Searched refs:vzero_cutoff (Results 1 – 12 of 12) sorted by relevance
/external/XNNPACK/src/math/ |
D | exp-avx512f-rr2-p5-scalef.c | 23 const __m512 vzero_cutoff = _mm512_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__avx512f_rr2_p5_scalef() local 50 const __mmask16 vinvuf = _mm512_cmp_ps_mask(vx, vzero_cutoff, _CMP_NLT_UQ); in xnn_math_f32_exp__avx512f_rr2_p5_scalef()
|
D | exp-avx2-rr2-lut8-p3-perm.c | 23 const __m256 vzero_cutoff = _mm256_set1_ps(-0x1.9FE368p6f); in xnn_math_f32_exp__avx2_rr2_lut8_p3_perm() local 93 vf = _mm256_andnot_ps(_mm256_cmp_ps(vx, vzero_cutoff, _CMP_LT_OS), vf); in xnn_math_f32_exp__avx2_rr2_lut8_p3_perm()
|
D | exp-avx512f-rr2-lut32-p2-perm2.c | 24 const __m512 vzero_cutoff = _mm512_set1_ps(-0x1.9FE368p6f); in xnn_math_f32_exp__avx512f_rr2_lut32_p2_perm2() local 62 const __mmask16 vinvuf = _mm512_cmp_ps_mask(vx, vzero_cutoff, _CMP_NLT_UQ); in xnn_math_f32_exp__avx512f_rr2_lut32_p2_perm2()
|
D | exp-avx512f-rr2-lut16-p3-perm.c | 24 const __m512 vzero_cutoff = _mm512_set1_ps(-0x1.9FE368p6f); in xnn_math_f32_exp__avx512f_rr2_lut16_p3_perm() local 57 const __mmask16 vinvuf = _mm512_cmp_ps_mask(vx, vzero_cutoff, _CMP_NLT_UQ); in xnn_math_f32_exp__avx512f_rr2_lut16_p3_perm()
|
D | exp-avx2-rr2-p5.c | 23 const __m256 vzero_cutoff = _mm256_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__avx2_rr2_p5() local 89 vf = _mm256_andnot_ps(_mm256_cmp_ps(vx, vzero_cutoff, _CMP_LT_OS), vf); in xnn_math_f32_exp__avx2_rr2_p5()
|
D | exp-avx512f-rr2-p5.c | 23 const __m512 vzero_cutoff = _mm512_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__avx512f_rr2_p5() local 88 … __m512 vf = _mm512_maskz_fmadd_ps(_mm512_cmp_ps_mask(vx, vzero_cutoff, _CMP_NLT_US), vt, vp, vso); in xnn_math_f32_exp__avx512f_rr2_p5()
|
D | exp-neonfma-rr2-p5.c | 24 const float32x4_t vzero_cutoff = vmovq_n_f32(-0x1.9FE368p+6f); in xnn_math_f32_exp__neonfma_rr2_p5() local 90 vf = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(vf), vcltq_f32(vx, vzero_cutoff))); in xnn_math_f32_exp__neonfma_rr2_p5()
|
D | exp-sse2-rr2-p5.c | 24 const __m128 vzero_cutoff = _mm_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__sse2_rr2_p5() local 90 vf = _mm_andnot_ps(_mm_cmplt_ps(vx, vzero_cutoff), vf); in xnn_math_f32_exp__sse2_rr2_p5()
|
D | exp-avx2-rr2-lut8-p4-perm.c | 24 const __m256 vzero_cutoff = _mm256_set1_ps(-0x1.9FE368p6f); in xnn_math_f32_exp__avx2_rr2_lut8_p4_perm() local 97 vf = _mm256_andnot_ps(_mm256_cmp_ps(vx, vzero_cutoff, _CMP_LT_OS), vf); in xnn_math_f32_exp__avx2_rr2_lut8_p4_perm()
|
D | exp-neonfma-rr2-lut64-p2.c | 28 const float32x4_t vzero_cutoff = vmovq_n_f32(-0x1.9FE368p6f); in xnn_math_f32_exp__neonfma_rr2_lut64_p2() local 101 vf = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(vf), vcltq_f32(vx, vzero_cutoff))); in xnn_math_f32_exp__neonfma_rr2_lut64_p2()
|
D | exp-avx-rr2-p5.c | 24 const __m256 vzero_cutoff = _mm256_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__avx_rr2_p5() local 98 vf = _mm256_andnot_ps(_mm256_cmp_ps(vx, vzero_cutoff, _CMP_LT_OS), vf); in xnn_math_f32_exp__avx_rr2_p5()
|
D | exp-sse2-rr2-lut64-p2.c | 28 const __m128 vzero_cutoff = _mm_set1_ps(-0x1.9FE368p+6f); in xnn_math_f32_exp__sse2_rr2_lut64_p2() local 111 vf = _mm_andnot_ps(_mm_cmplt_ps(vx, vzero_cutoff), vf); in xnn_math_f32_exp__sse2_rr2_lut64_p2()
|