| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 3x16c8-minmax-neon-mlal-padal.c | 285 int16x8_t vprod2x15 = vmull_s8(vb15x0, va2x0); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local
288 vprod2x15 = vmlal_s8(vprod2x15, vb15x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
291 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
410 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal() local
413 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
| D | 3x16c16-minmax-neon-mlal-padal.c | 266 int16x8_t vprod2x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal() local
269 vprod2x15 = vmlal_s8(vprod2x15, vget_high_s8(vb15), vget_high_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
272 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 354 int16x8_t vprod2x15 = vmull_s8(vb15x0, va2x0); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
358 vprod2x15 = vmlal_s8(vprod2x15, vb15x1, va2x1); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
362 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
513 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
517 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 334 int16x8_t vprod2x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
338 vprod2x15 = vmlal_s8(vprod2x15, vget_high_s8(vb15), vget_high_s8(va2)); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
342 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
| D | 3x16c8-minmax-neon-mull-padal.c | 221 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal() local
224 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 274 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal() local
278 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_gemm_minmax_ukernel_4x16c8__neon_mull_padal()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 3x16c8-minmax-neon-mlal-padal.c | 300 int16x8_t vprod2x15 = vmull_s8(vb15x0, va2x0); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local
303 vprod2x15 = vmlal_s8(vprod2x15, vb15x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
306 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
425 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal() local
428 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mlal_padal()
|
| D | 3x16c16-minmax-neon-mlal-padal.c | 281 int16x8_t vprod2x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal() local
284 vprod2x15 = vmlal_s8(vprod2x15, vget_high_s8(vb15), vget_high_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
287 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c16__neon_mlal_padal()
|
| D | 4x16c8-minmax-neon-mlal-padal.c | 371 int16x8_t vprod2x15 = vmull_s8(vb15x0, va2x0); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
375 vprod2x15 = vmlal_s8(vprod2x15, vb15x1, va2x1); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
379 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
530 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal() local
534 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mlal_padal()
|
| D | 4x16c16-minmax-neon-mlal-padal.c | 351 int16x8_t vprod2x15 = vmull_s8(vget_low_s8(vb15), vget_low_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal() local
355 vprod2x15 = vmlal_s8(vprod2x15, vget_high_s8(vb15), vget_high_s8(va2)); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
359 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c16__neon_mlal_padal()
|
| D | 3x16c8-minmax-neon-mull-padal.c | 236 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal() local
239 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_3x16c8__neon_mull_padal()
|
| D | 4x16c8-minmax-neon-mull-padal.c | 291 const int16x8_t vprod2x15 = vmull_s8(vb15, va2); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal() local
295 vacc2x15 = vpadalq_s16(vacc2x15, vprod2x15); in xnn_qs8_igemm_minmax_ukernel_4x16c8__neon_mull_padal()
|