/external/XNNPACK/src/f32-gemm/gen/ |
D | 1x8-neon-dup-ld64.c | 53 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemm_ukernel_1x8__neon_dup_ld64() local 54 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_1x8__neon_dup_ld64() 55 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_1x8__neon_dup_ld64()
|
D | 1x8-neonfma-dup-ld64.c | 53 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemm_ukernel_1x8__neonfma_dup_ld64() local 54 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_1x8__neonfma_dup_ld64() 55 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_1x8__neonfma_dup_ld64()
|
D | 1x8-psimd-splat.c | 51 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_gemm_ukernel_1x8__psimd_splat() local 56 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_1x8__psimd_splat() 57 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_1x8__psimd_splat()
|
D | 4x8-neon-dup-ld64.c | 80 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld64() local 84 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld64() 88 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld64()
|
D | 4x8-neonfma-dup-ld64.c | 80 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld64() local 84 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld64() 88 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld64()
|
D | 4x8-psimd-splat.c | 81 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_gemm_ukernel_4x8__psimd_splat() local 89 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_4x8__psimd_splat() 93 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_4x8__psimd_splat()
|
D | 4x8-neonfma-dup-ld128.c | 81 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld128() local 85 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld128() 89 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8-neon-dup-ld128.c | 81 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld128() local 85 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld128() 89 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemm_ukernel_4x8__neon_dup_ld128()
|
D | 4x2-neonfma-lane-ld64.c | 81 const float32x2_t va0c0 = vdup_lane_f32(va0, 0); in xnn_f32_gemm_ukernel_4x2__neonfma_lane_ld64() local 85 vacc0x01 = vfma_f32(vacc0x01, va0c0, vb01c0); in xnn_f32_gemm_ukernel_4x2__neonfma_lane_ld64()
|
/external/XNNPACK/src/f32-gemm/gen-inc/ |
D | 1x8-neon-dup-ld64.c | 55 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemminc_ukernel_1x8__neon_dup_ld64() local 56 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_1x8__neon_dup_ld64() 57 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_1x8__neon_dup_ld64()
|
D | 1x8-neonfma-dup-ld64.c | 55 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemminc_ukernel_1x8__neonfma_dup_ld64() local 56 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_1x8__neonfma_dup_ld64() 57 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_1x8__neonfma_dup_ld64()
|
D | 1x8-psimd-splat.c | 53 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_gemminc_ukernel_1x8__psimd_splat() local 58 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_1x8__psimd_splat() 59 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_1x8__psimd_splat()
|
D | 4x8-neon-dup-ld64.c | 82 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld64() local 86 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld64() 90 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld64()
|
D | 4x8-neonfma-dup-ld64.c | 82 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld64() local 86 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld64() 90 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld64()
|
D | 4x8-neonfma-dup-ld128.c | 83 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld128() local 87 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld128() 91 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8-psimd-splat.c | 83 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_gemminc_ukernel_4x8__psimd_splat() local 91 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_4x8__psimd_splat() 95 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_4x8__psimd_splat()
|
D | 4x8-neon-dup-ld128.c | 83 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld128() local 87 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld128() 91 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_gemminc_ukernel_4x8__neon_dup_ld128()
|
/external/XNNPACK/src/f32-igemm/gen/ |
D | 1x8-neon-dup-ld64.c | 66 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_igemm_ukernel_1x8__neon_dup_ld64() local 67 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_1x8__neon_dup_ld64() 68 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_1x8__neon_dup_ld64()
|
D | 1x8-neonfma-dup-ld64.c | 66 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_igemm_ukernel_1x8__neonfma_dup_ld64() local 67 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_1x8__neonfma_dup_ld64() 68 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_1x8__neonfma_dup_ld64()
|
D | 1x8-psimd-splat.c | 64 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_igemm_ukernel_1x8__psimd_splat() local 69 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_1x8__psimd_splat() 70 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_1x8__psimd_splat()
|
D | 4x8-neon-dup-ld64.c | 102 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld64() local 106 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld64() 110 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld64()
|
D | 4x8-neonfma-dup-ld64.c | 102 const float32x4_t va0c0 = vdupq_lane_f32(va0, 0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld64() local 106 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld64() 110 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld64()
|
D | 4x8-psimd-splat.c | 103 const psimd_f32 va0c0 = psimd_splat0_f32(va0); in xnn_f32_igemm_ukernel_4x8__psimd_splat() local 111 vacc0x0123 = psimd_qfma_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_4x8__psimd_splat() 115 vacc0x4567 = psimd_qfma_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_4x8__psimd_splat()
|
D | 4x8-neonfma-dup-ld128.c | 103 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld128() local 107 vacc0x0123 = vfmaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld128() 111 vacc0x4567 = vfmaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_4x8__neonfma_dup_ld128()
|
D | 4x8-neon-dup-ld128.c | 103 const float32x4_t va0c0 = vdupq_lane_f32(vget_low_f32(va0), 0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld128() local 107 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c0, vb0123c0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld128() 111 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c0, vb4567c0); in xnn_f32_igemm_ukernel_4x8__neon_dup_ld128()
|