| /external/XNNPACK/src/f32-dwconv/gen/ |
| D | up8x3-wasmrelaxedsimd-fma.c | 71 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_ukernel_up8x3__wasmrelaxedsimd_fma() local
73 vacc4567p0 = __builtin_wasm_fma_f32x4(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_ukernel_up8x3__wasmrelaxedsimd_fma()
|
| D | up8x3-wasmsimd.c | 71 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_ukernel_up8x3__wasmsimd() local
73 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_ukernel_up8x3__wasmsimd()
|
| D | up8x3-minmax-wasmrelaxedsimd-fma.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_fma() local
75 vacc4567p0 = __builtin_wasm_fma_f32x4(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_fma()
|
| D | up8x3-minmax-wasmsimd-x86-acc2.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_x86_acc2() local
75 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_x86_acc2()
|
| D | up8x3-minmax-wasmrelaxedsimd-fma-acc2.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_fma_acc2() local
75 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_fma_acc2()
|
| D | up8x3-minmax-sse.c | 73 const __m128 vk1x4567 = _mm_load_ps(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__sse() local
75 vacc4567p0 = _mm_add_ps(vacc4567p0, _mm_mul_ps(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x3__sse()
|
| D | up8x3-minmax-neon.c | 70 const float32x4_t vk1x4567 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x3__neon() local
72 vacc4567p0 = vmlaq_f32(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__neon()
|
| D | up8x3-minmax-sse-acc2.c | 73 const __m128 vk1x4567 = _mm_load_ps(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__sse_acc2() local
75 __m128 vacc4567p1 = _mm_mul_ps(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__sse_acc2()
|
| D | up8x3-minmax-neonfma.c | 70 const float32x4_t vk1x4567 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x3__neonfma() local
72 vacc4567p0 = vfmaq_f32(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__neonfma()
|
| D | up8x3-minmax-wasmsimd-x86.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_x86() local
75 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_x86()
|
| D | up8x3-minmax-neonfma-acc2.c | 70 const float32x4_t vk1x4567 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x3__neonfma_acc2() local
72 float32x4_t vacc4567p1 = vmulq_f32(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__neonfma_acc2()
|
| D | up8x3-minmax-wasmsimd-arm-acc2.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_arm_acc2() local
75 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_arm_acc2()
|
| D | up8x3-minmax-wasmrelaxedsimd.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd() local
75 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd()
|
| D | up8x3-minmax-neon-acc2.c | 70 const float32x4_t vk1x4567 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x3__neon_acc2() local
72 float32x4_t vacc4567p1 = vmulq_f32(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__neon_acc2()
|
| D | up8x3-minmax-wasmrelaxedsimd-acc2.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_acc2() local
75 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmrelaxedsimd_acc2()
|
| D | up8x4-wasmrelaxedsimd-fma.c | 76 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_ukernel_up8x4__wasmrelaxedsimd_fma() local
78 vacc4567p0 = __builtin_wasm_fma_f32x4(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_ukernel_up8x4__wasmrelaxedsimd_fma()
|
| D | up8x3-minmax-wasmsimd-arm.c | 73 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_arm() local
75 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x3__wasmsimd_arm()
|
| D | up8x4-minmax-wasmrelaxedsimd-fma.c | 78 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd_fma() local
80 vacc4567p0 = __builtin_wasm_fma_f32x4(vacc4567p0, vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd_fma()
|
| D | up8x4-wasmsimd.c | 76 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd() local
78 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_ukernel_up8x4__wasmsimd()
|
| D | up8x4-minmax-wasmrelaxedsimd-acc2.c | 78 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd_acc2() local
80 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd_acc2()
|
| D | up8x4-minmax-wasmsimd-x86-acc2.c | 78 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2() local
80 v128_t vacc4567p1 = wasm_f32x4_mul(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86_acc2()
|
| D | up8x4-minmax-wasmsimd-x86.c | 78 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86() local
80 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmsimd_x86()
|
| D | up8x4-minmax-neonfma-acc2.c | 75 const float32x4_t vk1x4567 = vld1q_f32(w); w += 4; in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2() local
77 float32x4_t vacc4567p1 = vmulq_f32(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x4__neonfma_acc2()
|
| D | up8x4-minmax-wasmrelaxedsimd.c | 78 const v128_t vk1x4567 = wasm_v128_load(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd() local
80 vacc4567p0 = wasm_f32x4_add(vacc4567p0, wasm_f32x4_mul(vi1x4567, vk1x4567)); in xnn_f32_dwconv_minmax_ukernel_up8x4__wasmrelaxedsimd()
|
| D | up8x4-minmax-sse-acc2.c | 78 const __m128 vk1x4567 = _mm_load_ps(w + 20); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2() local
80 __m128 vacc4567p1 = _mm_mul_ps(vi1x4567, vk1x4567); in xnn_f32_dwconv_minmax_ukernel_up8x4__sse_acc2()
|