/external/XNNPACK/src/f32-gemm/gen/ |
D | 1x8-wasmsimd-splat.c | 72 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat() local 77 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat() 78 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_ukernel_1x8__wasmsimd_splat()
|
D | 1x8-relu-wasmsimd-splat.c | 72 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat() local 77 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat() 78 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_relu_ukernel_1x8__wasmsimd_splat()
|
D | 1x8-minmax-wasmsimd-arm-splat.c | 74 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat() local 79 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat() 80 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
D | 1x8-minmax-wasmsimd-x86-splat.c | 72 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat() local 77 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat() 78 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
D | 3x8-minmax-wasmsimd-arm-splat.c | 112 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local 119 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat() 122 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
D | 3x8-minmax-wasmsimd-x86-splat.c | 110 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local 117 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat() 120 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
D | 4x8-wasmsimd-splat.c | 129 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() local 137 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat() 141 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemm_ukernel_4x8__wasmsimd_splat()
|
D | 4x8-minmax-neon-dup-ld128.c | 129 const float32x4_t va0c3 = vdupq_lane_f32(vget_high_f32(va0), 1); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() local 133 vacc0x0123 = vmlaq_f32(vacc0x0123, va0c3, vb0123c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128() 137 vacc0x4567 = vmlaq_f32(vacc0x4567, va0c3, vb4567c3); in xnn_f32_gemm_minmax_ukernel_4x8__neon_dup_ld128()
|
/external/XNNPACK/src/f32-igemm/gen/ |
D | 1x8-wasmsimd-splat.c | 85 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat() local 90 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat() 91 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_ukernel_1x8__wasmsimd_splat()
|
D | 1x8-relu-wasmsimd-splat.c | 85 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat() local 90 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat() 91 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_relu_ukernel_1x8__wasmsimd_splat()
|
D | 1x8-minmax-wasmsimd-x86-splat.c | 85 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat() local 90 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat() 91 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
D | 1x8-minmax-wasmsimd-arm-splat.c | 87 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat() local 92 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat() 93 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
D | 3x8-minmax-wasmsimd-arm-splat.c | 131 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() local 138 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat() 141 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
D | 3x8-minmax-wasmsimd-x86-splat.c | 129 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() local 136 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat() 139 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
D | 4x8-relu-wasmsimd-splat.c | 151 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() local 159 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat() 163 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_igemm_relu_ukernel_4x8__wasmsimd_splat()
|
/external/XNNPACK/src/f32-gemm/gen-inc/ |
D | 1x8inc-minmax-wasmsimd-arm-splat.c | 76 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat() local 81 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat() 82 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_arm_splat()
|
D | 1x8inc-minmax-wasmsimd-x86-splat.c | 74 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat() local 79 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat() 80 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_1x8__wasmsimd_x86_splat()
|
D | 3x8inc-minmax-wasmsimd-arm-splat.c | 114 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() local 121 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat() 124 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_arm_splat()
|
D | 3x8inc-minmax-wasmsimd-x86-splat.c | 112 const v128_t va0c3 = wasm_v32x4_shuffle(va0, va0, 3, 3, 3, 3); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() local 119 vacc0x0123 = wasm_f32x4_add(vacc0x0123, wasm_f32x4_mul(va0c3, vb0123c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat() 122 vacc0x4567 = wasm_f32x4_add(vacc0x4567, wasm_f32x4_mul(va0c3, vb4567c3)); in xnn_f32_gemminc_minmax_ukernel_3x8__wasmsimd_x86_splat()
|
/external/XNNPACK/src/f16-gemm/gen/ |
D | 1x16-minmax-neonfp16arith-ld64.c | 95 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64() local 97 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64() 98 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c3, vb89ABCDEFc3); in xnn_f16_gemm_minmax_ukernel_1x16__neonfp16arith_ld64()
|
D | 1x8-minmax-neonfp16arith-ld64.c | 83 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64() local 85 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_gemm_minmax_ukernel_1x8__neonfp16arith_ld64()
|
/external/XNNPACK/src/f16-gemm/gen-inc/ |
D | 1x16inc-minmax-neonfp16arith-ld64.c | 97 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64() local 99 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64() 100 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c3, vb89ABCDEFc3); in xnn_f16_gemminc_minmax_ukernel_1x16__neonfp16arith_ld64()
|
D | 1x8inc-minmax-neonfp16arith-ld64.c | 85 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_gemminc_minmax_ukernel_1x8__neonfp16arith_ld64() local 87 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_gemminc_minmax_ukernel_1x8__neonfp16arith_ld64()
|
/external/XNNPACK/src/f16-igemm/gen/ |
D | 1x16-minmax-neonfp16arith-ld64.c | 106 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64() local 108 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64() 109 vacc0x89ABCDEF = vfmaq_f16(vacc0x89ABCDEF, va0c3, vb89ABCDEFc3); in xnn_f16_igemm_minmax_ukernel_1x16__neonfp16arith_ld64()
|
D | 1x8-minmax-neonfp16arith-ld64.c | 94 const float16x8_t va0c3 = vdupq_lane_f16(va0, 3); in xnn_f16_igemm_minmax_ukernel_1x8__neonfp16arith_ld64() local 96 vacc0x01234567 = vfmaq_f16(vacc0x01234567, va0c3, vb01234567c3); in xnn_f16_igemm_minmax_ukernel_1x8__neonfp16arith_ld64()
|