/external/XNNPACK/src/qc8-gemm/gen/ |
D | 1x16c4-minmax-fp32-neondot.c | 103 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16c4__neondot() local 112 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16c4__neondot() 117 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16c4__neondot()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 229 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 238 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 243 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
D | 1x16-minmax-fp32-neon-mlal-lane-prfm.c | 230 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm() local 239 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm() 245 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane-prfm.c | 231 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm() local 240 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm() 245 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm()
|
D | 1x16-minmax-fp32-neon-mlal-lane.c | 228 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 237 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 243 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qc8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|
D | 4x16c4-minmax-fp32-neondot.c | 175 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_4x16c4__neondot() local 205 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_4x16c4__neondot() 213 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_gemm_minmax_fp32_ukernel_4x16c4__neondot()
|
/external/XNNPACK/src/qc8-igemm/gen/ |
D | 1x16c4-minmax-fp32-neondot.c | 113 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16c4__neondot() local 122 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16c4__neondot() 127 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16c4__neondot()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane-prfm.c | 245 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm() local 254 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm() 259 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane_prfm()
|
D | 1x16-minmax-fp32-neon-mlal-lane.c | 242 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 251 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 257 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|
D | 1x16-minmax-fp32-neon-mlal-lane-prfm.c | 244 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm() local 253 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm() 259 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane_prfm()
|
D | 4x16c4-minmax-fp32-neondot.c | 191 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_4x16c4__neondot() local 221 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_4x16c4__neondot() 229 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_4x16c4__neondot()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 243 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 252 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscaleCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 257 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qc8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
/external/XNNPACK/src/qu8-igemm/gen/ |
D | 1x16c4-minmax-fp32-neondot.c | 128 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16c4__neondot() local 134 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16c4__neondot() 139 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16c4__neondot()
|
D | 2x16c4-minmax-fp32-neondot.c | 163 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_2x16c4__neondot() local 173 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_igemm_minmax_fp32_ukernel_2x16c4__neondot() 182 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_2x16c4__neondot()
|
D | 1x16-minmax-fp32-neon-mlal-lane.c | 243 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 249 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 255 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 244 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 250 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 255 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
D | 4x16c4-minmax-fp32-neondot.c | 233 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_4x16c4__neondot() local 251 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_igemm_minmax_fp32_ukernel_4x16c4__neondot() 268 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_igemm_minmax_fp32_ukernel_4x16c4__neondot()
|
/external/XNNPACK/src/qu8-gemm/gen/ |
D | 1x16c4-minmax-fp32-neondot.c | 116 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16c4__neondot() local 122 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16c4__neondot() 127 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16c4__neondot()
|
D | 2x16c4-minmax-fp32-neondot.c | 149 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_2x16c4__neondot() local 159 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_gemm_minmax_fp32_ukernel_2x16c4__neondot() 168 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_2x16c4__neondot()
|
D | 1x16-minmax-fp32-neon-mlal-lane.c | 229 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 235 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 241 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 230 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 236 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 241 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qu8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
/external/XNNPACK/src/qs8-gemm/gen/ |
D | 1x16-minmax-fp32-neon-mlal-lane.c | 228 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 234 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 240 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 229 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 235 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 240 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qs8_gemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
/external/XNNPACK/src/qs8-igemm/gen/ |
D | 1x16-minmax-fp32-neonv8-mlal-lane.c | 243 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() local 249 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane() 254 vacc0xCDEF = vcvtnq_s32_f32(vfpacc0xCDEF); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neonv8_mlal_lane()
|
D | 1x16-minmax-fp32-neon-mlal-lane.c | 242 float32x4_t vfpacc0xCDEF = vcvtq_f32_s32(vacc0xCDEF); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() local 248 vfpacc0xCDEF = vmulq_f32(vfpacc0xCDEF, vscale); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane() 254 vacc0xCDEF = vreinterpretq_s32_f32(vaddq_f32(vfpacc0xCDEF, vmagic_bias)); in xnn_qs8_igemm_minmax_fp32_ukernel_1x16__neon_mlal_lane()
|