/external/XNNPACK/src/f32-dwconv/gen/ |
D | up8x25-minmax-fma3-acc2.c | 257 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3_acc2() local 258 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3_acc2() 395 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3_acc2() local 396 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3_acc2()
|
D | up8x25-minmax-fma3.c | 257 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3() local 258 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3() 393 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3() local 394 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up8x25__fma3()
|
D | up8x25-minmax-avx.c | 257 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx() local 258 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx() 393 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx() local 394 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx()
|
D | up16x25-minmax-fma3.c | 301 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3() local 303 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3() 498 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3() local 499 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3() 634 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3() local 635 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3()
|
D | up16x25-minmax-fma3-acc2.c | 301 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2() local 303 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2() 501 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2() local 502 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2() 639 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2() local 640 vacc01234567p0 = _mm256_fmadd_ps(vi14x01234567, vk14x01234567, vacc01234567p0); in xnn_f32_dwconv_minmax_ukernel_up16x25__fma3_acc2()
|
D | up16x25-minmax-avx.c | 301 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx() local 303 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx() 498 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx() local 499 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx() 634 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx() local 635 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx()
|
D | up8x25-minmax-avx-acc2.c | 257 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx_acc2() local 258 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx_acc2() 395 const __m256 vk14x01234567 = _mm256_load_ps(w + 120); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx_acc2() local 396 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up8x25__avx_acc2()
|
D | up16x25-minmax-avx-acc2.c | 301 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2() local 303 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2() 501 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2() local 502 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2() 639 const __m256 vk14x01234567 = _mm256_load_ps(w + 240); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2() local 640 vacc01234567p0 = _mm256_add_ps(vacc01234567p0, _mm256_mul_ps(vi14x01234567, vk14x01234567)); in xnn_f32_dwconv_minmax_ukernel_up16x25__avx_acc2()
|
/external/XNNPACK/src/f16-dwconv/gen/ |
D | up8x25-minmax-neonfp16arith-acc2.c | 227 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith_acc2() local 228 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith_acc2() 339 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith_acc2() local 340 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith_acc2()
|
D | up8x25-minmax-neonfp16arith.c | 227 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith() local 228 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith() 337 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith() local 338 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up8x25__neonfp16arith()
|
D | up16x25-minmax-neonfp16arith-acc2.c | 271 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2() local 273 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2() 419 const float16x8_t vk14x01234567 = vld1q_f16(w + 232); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2() local 420 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2() 531 const float16x8_t vk14x01234567 = vld1q_f16(w + 240); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2() local 532 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith_acc2()
|
D | up16x25-minmax-neonfp16arith.c | 271 const float16x8_t vk14x01234567 = vld1q_f16(w); w += 8; in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith() local 273 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith() 416 const float16x8_t vk14x01234567 = vld1q_f16(w + 232); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith() local 417 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith() 526 const float16x8_t vk14x01234567 = vld1q_f16(w + 240); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith() local 527 vacc01234567p0 = vfmaq_f16(vacc01234567p0, vi14x01234567, vk14x01234567); in xnn_f16_dwconv_minmax_ukernel_up16x25__neonfp16arith()
|