Fixed #28. Convert the result to double precision in the end of dsdot kernel.
This commit is contained in:
parent
1d60510959
commit
b206fc7075
|
@ -19,7 +19,8 @@ common:
|
||||||
* Fixed #25 a wrong result of rotmg.
|
* Fixed #25 a wrong result of rotmg.
|
||||||
|
|
||||||
x86/x86_64:
|
x86/x86_64:
|
||||||
*
|
* Fixed #28 a wrong result of dsdot on x86_64.
|
||||||
|
|
||||||
MIPS64:
|
MIPS64:
|
||||||
*
|
*
|
||||||
====================================================================
|
====================================================================
|
||||||
|
|
|
@ -668,7 +668,7 @@ $(KDIR)qdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)qdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNEL
|
||||||
$(CC) -c $(CFLAGS) -UCOMPLEX -DXDOUBLE $< -o $@
|
$(CC) -c $(CFLAGS) -UCOMPLEX -DXDOUBLE $< -o $@
|
||||||
|
|
||||||
$(KDIR)dsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)dsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
|
$(KDIR)dsdot_k$(TSUFFIX).$(SUFFIX) $(KDIR)dsdot_k$(TPSUFFIX).$(PSUFFIX) : $(KERNELDIR)/$(SDOTKERNEL)
|
||||||
$(CC) -c $(CFLAGS) -UCOMPLEX -UDOUBLE $< -o $@
|
$(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 $@
|
||||||
|
|
|
@ -1286,6 +1286,10 @@
|
||||||
haddps %xmm0, %xmm0
|
haddps %xmm0, %xmm0
|
||||||
#endif
|
#endif
|
||||||
|
|
||||||
|
#ifdef DSDOT
|
||||||
|
cvtss2sd %xmm0, %xmm0
|
||||||
|
#endif
|
||||||
|
|
||||||
RESTOREREGISTERS
|
RESTOREREGISTERS
|
||||||
|
|
||||||
ret
|
ret
|
||||||
|
|
Loading…
Reference in New Issue