floraachy
/
OpenBLAS
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

added DSDOT definition and enabled optimized sdot kernel

This commit is contained in:
wernsaar 2014-06-27 11:30:29 +02:00
parent a6ae079b17
commit 578d1b6219
2 changed files with 22 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
kernel/Makefile.L1
View File

@ -687,15 +687,27 @@ $(KDIR)ddot_k$(TSUFFIX).$(SUFFIX) $(KDIR)ddot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNEL
$(KDIR)qdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)qdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(QDOTKERNEL) $(KDIR)qdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)qdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(QDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -DXDOUBLE $< -o $@ $(CC) -c $(CFLAGS) -UCOMPLEX -DXDOUBLE $< -o $@
$(KDIR)dsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)dsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@
$(KDIR)sdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)sdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL) $(KDIR)sdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)sdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE $< -o $@ $(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE $< -o $@
ifdef DSDOTKERNEL
$(KDIR)dsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)dsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(DSDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@
$(KDIR)sdsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)sdsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(DSDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@
else
$(KDIR)dsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)dsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@
$(KDIR)sdsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)sdsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL) $(KDIR)sdsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)sdsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@ $(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE -DDSDOT $< -o $@
endif
$(KDIR)zdotu_k$(TSUFFIX).$(SUFFIX) $(KDIR)zdotu_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(ZDOTKERNEL) $(KDIR)zdotu_k$(TSUFFIX).$(SUFFIX) $(KDIR)zdotu_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(ZDOTKERNEL)
$(CC) -c $(CFLAGS) -DCOMPLEX -DDOUBLE -UCONJ $< -o $@ $(CC) -c $(CFLAGS) -DCOMPLEX -DDOUBLE -UCONJ $< -o $@

8
kernel/x86_64/KERNEL
View File

@ -119,9 +119,15 @@ XCOPYKERNEL = zcopy.S
endif endif
ifndef SDOTKERNEL ifndef SDOTKERNEL
SDOTKERNEL = ../arm/dot.c SDOTKERNEL = dot_sse.S
endif endif
ifndef DSDOTKERNEL
DSDOTKERNEL = ../arm/dot.c
endif
ifndef DDOTKERNEL ifndef DDOTKERNEL
DDOTKERNEL = dot_sse2.S DDOTKERNEL = dot_sse2.S
endif endif