/external/XNNPACK/src/f32-argmaxpool/ |
D | 9p8x-neon-c4.c | 159 const uint32x4_t vidx3 = vaddq_u32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local 161 vidx = vbslq_u32(vm3, vidx3, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() 164 const uint32x4_t vidx4 = vaddq_u32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() 262 const uint32x4_t vidx3 = vaddq_u32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local 264 vidx = vbslq_u32(vm3, vidx3, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() 267 const uint32x4_t vidx4 = vaddq_u32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() 317 const uint32x4_t vidx3 = vaddq_u32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() local 319 vidx = vbslq_u32(vm3, vidx3, vidx); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4() 322 const uint32x4_t vidx4 = vaddq_u32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__neon_c4()
|
D | 9p8x-wasmsimd-c4.c | 178 const v128_t vidx3 = wasm_i32x4_add(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local 180 vidx = wasm_v128_bitselect(vidx3, vidx, vm3); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() 183 const v128_t vidx4 = wasm_i32x4_add(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() 293 const v128_t vidx3 = wasm_i32x4_add(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local 295 vidx = wasm_v128_bitselect(vidx3, vidx, vm3); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() 298 const v128_t vidx4 = wasm_i32x4_add(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() 350 const v128_t vidx3 = wasm_i32x4_add(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() local 352 vidx = wasm_v128_bitselect(vidx3, vidx, vm3); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4() 355 const v128_t vidx4 = wasm_i32x4_add(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__wasmsimd_c4()
|
D | 9p8x-sse2-c4.c | 178 const __m128i vidx3 = _mm_add_epi32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local 180 vidx = _mm_or_si128(_mm_andnot_si128(vm3, vidx), _mm_and_si128(vm3, vidx3)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() 183 const __m128i vidx4 = _mm_add_epi32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() 293 const __m128i vidx3 = _mm_add_epi32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local 295 vidx = _mm_or_si128(_mm_andnot_si128(vm3, vidx), _mm_and_si128(vm3, vidx3)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() 298 const __m128i vidx4 = _mm_add_epi32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() 350 const __m128i vidx3 = _mm_add_epi32(vidx2, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() local 352 vidx = _mm_or_si128(_mm_andnot_si128(vm3, vidx), _mm_and_si128(vm3, vidx3)); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4() 355 const __m128i vidx4 = _mm_add_epi32(vidx3, v1); in xnn_f32_argmaxpool_ukernel_9p8x__sse2_c4()
|
/external/XNNPACK/src/math/ |
D | expm1minus-sse2-rr2-lut16-p3.c | 83 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx, 6); in xnn_math_f32_expm1minus__sse2_rr2_lut16_p3() local 87 …m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx3))); in xnn_math_f32_expm1minus__sse2_rr2_lut16_p3()
|
D | sigmoid-sse2-rr2-lut64-p2-div.c | 88 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx, 6); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_div() local 92 …m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_64 + vidx3))); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_div()
|
D | sigmoid-sse2-rr2-lut64-p2-nr2.c | 89 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx, 6); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_nr2() local 93 …m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_64 + vidx3))); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_nr2()
|
D | exp-sse2-rr2-lut64-p2.c | 81 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx, 6); in xnn_math_f32_exp__sse2_rr2_lut64_p2() local 85 …st __m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2_k_over_64 + vidx3))); in xnn_math_f32_exp__sse2_rr2_lut64_p2()
|
D | sigmoid-sse2-rr2-lut64-p2-nr1.c | 89 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx, 6); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_nr1() local 93 …m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_64 + vidx3))); in xnn_math_f32_sigmoid__sse2_rr2_lut64_p2_nr1()
|
/external/XNNPACK/src/f32-raddstoreexpminusmax/gen/ |
D | scalar-lut64-p2-x4.c | 88 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4() local 93 const float vs3 = fp32_from_bits(xnn_table_exp2_k_over_64[vidx3] + ve3); in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4()
|
D | scalar-lut64-p2-x4-acc2.c | 89 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2() local 94 const float vs3 = fp32_from_bits(xnn_table_exp2_k_over_64[vidx3] + ve3); in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc2()
|
D | scalar-lut64-p2-x4-acc4.c | 91 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4() local 96 const float vs3 = fp32_from_bits(xnn_table_exp2_k_over_64[vidx3] + ve3); in xnn_f32_raddstoreexpminusmax_ukernel__scalar_lut64_p2_x4_acc4()
|
/external/XNNPACK/src/f32-sigmoid/gen/ |
D | scalar-lut2048-p1-div-x4.c | 67 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_sigmoid_ukernel__scalar_lut2048_p1_div_x4() local 68 const float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_2048[vidx3] + ve3); in xnn_f32_sigmoid_ukernel__scalar_lut2048_p1_div_x4()
|
D | scalar-lut64-p2-div-x4.c | 67 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_sigmoid_ukernel__scalar_lut64_p2_div_x4() local 68 const float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_64[vidx3] + ve3); in xnn_f32_sigmoid_ukernel__scalar_lut64_p2_div_x4()
|
D | sse41-lut64-p2-div-x8.c | 74 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx0123, 6); in xnn_f32_sigmoid_ukernel__sse41_lut64_p2_div_x8() local 78 …23 = _mm_insert_epi32(vl2, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_64 + vidx3)), 1); in xnn_f32_sigmoid_ukernel__sse41_lut64_p2_div_x8()
|
D | sse2-lut64-p2-div-x8.c | 78 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx0123, 6); in xnn_f32_sigmoid_ukernel__sse2_lut64_p2_div_x8() local 83 …m128i vl3 = _mm_cvtsi32_si128(*((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_64 + vidx3))); in xnn_f32_sigmoid_ukernel__sse2_lut64_p2_div_x8()
|
/external/XNNPACK/src/f32-velu/gen/ |
D | velu-scalar-rr2-lut16-p3-x4.c | 70 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4() local 80 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x4()
|
D | velu-wasm-rr2-lut16-p3-x4.c | 70 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4() local 80 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x4()
|
D | velu-scalar-rr2-lut16-p3-x5.c | 73 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5() local 86 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x5()
|
D | velu-wasm-rr2-lut16-p3-x5.c | 73 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5() local 86 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x5()
|
D | velu-scalar-rr2-lut16-p3-x6.c | 76 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6() local 92 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__scalar_rr2_lut16_p3_x6()
|
D | velu-wasm-rr2-lut16-p3-x6.c | 76 const uint32_t vidx3 = fp32_to_bits(vn3) & vindex_mask; in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x6() local 92 float vs3 = fp32_from_bits(xnn_table_exp2minus_k_over_16[vidx3] + ven3); in xnn_f32_velu_ukernel__wasm_rr2_lut16_p3_x6()
|
D | velu-avx2-rr1-lut16-p3-gather-x32.c | 65 const __m256i vidx3 = _mm256_and_si256(_mm256_castps_si256(vn3), vindex_mask); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x32() local 66 const __m256i vl3 = _mm256_i32gather_epi32(xnn_table_exp2minus_k_over_16, vidx3, sizeof(float)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x32()
|
D | velu-avx2-rr1-lut16-p3-gather-x40.c | 68 const __m256i vidx3 = _mm256_and_si256(_mm256_castps_si256(vn3), vindex_mask); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x40() local 69 const __m256i vl3 = _mm256_i32gather_epi32(xnn_table_exp2minus_k_over_16, vidx3, sizeof(float)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x40()
|
D | velu-avx2-rr1-lut16-p3-gather-x48.c | 71 const __m256i vidx3 = _mm256_and_si256(_mm256_castps_si256(vn3), vindex_mask); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x48() local 72 const __m256i vl3 = _mm256_i32gather_epi32(xnn_table_exp2minus_k_over_16, vidx3, sizeof(float)); in xnn_f32_velu_ukernel__avx2_rr1_lut16_p3_gather_x48()
|
D | velu-sse41-rr2-lut16-p3-x8.c | 80 const uint32_t vidx3 = (uint32_t) _mm_extract_epi16(vidx0123, 6); in xnn_f32_velu_ukernel__sse41_rr2_lut16_p3_x8() local 84 …23 = _mm_insert_epi32(vl2, *((const int*) ((uintptr_t) xnn_table_exp2minus_k_over_16 + vidx3)), 1); in xnn_f32_velu_ukernel__sse41_rr2_lut16_p3_x8()
|