/external/XNNPACK/src/bf16-gemm/gen/ |
D | 1x4c8-minmax-neonfma-zip.c | 74 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip() local 117 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_zip() local
|
D | 1x4c8-minmax-neonfma-shland.c | 74 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland() local 117 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_1x4c8__neonfma_shland() local
|
D | 2x4c8-minmax-neonfma-shland.c | 91 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland() local 151 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_shland() local
|
D | 2x4c8-minmax-neonfma-zip.c | 91 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip() local 151 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_2x4c8__neonfma_zip() local
|
D | 3x4c8-minmax-neonfma-shland.c | 108 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonfma_shland() local 185 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonfma_shland() local
|
D | 3x4c8-minmax-neonfma-zip.c | 108 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonfma_zip() local 185 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_3x4c8__neonfma_zip() local
|
D | 4x4c8-minmax-neonfma-zip.c | 125 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip() local 219 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_zip() local
|
D | 4x4c8-minmax-neonfma-shland.c | 125 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland() local 219 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_4x4c8__neonfma_shland() local
|
D | 5x4c8-minmax-neonfma-zip.c | 142 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonfma_zip() local 253 const float32x4_t vb1o = vreinterpretq_f32_u16(vzip2q_u16(vzero, vb1)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonfma_zip() local
|
D | 5x4c8-minmax-neonfma-shland.c | 142 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonfma_shland() local 253 const float32x4_t vb1o = vreinterpretq_f32_u16(vandq_u16(vb1, vmask)); in xnn_bf16_gemm_minmax_ukernel_5x4c8__neonfma_shland() local
|