| /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/f32-vmulcaddc/gen/ |
| D | c2-minmax-wasm-2x.c | 52 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x() local
62 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
67 vacc0x1 = __builtin_wasm_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
72 vacc0x1 = __builtin_wasm_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
77 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__wasm_2x()
|
| D | c2-minmax-scalar-2x.c | 52 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x() local
62 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
67 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
72 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
77 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c2__scalar_2x()
|
| D | c4-minmax-wasm-2x.c | 54 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x() local
70 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
79 vacc0x1 = __builtin_wasm_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
88 vacc0x1 = __builtin_wasm_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
97 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__wasm_2x()
|
| D | c4-minmax-scalar-2x.c | 54 float vacc0x1 = i0[1]; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x() local
70 vacc0x1 = vacc0x1 * vscale1 + vbias1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
79 vacc0x1 = math_max_f32(vacc0x1, vmin); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
88 vacc0x1 = math_min_f32(vacc0x1, vmax); in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_2x()
97 o0[1] = vacc0x1; in xnn_f32_vmulcaddc_minmax_ukernel_c4__scalar_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/qs8-gemm/gen/ |
| D | 1x4c8-minmax-wasmsimd-ld64.c | 46 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64() local
64 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
65 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
82 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld128.c | 46 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128() local
64 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
67 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
86 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
|
| D | 1x4c8-xw-minmax-wasmsimd.c | 46 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd() local
64 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
65 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
82 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_xw_minmax_ukernel_1x4c8__wasmsimd()
|
| D | 2x4c8-xw-minmax-wasmsimd.c | 52 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd() local
56 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
79 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
80 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
106 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_xw_minmax_ukernel_2x4c8__wasmsimd()
|
| D | 2x4c8-minmax-wasmsimd-ld128.c | 52 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128() local
56 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
78 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
84 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
110 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
| D | 2x4c8-minmax-wasmsimd-ld64.c | 52 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64() local
56 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
79 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
80 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
106 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
| D | 3x4c8-xw-minmax-wasmsimd.c | 58 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd() local
62 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
66 v128_t vacc2x1 = vacc0x1; in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
94 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
95 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
130 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_xw_minmax_ukernel_3x4c8__wasmsimd()
|
| D | 3x4c8-minmax-wasmsimd-ld128.c | 58 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128() local
62 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
66 v128_t vacc2x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
92 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
101 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
134 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
|
| D | 3x4c8-minmax-wasmsimd-ld64.c | 58 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64() local
62 v128_t vacc1x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
66 v128_t vacc2x1 = vacc0x1; in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
94 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
95 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
130 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_gemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
|
| /external/XNNPACK/src/qs8-igemm/gen/ |
| D | 1x4c8-minmax-wasmsimd-ld64.c | 49 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64() local
75 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
76 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
95 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld64()
|
| D | 1x4c8-minmax-wasmsimd-ld128.c | 49 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128() local
75 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
78 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
99 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_1x4c8__wasmsimd_ld128()
|
| D | 2x4c8-minmax-wasmsimd-ld128.c | 53 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128() local
57 v128_t vacc1x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
91 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
97 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
125 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld128()
|
| D | 2x4c8-minmax-wasmsimd-ld64.c | 53 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64() local
57 v128_t vacc1x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
92 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
93 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
121 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_2x4c8__wasmsimd_ld64()
|
| D | 3x4c8-minmax-wasmsimd-ld128.c | 57 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128() local
61 v128_t vacc1x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
65 v128_t vacc2x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
107 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
116 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
151 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld128()
|
| D | 3x4c8-minmax-wasmsimd-ld64.c | 57 v128_t vacc0x1 = wasm_f32x4_replace_lane(vzero, 0, ((const float*) w)[1]); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64() local
61 v128_t vacc1x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
65 v128_t vacc2x1 = vacc0x1; in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
109 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_low_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
110 vacc0x1 = wasm_i32x4_add(vacc0x1, wasm_i32x4_widen_high_i16x8(vprod0x1)); in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
147 …_t vacc0x13 = wasm_i32x4_add(wasm_v32x4_shuffle(vacc0x1, vacc0x3, 0, 4, 1, 5), wasm_v32x4_shuffle(… in xnn_qs8_igemm_minmax_ukernel_3x4c8__wasmsimd_ld64()
|