| /external/XNNPACK/src/f32-ppmm/gen/ |
| D | 3x3-minmax-scalar.c | 44 float vacc0x0 = w[0]; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar() local
47 float vacc1x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
50 float vacc2x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
67 vacc0x0 += va0 * vb0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
81 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
92 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
109 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
126 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
131 vacc0x0 = vacc0x2; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
140 *c0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_3x3__scalar()
|
| D | 2x4-minmax-scalar.c | 40 float vacc0x0 = w[0]; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar() local
44 float vacc1x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
62 vacc0x0 += va0 * vb0; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
75 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
85 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
99 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
114 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
118 vacc0x0 = vacc0x2; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
125 *c0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_2x4__scalar()
|
| D | 4x2-minmax-scalar.c | 48 float vacc0x0 = w[0]; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar() local
50 float vacc1x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
52 float vacc2x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
54 float vacc3x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
70 vacc0x0 += va0 * vb0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
83 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
93 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
109 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
125 *c0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x2__scalar()
|
| D | 4x4-minmax-scalar.c | 48 float vacc0x0 = w[0]; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar() local
52 float vacc1x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
56 float vacc2x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
60 float vacc3x0 = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
80 vacc0x0 += va0 * vb0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
101 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
119 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
149 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
170 c0[0] = vacc0x0; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
176 vacc0x0 = vacc0x2; in xnn_f32_ppmm_minmax_ukernel_4x4__scalar()
[all …]
|
| /external/XNNPACK/src/f32-spmm/gen/ |
| D | 8x1-minmax-scalar.c | 41 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_8x1__scalar() local
42 float vacc1x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
43 float vacc2x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
45 float vacc4x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
46 float vacc5x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
47 float vacc6x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
48 float vacc7x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
62 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
72 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x1__scalar()
[all …]
|
| D | 4x1-minmax-scalar.c | 41 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_4x1__scalar() local
42 float vacc1x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
43 float vacc2x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
54 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
60 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
130 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_4x1__scalar() local
131 float vacc1x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
139 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
143 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_4x1__scalar()
[all …]
|
| D | 8x2-minmax-scalar.c | 41 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_8x2__scalar() local
42 float vacc1x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
43 float vacc2x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
44 float vacc3x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
45 float vacc4x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
46 float vacc5x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
47 float vacc6x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
48 float vacc7x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
71 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
89 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_8x2__scalar()
[all …]
|
| D | 2x1-minmax-scalar.c | 41 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_2x1__scalar() local
42 float vacc1x0 = vacc0x0; in xnn_f32_spmm_minmax_ukernel_2x1__scalar()
50 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_2x1__scalar()
54 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_2x1__scalar()
104 float vacc0x0 = *w++; in xnn_f32_spmm_minmax_ukernel_2x1__scalar() local
111 vacc0x0 += vi0 * vw0; in xnn_f32_spmm_minmax_ukernel_2x1__scalar()
114 float vout0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_spmm_minmax_ukernel_2x1__scalar()
|
| /external/XNNPACK/src/bf16-gemm/gen/ |
| D | 1x4c8-minmax-neonbf16-bfmlal.c | 44 …float32x4_t vacc0x0 = 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
58 vacc0x0 = vbfmlalbq_f32(vacc0x0, va0, vb0); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
63 vacc0x0 = vbfmlaltq_f32(vacc0x0, va0, vb0); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
82 vacc0x0 = vbfmlalbq_f32(vacc0x0, va0x0, vb0); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
83 vacc0x0 = vbfmlaltq_f32(vacc0x0, va0x0, vb0); 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()
101 const float32x2_t vsum0x0 = vadd_f32(vget_low_f32(vacc0x0), vget_high_f32(vacc0x0)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfmlal()
|
| D | 1x4c8-minmax-neonfma-zip.c | 45 …float32x4_t vacc0x0 = 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
66 vacc0x0 = vfmaq_f32(vacc0x0, va0e, vb0e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
78 vacc0x0 = vfmaq_f32(vacc0x0, va0o, vb0o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
111 vacc0x0 = vfmaq_f32(vacc0x0, va0x0e, vb0e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
126 vacc0x0 = vfmaq_f32(vacc0x0, va0x0o, vb0o); 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()
138 const float32x2_t vsum0x0 = vadd_f32(vget_low_f32(vacc0x0), vget_high_f32(vacc0x0)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip()
|
| D | 2x4c8-minmax-neonbf16-bfmlal.c | 50 …float32x4_t vacc0x0 = 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
54 float32x4_t vacc1x0 = vacc0x0; in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
69 vacc0x0 = vbfmlalbq_f32(vacc0x0, va0, vb0); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
78 vacc0x0 = vbfmlaltq_f32(vacc0x0, va0, vb0); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
102 vacc0x0 = vbfmlalbq_f32(vacc0x0, va0x0, vb0); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
103 vacc0x0 = vbfmlaltq_f32(vacc0x0, va0x0, vb0); 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()
136 const float32x2_t vsum0x0 = vadd_f32(vget_low_f32(vacc0x0), vget_high_f32(vacc0x0)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonbf16_bfmlal()
|
| D | 1x4c8-minmax-neonfma-shland.c | 45 …float32x4_t vacc0x0 = 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
66 vacc0x0 = vfmaq_f32(vacc0x0, va0e, vb0e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
78 vacc0x0 = vfmaq_f32(vacc0x0, va0o, vb0o); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
111 vacc0x0 = vfmaq_f32(vacc0x0, va0x0e, vb0e); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
126 vacc0x0 = vfmaq_f32(vacc0x0, va0x0o, vb0o); 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()
138 const float32x2_t vsum0x0 = vadd_f32(vget_low_f32(vacc0x0), vget_high_f32(vacc0x0)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland()
|
| D | 1x4c8-minmax-neonbf16-bfdot.c | 44 …float32x4_t vacc0x0 = 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
58 vacc0x0 = vbfdotq_f32(vacc0x0, va0, vb0); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
77 vacc0x0 = vbfdotq_f32(vacc0x0, va0x0, vb0); 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()
92 const float32x2_t vsum0x0 = vadd_f32(vget_low_f32(vacc0x0), vget_high_f32(vacc0x0)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonbf16_bfdot()
|
| /external/XNNPACK/src/f32-vmulcaddc/gen/ |
| D | c2-minmax-wasm-2x.c | 52 float vacc0x0 = i0[0]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x() local
62 vacc0x0 = vacc0x0 * vscale0 + vbias0; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
67 vacc0x0 = __builtin_wasm_max_f32(vacc0x0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
72 vacc0x0 = __builtin_wasm_min_f32(vacc0x0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
77 o0[0] = vacc0x0; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
|
| D | c2-minmax-scalar-2x.c | 52 float vacc0x0 = i0[0]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x() local
62 vacc0x0 = vacc0x0 * vscale0 + vbias0; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
67 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
72 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
77 o0[0] = vacc0x0; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
|
| D | c4-minmax-scalar-2x.c | 54 float vacc0x0 = i0[0]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x() local
70 vacc0x0 = vacc0x0 * vscale0 + vbias0; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
79 vacc0x0 = math_max_f32(vacc0x0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
88 vacc0x0 = math_min_f32(vacc0x0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
97 o0[0] = vacc0x0; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
|
| D | c4-minmax-wasm-2x.c | 54 float vacc0x0 = i0[0]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x() local
70 vacc0x0 = vacc0x0 * vscale0 + vbias0; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
79 vacc0x0 = __builtin_wasm_max_f32(vacc0x0, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
88 vacc0x0 = __builtin_wasm_min_f32(vacc0x0, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
97 o0[0] = vacc0x0; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
|
| /external/XNNPACK/src/qs8-gemm/gen/ |
| D | 1x1c4-minmax-fp32-armsimd32.c | 44 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32() local
57 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
60 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
65 float vfpacc0x0 = (float) vacc0x0; in xnn_qs8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
|
| D | 2x1c4-minmax-fp32-armsimd32.c | 50 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qs8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32() local
51 int32_t vacc1x0 = vacc0x0; in xnn_qs8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
67 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
71 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qs8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
77 float vfpacc0x0 = (float) vacc0x0; in xnn_qs8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
|
| /external/XNNPACK/src/qc8-gemm/gen/ |
| D | 1x1c4-minmax-fp32-armsimd32.c | 43 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qc8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32() local
56 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qc8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
59 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qc8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
64 float vfpacc0x0 = (float) vacc0x0; in xnn_qc8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
|
| D | 2x1c4-minmax-fp32-armsimd32.c | 49 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qc8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32() local
50 int32_t vacc1x0 = vacc0x0; in xnn_qc8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
66 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
70 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qc8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
76 float vfpacc0x0 = (float) vacc0x0; in xnn_qc8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
|
| /external/XNNPACK/src/qu8-gemm/gen/ |
| D | 1x1c4-minmax-fp32-armsimd32.c | 45 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qu8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32() local
58 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qu8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
61 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qu8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
66 float vfpacc0x0 = (float) vacc0x0; in xnn_qu8_gemm_minmax_fp32_ukernel_1x1c4__armsimd32()
|
| D | 2x1c4-minmax-fp32-armsimd32.c | 51 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qu8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32() local
52 int32_t vacc1x0 = vacc0x0; in xnn_qu8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
68 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qu8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
72 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qu8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
78 float vfpacc0x0 = (float) vacc0x0; in xnn_qu8_gemm_minmax_fp32_ukernel_2x1c4__armsimd32()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 1x1c4-minmax-fp32-armsimd32.c | 50 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qs8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32() local
72 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qs8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
75 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qs8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
82 float vfpacc0x0 = (float) vacc0x0; in xnn_qs8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
|
| /external/XNNPACK/src/qc8-igemm/gen/ |
| D | 1x1c4-minmax-fp32-armsimd32.c | 49 int32_t vacc0x0 = ((const int32_t*) w)[0]; in xnn_qc8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32() local
71 vacc0x0 = __smlad(va0c02, vb0c02, vacc0x0); in xnn_qc8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
74 vacc0x0 = __smlad(va0c13, vb0c13, vacc0x0); in xnn_qc8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
81 float vfpacc0x0 = (float) vacc0x0; in xnn_qc8_igemm_minmax_fp32_ukernel_1x1c4__armsimd32()
|