43 #define SCALE_COL_T(T,CAST,A,ROW,v,i)        \  argument
54      *data *= CAST v;                   \
96 #define SCALE_ROW_F16(A,COL,v,i)                   \  argument
108        tmpa = vmulq_n_f16(tmpa,(_Float16)v);                  \
115 #define MAC_ROW_F16(COL,A,i,v,B,j)                   \  argument
129        tmpa = vfmaq_n_f16(tmpa,tmpb,v);              \
136 #define MAS_ROW_F16(COL,A,i,v,B,j)                   \  argument
143   f16x8_t vec=vdupq_n_f16(v);                        \
181 #define SCALE_ROW_F16(A,COL,v,i)       \
192      *data++ *= (_Float16)v;           \
197 #define MAC_ROW_F16(COL,A,i,v,B,j)                \
210      *dataA++ += (_Float16)v * (_Float16)*dataB++;\
214 #define MAS_ROW_F16(COL,A,i,v,B,j)                \
227      *dataA++ -= (_Float16)v * (_Float16)*dataB++;\
237 #define SCALE_COL_F16(A,ROW,v,i)        \  argument
238   SCALE_COL_T(float16_t,(_Float16),A,ROW,v,i)
266 #define MAC_ROW_F32(COL,A,i,v,B,j)                   \  argument
280        tmpa = vfmaq_n_f32(tmpa,tmpb,v);              \
287 #define MAS_ROW_F32(COL,A,i,v,B,j)                   \  argument
294   f32x4_t vec=vdupq_n_f32(v);                        \
309 #define SCALE_ROW_F32(A,COL,v,i)                    \  argument
321        tmpa = vmulq_n_f32(tmpa,v);                  \
351 #define MAC_ROW_F32(COL,A,i,v,B,j)     \
358   f32x4_t vec = vdupq_n_f32(v);        \
380      *dataA++ += v* *dataB++;          \
385 #define MAS_ROW_F32(COL,A,i,v,B,j)     \
392   f32x4_t vec = vdupq_n_f32(v);        \
414      *dataA++ -= v* *dataB++;          \
419 #define SCALE_ROW_F32(A,COL,v,i)        \
425   f32x4_t vec = vdupq_n_f32(v);         \
443      *data++ *= v;                      \
472 #define SCALE_ROW_F32(A,COL,v,i)       \
483      *data++ *= v;                     \
488 #define MAC_ROW_F32(COL,A,i,v,B,j)     \
501      *dataA++ += v* *dataB++;          \
505 #define MAS_ROW_F32(COL,A,i,v,B,j)     \
518      *dataA++ -= v* *dataB++;          \
547 #define SCALE_COL_F32(A,ROW,v,i)        \  argument
548   SCALE_COL_T(float32_t,,A,ROW,v,i)
584 #define SCALE_ROW_F64(A,COL,v,i)       \  argument
595      *data++ *= v;                     \
599 #define SCALE_COL_F64(A,ROW,v,i)        \  argument
600   SCALE_COL_T(float64_t,,A,ROW,v,i)
602 #define MAC_ROW_F64(COL,A,i,v,B,j)      \  argument
615      *dataA++ += v* *dataB++;           \
619 #define MAS_ROW_F64(COL,A,i,v,B,j)      \  argument
632      *dataA++ -= v* *dataB++;           \