtbmv dtbsv
|
||||
dtpmv dtpsv dtrmm dtrmv dtrsm dtrsv
|
||||
idamax idamin idmax idmin dgeadd dsum dgemmt"
|
||||
idamax idamin idmax idmin dgeadd dsum dgemmt dgemmtr"
|
||||
|
||||
blasobjss="
|
||||
isamax isamin ismax ismin
|
||||
|
@ -38,7 +38,7 @@ blasobjss="
|
|||
smax smin snrm2 simatcopy somatcopy
|
||||
srot srotg srotm srotmg ssbmv sscal sspmv sspr2 sspr sswap
|
||||
ssymm ssymv ssyr2 ssyr2k ssyr ssyrk stbmv stbsv stpmv stpsv
|
||||
strmm strmv strsm strsv sgeadd ssum sgemmt"
|
||||
strmm strmv strsm strsv sgeadd ssum sgemmt sgemmtr"
|
||||
|
||||
blasobjsz="
|
||||
izamax izamin
|
||||
|
@ -48,17 +48,17 @@ blasobjsz="
|
|||
zhpr zrotg zscal zswap zsymm zsyr2k zsyrk ztbmv
|
||||
ztbsv ztpmv ztpsv ztrmm ztrmv ztrsm ztrsv
|
||||
zomatcopy zimatcopy dzamax dzamin dzasum dznrm2
|
||||
zgeadd dzsum zgemmt"
|
||||
zgeadd dzsum zgemmt zgemmtr"
|
||||
|
||||
blasobjs="lsame xerbla"
|
||||
bfblasobjs="sbgemm sbgemv sbdot sbstobf16 sbdtobf16 sbf16tos dbf16tod"
|
||||
bfblasobjs="sbgemm sbgemmt sbgemmtr sbgemv sbdot sbstobf16 sbdtobf16 sbf16tos dbf16tod"
|
||||
cblasobjsc="
|
||||
cblas_caxpy cblas_ccopy cblas_cdotc cblas_cdotu cblas_cgbmv cblas_cgemm cblas_cgemv
|
||||
cblas_cgerc cblas_cgeru cblas_chbmv cblas_chemm cblas_chemv cblas_cher2 cblas_cher2k
|
||||
cblas_cher cblas_cherk cblas_chpmv cblas_chpr2 cblas_chpr cblas_cscal cblas_caxpby
|
||||
cblas_csscal cblas_cswap cblas_csymm cblas_csyr2k cblas_csyrk cblas_ctbmv cblas_cgeadd
|
||||
cblas_ctbsv cblas_ctpmv cblas_ctpsv cblas_ctrmm cblas_ctrmv cblas_ctrsm cblas_ctrsv
|
||||
cblas_scnrm2 cblas_scasum cblas_cgemmt
|
||||
cblas_scnrm2 cblas_scasum cblas_cgemmt cblas_cgemmtr
|
||||
cblas_icamax cblas_icamin cblas_icmin cblas_icmax cblas_scsum cblas_cimatcopy cblas_comatcopy
|
||||
cblas_caxpyc cblas_crotg cblas_csrot cblas_scamax cblas_scamin cblas_cgemm_batch
|
||||
"
|
||||
|
@ -68,7 +68,7 @@ cblasobjsd="
|
|||
cblas_drot cblas_drotg cblas_drotm cblas_drotmg cblas_dsbmv cblas_dscal cblas_dsdot
|
||||
cblas_dspmv cblas_dspr2 cblas_dspr cblas_dswap cblas_dsymm cblas_dsymv cblas_dsyr2
|
||||
cblas_dsyr2k cblas_dsyr cblas_dsyrk cblas_dtbmv cblas_dtbsv cblas_dtpmv cblas_dtpsv
|
||||
cblas_dtrmm cblas_dtrmv cblas_dtrsm cblas_dtrsv cblas_daxpby cblas_dgeadd cblas_dgemmt
|
||||
cblas_dtrmm cblas_dtrmv cblas_dtrsm cblas_dtrsv cblas_daxpby cblas_dgeadd cblas_dgemmt cblas_dgemmtr
|
||||
cblas_idamax cblas_idamin cblas_idmin cblas_idmax cblas_dsum cblas_dimatcopy cblas_domatcopy
|
||||
cblas_damax cblas_damin cblas_dgemm_batch
|
||||
"
|
||||
|
@ -80,7 +80,7 @@ cblasobjss="
|
|||
cblas_srotm cblas_srotmg cblas_ssbmv cblas_sscal cblas_sspmv cblas_sspr2 cblas_sspr
|
||||
cblas_sswap cblas_ssymm cblas_ssymv cblas_ssyr2 cblas_ssyr2k cblas_ssyr cblas_ssyrk
|
||||
cblas_stbmv cblas_stbsv cblas_stpmv cblas_stpsv cblas_strmm cblas_strmv cblas_strsm
|
||||
cblas_strsv cblas_sgeadd cblas_sgemmt
|
||||
cblas_strsv cblas_sgeadd cblas_sgemmt cblas_sgemmtr
|
||||
cblas_isamax cblas_isamin cblas_ismin cblas_ismax cblas_ssum cblas_simatcopy cblas_somatcopy
|
||||
cblas_samax cblas_samin cblas_sgemm_batch
|
||||
"
|
||||
|
@ -92,7 +92,7 @@ cblasobjsz="
|
|||
cblas_zhpr cblas_zscal cblas_zswap cblas_zsymm cblas_zsyr2k cblas_zsyrk
|
||||
cblas_ztbmv cblas_ztbsv cblas_ztpmv cblas_ztpsv cblas_ztrmm cblas_ztrmv cblas_ztrsm
|
||||
cblas_ztrsv cblas_cdotc_sub cblas_cdotu_sub cblas_zdotc_sub cblas_zdotu_sub
|
||||
cblas_zaxpby cblas_zgeadd cblas_zgemmt
|
||||
cblas_zaxpby cblas_zgeadd cblas_zgemmt cblas_zgemmtr
|
||||
cblas_izamax cblas_izamin cblas_izmin cblas_izmax cblas_dzsum cblas_zimatcopy cblas_zomatcopy
|
||||
cblas_zaxpyc cblas_zdrot cblas_zrotg cblas_dzamax cblas_dzamin cblas_zgemm_batch
|
||||
"
|
||||
|
|
|
@ -108,6 +108,9 @@ endif ()
|
|||
# trmm is trsm with a compiler flag set
|
||||
GenerateNamedObjects("trsm.c" "TRMM" "trmm" ${CBLAS_FLAG})
|
||||
|
||||
# gemmtr is gemmt under the name adopted by the Reference BLAS
|
||||
GenerateNamedObjects("gemm.c" "" "gemmtr" ${CBLAS_FLAG})
|
||||
|
||||
# max and imax are compiled 4 times
|
||||
GenerateNamedObjects("max.c" "" "" ${CBLAS_FLAG})
|
||||
GenerateNamedObjects("max.c" "USE_ABS" "amax" ${CBLAS_FLAG})
|
||||
|
@ -123,6 +126,7 @@ if (BUILD_BFLOAT16)
|
|||
GenerateNamedObjects("bf16dot.c" "" "sbdot" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("gemm.c" "" "sbgemm" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("gemmt.c" "" "sbgemmt" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("gemmt.c" "" "sbgemmtr" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("sbgemv.c" "" "sbgemv" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("tobf16.c" "SINGLE_PREC" "sbstobf16" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
GenerateNamedObjects("tobf16.c" "DOUBLE_PREC" "sbdtobf16" ${CBLAS_FLAG} "" "" true "BFLOAT16")
|
||||
|
|
|
@ -44,12 +44,12 @@ SBLAS3OBJS = \
|
|||
sgemm.$(SUFFIX) ssymm.$(SUFFIX) strmm.$(SUFFIX) \
|
||||
strsm.$(SUFFIX) ssyrk.$(SUFFIX) ssyr2k.$(SUFFIX) \
|
||||
somatcopy.$(SUFFIX) simatcopy.$(SUFFIX)\
|
||||
sgeadd.$(SUFFIX) sgemmt.$(SUFFIX)
|
||||
sgeadd.$(SUFFIX) sgemmt.$(SUFFIX) sgemmtr.$(SUFFIX)
|
||||
|
||||
ifeq ($(BUILD_BFLOAT16),1)
|
||||
SBBLAS1OBJS = sbdot.$(SUFFIX)
|
||||
SBBLAS2OBJS = sbgemv.$(SUFFIX)
|
||||
SBBLAS3OBJS = sbgemm.$(SUFFIX) sbgemmt.$(SUFFIX)
|
||||
SBBLAS3OBJS = sbgemm.$(SUFFIX) sbgemmt.$(SUFFIX) sbgemmtr.$(SUFFIX)
|
||||
SBEXTOBJS = sbstobf16.$(SUFFIX) sbdtobf16.$(SUFFIX) sbf16tos.$(SUFFIX) dbf16tod.$(SUFFIX)
|
||||
endif
|
||||
|
||||
|
@ -76,7 +76,7 @@ DBLAS3OBJS = \
|
|||
dgemm.$(SUFFIX) dsymm.$(SUFFIX) dtrmm.$(SUFFIX) \
|
||||
dtrsm.$(SUFFIX) dsyrk.$(SUFFIX) dsyr2k.$(SUFFIX) \
|
||||
domatcopy.$(SUFFIX) dimatcopy.$(SUFFIX)\
|
||||
dgeadd.$(SUFFIX) dgemmt.$(SUFFIX)
|
||||
dgeadd.$(SUFFIX) dgemmt.$(SUFFIX) dgemmtr.$(SUFFIX)
|
||||
|
||||
CBLAS1OBJS = \
|
||||
caxpy.$(SUFFIX) caxpyc.$(SUFFIX) cswap.$(SUFFIX) \
|
||||
|
@ -105,7 +105,7 @@ CBLAS3OBJS = \
|
|||
ctrsm.$(SUFFIX) csyrk.$(SUFFIX) csyr2k.$(SUFFIX) \
|
||||
chemm.$(SUFFIX) cherk.$(SUFFIX) cher2k.$(SUFFIX) \
|
||||
comatcopy.$(SUFFIX) cimatcopy.$(SUFFIX)\
|
||||
cgeadd.$(SUFFIX) cgemmt.$(SUFFIX)
|
||||
cgeadd.$(SUFFIX) cgemmt.$(SUFFIX) cgemmtr.$(SUFFIX)
|
||||
|
||||
ZBLAS1OBJS = \
|
||||
zaxpy.$(SUFFIX) zaxpyc.$(SUFFIX) zswap.$(SUFFIX) \
|
||||
|
@ -134,7 +134,7 @@ ZBLAS3OBJS = \
|
|||
ztrsm.$(SUFFIX) zsyrk.$(SUFFIX) zsyr2k.$(SUFFIX) \
|
||||
zhemm.$(SUFFIX) zherk.$(SUFFIX) zher2k.$(SUFFIX) \
|
||||
zomatcopy.$(SUFFIX) zimatcopy.$(SUFFIX)\
|
||||
zgeadd.$(SUFFIX) zgemmt.$(SUFFIX)
|
||||
zgeadd.$(SUFFIX) zgemmt.$(SUFFIX) zgemmtr.$(SUFFIX)
|
||||
|
||||
ifeq ($(SUPPORT_GEMM3M), 1)
|
||||
|
||||
|
@ -282,12 +282,12 @@ CSBLAS2OBJS = \
|
|||
CSBLAS3OBJS = \
|
||||
cblas_sgemm.$(SUFFIX) cblas_ssymm.$(SUFFIX) cblas_strmm.$(SUFFIX) cblas_strsm.$(SUFFIX) \
|
||||
cblas_ssyrk.$(SUFFIX) cblas_ssyr2k.$(SUFFIX) cblas_somatcopy.$(SUFFIX) cblas_simatcopy.$(SUFFIX)\
|
||||
cblas_sgeadd.$(SUFFIX) cblas_sgemmt.$(SUFFIX) cblas_sgemm_batch.$(SUFFIX)
|
||||
cblas_sgeadd.$(SUFFIX) cblas_sgemmt.$(SUFFIX) cblas_sgemmtr.$(SUFFIX) cblas_sgemm_batch.$(SUFFIX)
|
||||
|
||||
ifeq ($(BUILD_BFLOAT16),1)
|
||||
CSBBLAS1OBJS = cblas_sbdot.$(SUFFIX)
|
||||
CSBBLAS2OBJS = cblas_sbgemv.$(SUFFIX)
|
||||
CSBBLAS3OBJS = cblas_sbgemm.$(SUFFIX) cblas_sbgemmt.$(SUFFIX) cblas_sbgemm_batch.$(SUFFIX)
|
||||
CSBBLAS3OBJS = cblas_sbgemm.$(SUFFIX) cblas_sbgemmt.$(SUFFIX) cblas_sbgemmtr.$(SUFFIX) cblas_sbgemm_batch.$(SUFFIX)
|
||||
CSBEXTOBJS = cblas_sbstobf16.$(SUFFIX) cblas_sbdtobf16.$(SUFFIX) cblas_sbf16tos.$(SUFFIX) cblas_dbf16tod.$(SUFFIX)
|
||||
endif
|
||||
|
||||
|
@ -308,7 +308,7 @@ CDBLAS2OBJS = \
|
|||
CDBLAS3OBJS += \
|
||||
cblas_dgemm.$(SUFFIX) cblas_dsymm.$(SUFFIX) cblas_dtrmm.$(SUFFIX) cblas_dtrsm.$(SUFFIX) \
|
||||
cblas_dsyrk.$(SUFFIX) cblas_dsyr2k.$(SUFFIX) cblas_domatcopy.$(SUFFIX) cblas_dimatcopy.$(SUFFIX) \
|
||||
cblas_dgeadd.$(SUFFIX) cblas_dgemmt.$(SUFFIX) cblas_dgemm_batch.$(SUFFIX)
|
||||
cblas_dgeadd.$(SUFFIX) cblas_dgemmt.$(SUFFIX) cblas_dgemmtr.$(SUFFIX) cblas_dgemm_batch.$(SUFFIX)
|
||||
|
||||
CCBLAS1OBJS = \
|
||||
cblas_icamax.$(SUFFIX) cblas_icamin.$(SUFFIX) cblas_scasum.$(SUFFIX) cblas_caxpy.$(SUFFIX) \
|
||||
|
@ -333,7 +333,7 @@ CCBLAS3OBJS = \
|
|||
cblas_csyrk.$(SUFFIX) cblas_csyr2k.$(SUFFIX) \
|
||||
cblas_chemm.$(SUFFIX) cblas_cherk.$(SUFFIX) cblas_cher2k.$(SUFFIX) \
|
||||
cblas_comatcopy.$(SUFFIX) cblas_cimatcopy.$(SUFFIX)\
|
||||
cblas_cgeadd.$(SUFFIX) cblas_cgemmt.$(SUFFIX) cblas_cgemm_batch.$(SUFFIX)
|
||||
cblas_cgeadd.$(SUFFIX) cblas_cgemmt.$(SUFFIX) cblas_cgemmtr.$(SUFFIX) cblas_cgemm_batch.$(SUFFIX)
|
||||
|
||||
CXERBLAOBJ = \
|
||||
cblas_xerbla.$(SUFFIX)
|
||||
|
@ -364,7 +364,7 @@ CZBLAS3OBJS = \
|
|||
cblas_zsyrk.$(SUFFIX) cblas_zsyr2k.$(SUFFIX) \
|
||||
cblas_zhemm.$(SUFFIX) cblas_zherk.$(SUFFIX) cblas_zher2k.$(SUFFIX)\
|
||||
cblas_zomatcopy.$(SUFFIX) cblas_zimatcopy.$(SUFFIX) \
|
||||
cblas_zgeadd.$(SUFFIX) cblas_zgemmt.$(SUFFIX) cblas_zgemm_batch.$(SUFFIX)
|
||||
cblas_zgeadd.$(SUFFIX) cblas_zgemmt.$(SUFFIX) cblas_zgemmtr.$(SUFFIX) cblas_zgemm_batch.$(SUFFIX)
|
||||
|
||||
|
||||
ifeq ($(SUPPORT_GEMM3M), 1)
|
||||
|
@ -1305,6 +1305,8 @@ sbgemm.$(SUFFIX) sbgemm.$(PSUFFIX) : gemm.c ../param.h
|
|||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
sbgemmt.$(SUFFIX) sbgemmt.$(PSUFFIX) : sbgemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
sbgemmtr.$(SUFFIX) sbgemmtr.$(PSUFFIX) : sbgemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
endif
|
||||
|
||||
sgemm.$(SUFFIX) sgemm.$(PSUFFIX) : gemm.c ../param.h
|
||||
|
@ -1340,7 +1342,19 @@ cgemmt.$(SUFFIX) cgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
|||
zgemmt.$(SUFFIX) zgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
xgemmt.$(SUFFIX) xgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
||||
sgemmtr.$(SUFFIX) sgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
dgemmtr.$(SUFFIX) dgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
qgemmtr.$(SUFFIX) qgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cgemmtr.$(SUFFIX) cgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
zgemmtr.$(SUFFIX) zgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
ssymm.$(SUFFIX) ssymm.$(PSUFFIX) : symm.c
|
||||
|
@ -1966,9 +1980,14 @@ cblas_zgemm.$(SUFFIX) cblas_zgemm.$(PSUFFIX) : gemm.c ../param.h
|
|||
cblas_sgemmt.$(SUFFIX) cblas_sgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cblas_sgemmtr.$(SUFFIX) cblas_sgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
ifeq ($(BUILD_BFLOAT16),1)
|
||||
cblas_sbgemmt.$(SUFFIX) cblas_sbgemmt.$(PSUFFIX) : sbgemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
cblas_sbgemmtr.$(SUFFIX) cblas_sbgemmtr.$(PSUFFIX) : sbgemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
endif
|
||||
|
||||
cblas_dgemmt.$(SUFFIX) cblas_dgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
||||
|
@ -1980,6 +1999,15 @@ cblas_cgemmt.$(SUFFIX) cblas_cgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
|||
cblas_zgemmt.$(SUFFIX) cblas_zgemmt.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cblas_dgemmtr.$(SUFFIX) cblas_dgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cblas_cgemmtr.$(SUFFIX) cblas_cgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cblas_zgemmtr.$(SUFFIX) cblas_zgemmtr.$(PSUFFIX) : gemmt.c ../param.h
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
cblas_ssymm.$(SUFFIX) cblas_ssymm.$(PSUFFIX) : symm.c
|
||||
$(CC) -DCBLAS -c $(CFLAGS) $< -o $(@F)
|
||||
|
||||
|
|
Loading…
Reference in New Issue