| /external/XNNPACK/src/f32-gemm/gen/ |
| D | 6x8-minmax-neonfma-dup-ld128.c | 148 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
154 vacc5x0123 = vfmaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
160 vacc5x4567 = vfmaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_gemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8-minmax-wasmsimd-arm-splat.c | 152 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat() local
162 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
168 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-neon-dup-ld128.c | 148 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128() local
154 vacc5x0123 = vmlaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
160 vacc5x4567 = vmlaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_gemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8-minmax-wasmsimd-x86-splat.c | 150 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_splat() local
160 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_splat()
166 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_gemm_minmax_ukernel_6x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f32-igemm/gen/ |
| D | 6x8-minmax-neon-dup-ld128.c | 176 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128() local
182 vacc5x0123 = vmlaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
188 vacc5x4567 = vmlaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_igemm_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8-minmax-neonfma-dup-ld128.c | 176 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128() local
182 vacc5x0123 = vfmaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
188 vacc5x4567 = vfmaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_igemm_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8-minmax-wasmsimd-arm-splat.c | 180 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_arm_splat() local
190 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
196 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| D | 6x8-minmax-wasmsimd-x86-splat.c | 178 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_x86_splat() local
188 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_x86_splat()
194 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_igemm_minmax_ukernel_6x8__wasmsimd_x86_splat()
|
| /external/XNNPACK/src/f32-gemm/gen-inc/ |
| D | 6x8inc-minmax-neon-dup-ld128.c | 150 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128() local
156 vacc5x0123 = vmlaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
162 vacc5x4567 = vmlaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_gemminc_minmax_ukernel_6x8__neon_dup_ld128()
|
| D | 6x8inc-minmax-neonfma-dup-ld128.c | 150 const float32x4_t va5c2 = vdupq_lane_f32(vget_high_f32(va5), 0); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128() local
156 vacc5x0123 = vfmaq_f32(vacc5x0123, va5c2, vb0123c2); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
162 vacc5x4567 = vfmaq_f32(vacc5x4567, va5c2, vb4567c2); in xnn_f32_gemminc_minmax_ukernel_6x8__neonfma_dup_ld128()
|
| D | 6x8inc-minmax-wasmsimd-x86-splat.c | 152 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat() local
162 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
168 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_x86_splat()
|
| D | 6x8inc-minmax-wasmsimd-arm-splat.c | 154 const v128_t va5c2 = wasm_v32x4_shuffle(va5, va5, 2, 2, 2, 2); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat() local
164 vacc5x0123 = wasm_f32x4_add(vacc5x0123, wasm_f32x4_mul(va5c2, vb0123c2)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
170 vacc5x4567 = wasm_f32x4_add(vacc5x4567, wasm_f32x4_mul(va5c2, vb4567c2)); in xnn_f32_gemminc_minmax_ukernel_6x8__wasmsimd_arm_splat()
|
| /external/XNNPACK/src/f16-gemm/gen/ |
| D | 6x16-minmax-neonfp16arith-ld64.c | 193 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemm_minmax_ukernel_6x16__neonfp16arith_ld64() local
200 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_6x16__neonfp16arith_ld64()
206 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_gemm_minmax_ukernel_6x16__neonfp16arith_ld64()
|
| D | 6x8-minmax-neonfp16arith-ld64.c | 154 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemm_minmax_ukernel_6x8__neonfp16arith_ld64() local
161 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_6x8__neonfp16arith_ld64()
|
| D | 8x16-minmax-neonfp16arith-ld64.c | 235 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemm_minmax_ukernel_8x16__neonfp16arith_ld64() local
244 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_8x16__neonfp16arith_ld64()
252 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_gemm_minmax_ukernel_8x16__neonfp16arith_ld64()
|
| D | 8x8-minmax-neonfp16arith-ld64.c | 184 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemm_minmax_ukernel_8x8__neonfp16arith_ld64() local
193 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemm_minmax_ukernel_8x8__neonfp16arith_ld64()
|
| /external/XNNPACK/src/f16-igemm/gen/ |
| D | 6x16-minmax-neonfp16arith-ld64.c | 219 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_igemm_minmax_ukernel_6x16__neonfp16arith_ld64() local
226 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_6x16__neonfp16arith_ld64()
232 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_igemm_minmax_ukernel_6x16__neonfp16arith_ld64()
|
| D | 6x8-minmax-neonfp16arith-ld64.c | 180 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_igemm_minmax_ukernel_6x8__neonfp16arith_ld64() local
187 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_6x8__neonfp16arith_ld64()
|
| D | 8x16-minmax-neonfp16arith-ld64.c | 267 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_igemm_minmax_ukernel_8x16__neonfp16arith_ld64() local
276 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_8x16__neonfp16arith_ld64()
284 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_igemm_minmax_ukernel_8x16__neonfp16arith_ld64()
|
| D | 8x8-minmax-neonfp16arith-ld64.c | 216 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_igemm_minmax_ukernel_8x8__neonfp16arith_ld64() local
225 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_igemm_minmax_ukernel_8x8__neonfp16arith_ld64()
|
| /external/XNNPACK/src/f16-gemm/gen-inc/ |
| D | 6x16inc-minmax-neonfp16arith-ld64.c | 195 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemminc_minmax_ukernel_6x16__neonfp16arith_ld64() local
202 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_6x16__neonfp16arith_ld64()
208 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_gemminc_minmax_ukernel_6x16__neonfp16arith_ld64()
|
| D | 6x8inc-minmax-neonfp16arith-ld64.c | 156 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemminc_minmax_ukernel_6x8__neonfp16arith_ld64() local
163 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_6x8__neonfp16arith_ld64()
|
| D | 8x16inc-minmax-neonfp16arith-ld64.c | 237 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemminc_minmax_ukernel_8x16__neonfp16arith_ld64() local
246 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_8x16__neonfp16arith_ld64()
254 vacc5x89ABCDEF = vfmaq_f16(vacc5x89ABCDEF, va5c2, vb89ABCDEFc2); in xnn_f16_gemminc_minmax_ukernel_8x16__neonfp16arith_ld64()
|
| D | 8x8inc-minmax-neonfp16arith-ld64.c | 186 const float16x8_t va5c2 = vdupq_lane_f16(va5, 2); in xnn_f16_gemminc_minmax_ukernel_8x8__neonfp16arith_ld64() local
195 vacc5x01234567 = vfmaq_f16(vacc5x01234567, va5c2, vb01234567c2); in xnn_f16_gemminc_minmax_ukernel_8x8__neonfp16arith_ld64()
|