lavu/floatdsp: RISC-V V vector_dmac_scalar
This commit is contained in:
		
							parent
							
								
									c3db27ba95
								
							
						
					
					
						commit
						d120ab5b91
					
				| @ -34,6 +34,8 @@ void ff_vector_fmul_scalar_rvv(float *dst, const float *src, float mul, | |||||||
| 
 | 
 | ||||||
| void ff_vector_dmul_rvv(double *dst, const double *src0, const double *src1, | void ff_vector_dmul_rvv(double *dst, const double *src0, const double *src1, | ||||||
|                          int len); |                          int len); | ||||||
|  | void ff_vector_dmac_scalar_rvv(double *dst, const double *src, double mul, | ||||||
|  |                                 int len); | ||||||
| void ff_vector_dmul_scalar_rvv(double *dst, const double *src, double mul, | void ff_vector_dmul_scalar_rvv(double *dst, const double *src, double mul, | ||||||
|                                 int len); |                                 int len); | ||||||
| 
 | 
 | ||||||
| @ -50,6 +52,7 @@ av_cold void ff_float_dsp_init_riscv(AVFloatDSPContext *fdsp) | |||||||
| 
 | 
 | ||||||
|     if (flags & AV_CPU_FLAG_RVV_F64) { |     if (flags & AV_CPU_FLAG_RVV_F64) { | ||||||
|         fdsp->vector_dmul = ff_vector_dmul_rvv; |         fdsp->vector_dmul = ff_vector_dmul_rvv; | ||||||
|  |         fdsp->vector_dmac_scalar = ff_vector_dmac_scalar_rvv; | ||||||
|         fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_rvv; |         fdsp->vector_dmul_scalar = ff_vector_dmul_scalar_rvv; | ||||||
|     } |     } | ||||||
| #endif | #endif | ||||||
|  | |||||||
| @ -91,6 +91,24 @@ func ff_vector_dmul_rvv, zve64d | |||||||
|         ret |         ret | ||||||
| endfunc | endfunc | ||||||
| 
 | 
 | ||||||
|  | // (a0) += (a1) * fa0 [0..a2-1] | ||||||
|  | func ff_vector_dmac_scalar_rvv, zve64d | ||||||
|  | NOHWD   fmv.d.x   fa0, a2 | ||||||
|  | NOHWD   mv        a2, a3 | ||||||
|  | 1: | ||||||
|  |         vsetvli   t0, a2, e64, m1, ta, ma | ||||||
|  |         vle64.v   v24, (a1) | ||||||
|  |         sub       a2, a2, t0 | ||||||
|  |         vle64.v   v16, (a0) | ||||||
|  |         sh3add    a1, t0, a1 | ||||||
|  |         vfmacc.vf v16, fa0, v24 | ||||||
|  |         vse64.v   v16, (a0) | ||||||
|  |         sh3add    a0, t0, a0 | ||||||
|  |         bnez      a2, 1b | ||||||
|  | 
 | ||||||
|  |         ret | ||||||
|  | endfunc | ||||||
|  | 
 | ||||||
| // (a0) = (a1) * fa0 [0..a2-1] | // (a0) = (a1) * fa0 [0..a2-1] | ||||||
| func ff_vector_dmul_scalar_rvv, zve64d | func ff_vector_dmul_scalar_rvv, zve64d | ||||||
| NOHWD   fmv.d.x  fa0, a2 | NOHWD   fmv.d.x  fa0, a2 | ||||||
|  | |||||||
		Loading…
	
	
			
			x
			
			
		
	
		Reference in New Issue
	
	Block a user