/external/XNNPACK/src/math/ |
D | exp-avx512f-rr2-p5-scalef.c | 25 const __m512 vinf_cutoff = _mm512_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__avx512f_rr2_p5_scalef() local 49 const __mmask16 vinvof = _mm512_cmp_ps_mask(vx, vinf_cutoff, _CMP_NGT_UQ); in xnn_math_f32_exp__avx512f_rr2_p5_scalef()
|
D | exp-avx2-rr2-lut8-p3-perm.c | 25 const __m256 vinf_cutoff = _mm256_set1_ps(0x1.62E42Ep6f); in xnn_math_f32_exp__avx2_rr2_lut8_p3_perm() local 96 vf = _mm256_blendv_ps(vf, vplus_inf, _mm256_cmp_ps(vx, vinf_cutoff, _CMP_GT_OS)); in xnn_math_f32_exp__avx2_rr2_lut8_p3_perm()
|
D | exp-avx512f-rr2-lut32-p2-perm2.c | 26 const __m512 vinf_cutoff = _mm512_set1_ps(0x1.62E42Ep6f); in xnn_math_f32_exp__avx512f_rr2_lut32_p2_perm2() local 61 const __mmask16 vinvof = _mm512_cmp_ps_mask(vx, vinf_cutoff, _CMP_NGT_UQ); in xnn_math_f32_exp__avx512f_rr2_lut32_p2_perm2()
|
D | exp-avx512f-rr2-lut16-p3-perm.c | 26 const __m512 vinf_cutoff = _mm512_set1_ps(0x1.62E42Ep6f); in xnn_math_f32_exp__avx512f_rr2_lut16_p3_perm() local 56 const __mmask16 vinvof = _mm512_cmp_ps_mask(vx, vinf_cutoff, _CMP_NGT_UQ); in xnn_math_f32_exp__avx512f_rr2_lut16_p3_perm()
|
D | exp-avx2-rr2-p5.c | 25 const __m256 vinf_cutoff = _mm256_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__avx2_rr2_p5() local 92 vf = _mm256_blendv_ps(vf, vplus_inf, _mm256_cmp_ps(vx, vinf_cutoff, _CMP_GT_OS)); in xnn_math_f32_exp__avx2_rr2_p5()
|
D | exp-avx512f-rr2-p5.c | 25 const __m512 vinf_cutoff = _mm512_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__avx512f_rr2_p5() local 91 vf = _mm512_mask_mul_ps(vplus_inf, _mm512_cmp_ps_mask(vx, vinf_cutoff, _CMP_NGT_US), vsn, vf); in xnn_math_f32_exp__avx512f_rr2_p5()
|
D | exp-neonfma-rr2-p5.c | 26 const float32x4_t vinf_cutoff = vmovq_n_f32(0x1.62E42Ep+6f); in xnn_math_f32_exp__neonfma_rr2_p5() local 93 vf = vbslq_f32(vcgtq_f32(vx, vinf_cutoff), vplus_inf, vf); in xnn_math_f32_exp__neonfma_rr2_p5()
|
D | exp-sse2-rr2-p5.c | 26 const __m128 vinf_cutoff = _mm_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__sse2_rr2_p5() local 93 const __m128 vm = _mm_cmpgt_ps(vx, vinf_cutoff); in xnn_math_f32_exp__sse2_rr2_p5()
|
D | exp-avx2-rr2-lut8-p4-perm.c | 26 const __m256 vinf_cutoff = _mm256_set1_ps(0x1.62E42Ep6f); in xnn_math_f32_exp__avx2_rr2_lut8_p4_perm() local 100 vf = _mm256_blendv_ps(vf, vplus_inf, _mm256_cmp_ps(vx, vinf_cutoff, _CMP_GT_OS)); in xnn_math_f32_exp__avx2_rr2_lut8_p4_perm()
|
D | exp-neonfma-rr2-lut64-p2.c | 30 const float32x4_t vinf_cutoff = vmovq_n_f32(0x1.62E42Ep6f); in xnn_math_f32_exp__neonfma_rr2_lut64_p2() local 104 vf = vbslq_f32(vcgtq_f32(vx, vinf_cutoff), vplus_inf, vf); in xnn_math_f32_exp__neonfma_rr2_lut64_p2()
|
D | exp-avx-rr2-p5.c | 26 const __m256 vinf_cutoff = _mm256_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__avx_rr2_p5() local 101 vf = _mm256_blendv_ps(vf, vplus_inf, _mm256_cmp_ps(vx, vinf_cutoff, _CMP_GT_OS)); in xnn_math_f32_exp__avx_rr2_p5()
|
D | exp-sse2-rr2-lut64-p2.c | 30 const __m128 vinf_cutoff = _mm_set1_ps(0x1.62E42Ep+6f); in xnn_math_f32_exp__sse2_rr2_lut64_p2() local 114 const __m128 vm = _mm_cmpgt_ps(vx, vinf_cutoff); in xnn_math_f32_exp__sse2_rr2_lut64_p2()
|