| /external/XNNPACK/src/f32-ppmm/gen/ |
| D | 4x2-minmax-scalar.c | 49 float vacc0x1 = w[1]; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar() local
51 float vacc1x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
53 float vacc2x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
55 float vacc3x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
74 vacc0x1 += va0 * vb1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
87 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
97 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
110 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
|
| D | 3x3-minmax-scalar.c | 45 float vacc0x1 = w[1]; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar() local
48 float vacc1x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
51 float vacc2x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
70 vacc0x1 += va0 * vb1; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
84 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
95 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
110 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
127 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
|
| D | 2x4-minmax-scalar.c | 41 float vacc0x1 = w[1]; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar() local
45 float vacc1x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
64 vacc0x1 += va0 * vb1; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
77 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
87 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
100 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
115 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
|
| D | 4x4-minmax-scalar.c | 49 float vacc0x1 = w[1]; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar() local
53 float vacc1x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
57 float vacc2x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
61 float vacc3x1 = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
84 vacc0x1 += va0 * vb1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
105 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
123 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
150 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
171 c0[1] = vacc0x1; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
|
| /external/XNNPACK/src/bf16-gemm/gen/ |
| D | 1x4c8-minmax-neonbf16-bfmlal.c | 45 …float32x4_t vacc0x1 = vcvt_f32_bf16(vld1_lane_bf16(w, vreinterpret_bf16_u16(vdup_n_u16(0)), 0)); w… in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal() local
59 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
64 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
85 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
86 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
96 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
102 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
|
| D | 1x4c8-minmax-neonfma-zip.c | 46 …float32x4_t vacc0x1 = vreinterpretq_f32_u32(vshll_n_u16(vld1_lane_u16(w, vdup_n_u16(0), 0), 16)); … in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip() local
67 vacc0x1 = vfmaq_f32(vacc0x1, va0e, vb1e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
79 vacc0x1 = vfmaq_f32(vacc0x1, va0o, vb1o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
112 vacc0x1 = vfmaq_f32(vacc0x1, va0x1e, vb1e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
127 vacc0x1 = vfmaq_f32(vacc0x1, va0x1o, vb1o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
133 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
139 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
|
| D | 2x4c8-minmax-neonbf16-bfmlal.c | 51 …float32x4_t vacc0x1 = vcvt_f32_bf16(vld1_lane_bf16(w, vreinterpret_bf16_u16(vdup_n_u16(0)), 0)); w… in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal() local
55 float32x4_t vacc1x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
71 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
80 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
108 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
109 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
128 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
138 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
|
| D | 1x4c8-minmax-neonfma-shland.c | 46 …float32x4_t vacc0x1 = vreinterpretq_f32_u32(vshll_n_u16(vld1_lane_u16(w, vdup_n_u16(0), 0), 16)); … in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland() local
67 vacc0x1 = vfmaq_f32(vacc0x1, va0e, vb1e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
79 vacc0x1 = vfmaq_f32(vacc0x1, va0o, vb1o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
112 vacc0x1 = vfmaq_f32(vacc0x1, va0x1e, vb1e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
127 vacc0x1 = vfmaq_f32(vacc0x1, va0x1o, vb1o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
133 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
139 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
|
| D | 1x4c8-minmax-neonbf16-bfdot.c | 45 …float32x4_t vacc0x1 = vcvt_f32_bf16(vld1_lane_bf16(w, vreinterpret_bf16_u16(vdup_n_u16(0)), 0)); w… in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot() local
59 vacc0x1 = vbfdotq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
79 vacc0x1 = vbfdotq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
87 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
93 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
|
| D | 3x4c8-minmax-neonbf16-bfmlal.c | 57 …float32x4_t vacc0x1 = vcvt_f32_bf16(vld1_lane_bf16(w, vreinterpret_bf16_u16(vdup_n_u16(0)), 0)); w… in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal() local
61 float32x4_t vacc1x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
65 float32x4_t vacc2x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
83 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
96 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
131 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
132 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
160 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
174 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonbf16_bfmlal()
|
| D | 2x4c8-minmax-neonfma-shland.c | 52 …float32x4_t vacc0x1 = vreinterpretq_f32_u32(vshll_n_u16(vld1_lane_u16(w, vdup_n_u16(0), 0), 16)); … in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland() local
56 float32x4_t vacc1x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
80 vacc0x1 = vfmaq_f32(vacc0x1, va0e, vb1e); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
97 vacc0x1 = vfmaq_f32(vacc0x1, va0o, vb1o); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
143 vacc0x1 = vfmaq_f32(vacc0x1, va0x1e, vb1e); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
166 vacc0x1 = vfmaq_f32(vacc0x1, va0x1o, vb1o); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
175 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
185 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland()
|
| D | 2x4c8-minmax-neonfma-zip.c | 52 …float32x4_t vacc0x1 = vreinterpretq_f32_u32(vshll_n_u16(vld1_lane_u16(w, vdup_n_u16(0), 0), 16)); … in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip() local
56 float32x4_t vacc1x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
80 vacc0x1 = vfmaq_f32(vacc0x1, va0e, vb1e); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
97 vacc0x1 = vfmaq_f32(vacc0x1, va0o, vb1o); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
143 vacc0x1 = vfmaq_f32(vacc0x1, va0x1e, vb1e); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
166 vacc0x1 = vfmaq_f32(vacc0x1, va0x1o, vb1o); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
175 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
185 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip()
|
| D | 4x4c8-minmax-neonbf16-bfmlal.c | 63 …float32x4_t vacc0x1 = vcvt_f32_bf16(vld1_lane_bf16(w, vreinterpret_bf16_u16(vdup_n_u16(0)), 0)); w… in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal() local
67 float32x4_t vacc1x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
71 float32x4_t vacc2x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
75 float32x4_t vacc3x1 = vacc0x1; in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
95 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
112 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0, vb1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
154 vacc0x1 = vbfmlalbq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
155 vacc0x1 = vbfmlaltq_f32(vacc0x1, va0x1, vb1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
192 const float32x4_t vacc0x01 = vpaddq_f32(vacc0x0, vacc0x1); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
210 const float32x2_t vsum0x1 = vadd_f32(vget_low_f32(vacc0x1), vget_high_f32(vacc0x1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonbf16_bfmlal()
|
| /external/XNNPACK/src/f32-vmulcaddc/gen/ |
| D | c2-minmax-wasm-2x.c | 53 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x() local
63 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
68 vacc0x1 = __builtin_wasm_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
73 vacc0x1 = __builtin_wasm_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
78 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
|
| D | c2-minmax-scalar-2x.c | 53 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x() local
63 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
68 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
73 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
78 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
|
| D | c4-minmax-scalar-2x.c | 55 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x() local
71 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
80 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
89 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
98 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
|
| D | c4-minmax-wasm-2x.c | 55 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x() local
71 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
80 vacc0x1 = __builtin_wasm_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
89 vacc0x1 = __builtin_wasm_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
98 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
|
| /external/XNNPACK/src/f32-spmm/gen/ |
| D | 8x2-minmax-scalar.c | 49 float vacc0x1 = *w++; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
50 float vacc1x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
51 float vacc2x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
52 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
53 float vacc4x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
54 float vacc5x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
55 float vacc6x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
56 float vacc7x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
79 vacc0x1 += vi0 * vw1; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
97 float vout0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
[all …]
|
| D | 8x4-minmax-scalar.c | 49 float vacc0x1 = *w++; in xnn_f32_spmm_minmax_ukernel_8x4__scalar() local
50 float vacc1x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
51 float vacc2x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
52 float vacc3x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
53 float vacc4x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
54 float vacc5x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
55 float vacc6x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
56 float vacc7x1 = vacc0x1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
97 vacc0x1 += vi0 * vw1; in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
131 float vout0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_spmm_minmax_ukernel_8x4__scalar()
[all …]
|
| /external/XNNPACK/src/qc8-gemm/gen/ |
| D | 1x2c4-minmax-fp32-armsimd32.c | 44 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qc8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32() local
64 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qc8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
67 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qc8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
73 float vfpacc0x1 = (float) vacc0x1; in xnn_qc8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
|
| D | 2x2c4-minmax-fp32-armsimd32.c | 50 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qc8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32() local
52 int32_t vacc1x1 = vacc0x1; in xnn_qc8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
77 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qc8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
81 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qc8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
88 float vfpacc0x1 = (float) vacc0x1; in xnn_qc8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 1x2c4-minmax-fp32-armsimd32.c | 45 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32() local
65 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
68 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
74 float vfpacc0x1 = (float) vacc0x1; in xnn_qs8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
|
| D | 2x2c4-minmax-fp32-armsimd32.c | 51 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qs8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32() local
53 int32_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
78 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qs8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
82 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qs8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
89 float vfpacc0x1 = (float) vacc0x1; in xnn_qs8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
|
| /external/XNNPACK/src/qu8-gemm/gen/ |
| D | 1x2c4-minmax-fp32-armsimd32.c | 46 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qu8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32() local
66 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qu8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
69 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qu8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
75 float vfpacc0x1 = (float) vacc0x1; in xnn_qu8_gemm_minmax_fp32_ukernel_1x2c4__armsimd32()
|
| D | 2x2c4-minmax-fp32-armsimd32.c | 52 int32_t vacc0x1 = ((const int32_t*) w)[1]; in xnn_qu8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32() local
54 int32_t vacc1x1 = vacc0x1; in xnn_qu8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
79 vacc0x1 = __smlad(va0c02, vb1c02, vacc0x1); in xnn_qu8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
83 vacc0x1 = __smlad(va0c13, vb1c13, vacc0x1); in xnn_qu8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
90 float vfpacc0x1 = (float) vacc0x1; in xnn_qu8_gemm_minmax_fp32_ukernel_2x2c4__armsimd32()
|