| /external/XNNPACK/src/f32-dwconv/gen/ |
| D | up1x4-wasm.c | 60 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x4__wasm() local
61 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x4__wasm()
|
| D | up1x4-scalar.c | 60 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x4__scalar() local
61 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x4__scalar()
|
| D | up1x4-wasm-acc2.c | 60 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x4__wasm_acc2() local
61 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x4__wasm_acc2()
|
| D | up1x4-scalar-acc2.c | 60 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x4__scalar_acc2() local
61 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x4__scalar_acc2()
|
| D | up1x9-scalar.c | 70 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x9__scalar() local
71 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x9__scalar()
|
| D | up1x9-wasm-acc2.c | 70 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x9__wasm_acc2() local
71 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x9__wasm_acc2()
|
| D | up1x9-scalar-acc2.c | 70 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x9__scalar_acc2() local
71 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x9__scalar_acc2()
|
| D | up1x9-wasm.c | 70 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x9__wasm() local
71 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x9__wasm()
|
| D | up2x4-scalar.c | 111 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x4__scalar() local
112 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x4__scalar()
|
| D | up2x4-wasm.c | 111 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x4__wasm() local
112 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x4__wasm()
|
| D | up2x4-wasm-acc2.c | 114 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x4__wasm_acc2() local
115 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x4__wasm_acc2()
|
| D | up2x4-scalar-acc2.c | 114 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x4__scalar_acc2() local
115 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x4__scalar_acc2()
|
| D | up2x9-wasm.c | 166 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x9__wasm() local
167 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x9__wasm()
|
| D | up2x9-scalar.c | 166 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x9__scalar() local
167 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x9__scalar()
|
| D | up2x9-scalar-acc2.c | 169 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x9__scalar_acc2() local
170 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x9__scalar_acc2()
|
| D | up2x9-wasm-acc2.c | 169 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x9__wasm_acc2() local
170 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x9__wasm_acc2()
|
| D | up1x25-wasm-acc2.c | 102 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x25__wasm_acc2() local
103 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x25__wasm_acc2()
|
| D | up1x25-wasm.c | 102 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x25__wasm() local
103 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x25__wasm()
|
| D | up1x25-scalar.c | 102 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x25__scalar() local
103 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x25__scalar()
|
| D | up1x25-scalar-acc2.c | 102 const float vk3 = w[4]; in xnn_f32_dwconv_ukernel_up1x25__scalar_acc2() local
103 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up1x25__scalar_acc2()
|
| D | up2x25-scalar.c | 342 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x25__scalar() local
343 vacc0p0 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x25__scalar()
|
| D | up2x25-scalar-acc2.c | 345 const float vk3 = w[7]; in xnn_f32_dwconv_ukernel_up2x25__scalar_acc2() local
346 vacc0p1 += vi3 * vk3; in xnn_f32_dwconv_ukernel_up2x25__scalar_acc2()
|
| /external/XNNPACK/src/q8-dwconv/ |
| D | up1x9-scalar.c | 63 const uint32_t vk3 = (uint32_t) ((const uint8_t*) w)[7]; in xnn_q8_dwconv_ukernel_up1x9__scalar() local
64 const int32_t vxk3 = (int32_t) vk3 - vkernel_zero_point; in xnn_q8_dwconv_ukernel_up1x9__scalar()
|
| D | up8x9-sse2.c | 75 const __m128i vk3 = _mm_loadl_epi64((const __m128i*) ((uintptr_t) w + 56)); in xnn_q8_dwconv_ukernel_up8x9__sse2() local
76 const __m128i vxk3 = _mm_sub_epi16(_mm_unpacklo_epi8(vk3, vzero), vkernel_zero_point); in xnn_q8_dwconv_ukernel_up8x9__sse2()
227 const __m128i vk3 = _mm_loadl_epi64((const __m128i*) ((uintptr_t) w + 56)); in xnn_q8_dwconv_ukernel_up8x9__sse2() local
228 const __m128i vxk3 = _mm_sub_epi16(_mm_unpacklo_epi8(vk3, vzero), vkernel_zero_point); in xnn_q8_dwconv_ukernel_up8x9__sse2()
|
| /external/XNNPACK/src/f32-dwconv/ |
| D | up4x9-aarch64-neonfma-cortex-a55.S | 102 # Load vk3.lo
108 # Load vk3.hi
189 # vacc.lo += vi3.lo * vk3.lo
197 # vacc.hi += vi3.hi * vk3.hi
410 # Load vk3.lo
418 # Load vk3.hi
518 # vacc.lo += vi3.lo * vk3.lo
526 # vacc.hi += vi3.hi * vk3.hi
|