Mike Klein | 267f507 | 2019-06-03 16:27:46 -0500 | [diff] [blame] | 1 | A8 over A8 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 2 | 12 values: |
| 3 | v0 = load8 arg(0) |
| 4 | v1 = to_f32 v0 |
| 5 | v2 = mul_f32 v1 3B808081 (0.0039215689) |
| 6 | v3 = load8 arg(1) |
| 7 | v4 = to_f32 v3 |
| 8 | v5 = mul_f32 v4 3B808081 (0.0039215689) |
| 9 | ↑ v6 = splat 3F800000 (1) |
| 10 | v7 = sub_f32 v6 v2 |
| 11 | v8 = mad_f32 v5 v7 v2 |
| 12 | v9 = mul_f32 v8 437F0000 (255) |
| 13 | v10 = round v9 |
| 14 | store8 arg(1) v10 |
Mike Klein | 22ea7e9 | 2019-06-10 12:05:48 -0500 | [diff] [blame] | 15 | |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 16 | 4 registers, 12 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 17 | 0 r0 = splat 3F800000 (1) |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 18 | loop: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 19 | 1 r1 = load8 arg(0) |
| 20 | 2 r1 = to_f32 r1 |
| 21 | 3 r1 = mul_f32 r1 3B808081 (0.0039215689) |
| 22 | 4 r2 = load8 arg(1) |
| 23 | 5 r2 = to_f32 r2 |
| 24 | 6 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 25 | 7 r3 = sub_f32 r0 r1 |
| 26 | 8 r1 = mad_f32 r2 r3 r1 |
| 27 | 9 r1 = mul_f32 r1 437F0000 (255) |
| 28 | 10 r1 = round r1 |
| 29 | 11 store8 arg(1) r1 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 30 | |
| 31 | A8 over G8 |
| 32 | 17 values: |
| 33 | v0 = load8 arg(1) |
| 34 | v1 = to_f32 v0 |
| 35 | v2 = mul_f32 v1 3B808081 (0.0039215689) |
| 36 | v3 = load8 arg(0) |
| 37 | v4 = to_f32 v3 |
| 38 | v5 = mul_f32 v4 3B808081 (0.0039215689) |
| 39 | ↑ v6 = splat 3F800000 (1) |
| 40 | v7 = sub_f32 v6 v5 |
| 41 | v8 = mul_f32 v2 v7 |
| 42 | ↑ v9 = splat 3E59B3D0 (0.21259999) |
| 43 | ↑ v10 = splat 3F371759 (0.71520001) |
| 44 | v11 = mul_f32 v8 3D93DD98 (0.0722) |
| 45 | v12 = mad_f32 v8 v10 v11 |
| 46 | v13 = mad_f32 v8 v9 v12 |
| 47 | v14 = mul_f32 v13 437F0000 (255) |
| 48 | v15 = round v14 |
| 49 | store8 arg(1) v15 |
| 50 | |
| 51 | 5 registers, 17 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 52 | 0 r0 = splat 3F800000 (1) |
| 53 | 1 r1 = splat 3E59B3D0 (0.21259999) |
| 54 | 2 r2 = splat 3F371759 (0.71520001) |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 55 | loop: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 56 | 3 r3 = load8 arg(1) |
| 57 | 4 r3 = to_f32 r3 |
| 58 | 5 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 59 | 6 r4 = load8 arg(0) |
| 60 | 7 r4 = to_f32 r4 |
| 61 | 8 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 62 | 9 r4 = sub_f32 r0 r4 |
| 63 | 10 r4 = mul_f32 r3 r4 |
| 64 | 11 r3 = mul_f32 r4 3D93DD98 (0.0722) |
| 65 | 12 r3 = mad_f32 r4 r2 r3 |
| 66 | 13 r3 = mad_f32 r4 r1 r3 |
| 67 | 14 r3 = mul_f32 r3 437F0000 (255) |
| 68 | 15 r3 = round r3 |
| 69 | 16 store8 arg(1) r3 |
Mike Klein | 267f507 | 2019-06-03 16:27:46 -0500 | [diff] [blame] | 70 | |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 71 | A8 over RGBA_8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 72 | 37 values: |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 73 | v0 = load32 arg(1) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 74 | v1 = bit_and v0 FF |
| 75 | v2 = to_f32 v1 |
| 76 | v3 = mul_f32 v2 3B808081 (0.0039215689) |
| 77 | v4 = load8 arg(0) |
| 78 | v5 = to_f32 v4 |
| 79 | v6 = mul_f32 v5 3B808081 (0.0039215689) |
| 80 | ↑ v7 = splat 3F800000 (1) |
| 81 | v8 = sub_f32 v7 v6 |
| 82 | v9 = mul_f32 v3 v8 |
| 83 | v10 = mul_f32 v9 437F0000 (255) |
| 84 | v11 = round v10 |
| 85 | v12 = shr_i32 v0 8 |
| 86 | v13 = bit_and v12 FF |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 87 | v14 = to_f32 v13 |
| 88 | v15 = mul_f32 v14 3B808081 (0.0039215689) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 89 | v16 = mul_f32 v15 v8 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 90 | v17 = mul_f32 v16 437F0000 (255) |
| 91 | v18 = round v17 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 92 | v19 = pack v11 v18 8 |
| 93 | v20 = shr_i32 v0 16 |
| 94 | v21 = bit_and v20 FF |
| 95 | v22 = to_f32 v21 |
| 96 | v23 = mul_f32 v22 3B808081 (0.0039215689) |
| 97 | v24 = mul_f32 v23 v8 |
| 98 | v25 = mul_f32 v24 437F0000 (255) |
| 99 | v26 = round v25 |
| 100 | v27 = shr_i32 v0 24 |
| 101 | v28 = bit_and v27 FF |
| 102 | v29 = to_f32 v28 |
| 103 | v30 = mul_f32 v29 3B808081 (0.0039215689) |
| 104 | v31 = mad_f32 v30 v8 v6 |
| 105 | v32 = mul_f32 v31 437F0000 (255) |
| 106 | v33 = round v32 |
| 107 | v34 = pack v26 v33 8 |
| 108 | v35 = pack v19 v34 16 |
| 109 | store32 arg(1) v35 |
Mike Klein | 22ea7e9 | 2019-06-10 12:05:48 -0500 | [diff] [blame] | 110 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 111 | 6 registers, 37 instructions: |
| 112 | 0 r0 = splat 3F800000 (1) |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 113 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 114 | 1 r1 = load32 arg(1) |
| 115 | 2 r2 = bit_and r1 FF |
| 116 | 3 r2 = to_f32 r2 |
| 117 | 4 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 118 | 5 r3 = load8 arg(0) |
| 119 | 6 r3 = to_f32 r3 |
| 120 | 7 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 121 | 8 r4 = sub_f32 r0 r3 |
| 122 | 9 r2 = mul_f32 r2 r4 |
| 123 | 10 r2 = mul_f32 r2 437F0000 (255) |
| 124 | 11 r2 = round r2 |
| 125 | 12 r5 = shr_i32 r1 8 |
| 126 | 13 r5 = bit_and r5 FF |
| 127 | 14 r5 = to_f32 r5 |
| 128 | 15 r5 = mul_f32 r5 3B808081 (0.0039215689) |
| 129 | 16 r5 = mul_f32 r5 r4 |
| 130 | 17 r5 = mul_f32 r5 437F0000 (255) |
| 131 | 18 r5 = round r5 |
| 132 | 19 r5 = pack r2 r5 8 |
| 133 | 20 r2 = shr_i32 r1 16 |
| 134 | 21 r2 = bit_and r2 FF |
| 135 | 22 r2 = to_f32 r2 |
| 136 | 23 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 137 | 24 r2 = mul_f32 r2 r4 |
| 138 | 25 r2 = mul_f32 r2 437F0000 (255) |
| 139 | 26 r2 = round r2 |
| 140 | 27 r1 = shr_i32 r1 24 |
| 141 | 28 r1 = bit_and r1 FF |
| 142 | 29 r1 = to_f32 r1 |
| 143 | 30 r1 = mul_f32 r1 3B808081 (0.0039215689) |
| 144 | 31 r3 = mad_f32 r1 r4 r3 |
| 145 | 32 r3 = mul_f32 r3 437F0000 (255) |
| 146 | 33 r3 = round r3 |
| 147 | 34 r3 = pack r2 r3 8 |
| 148 | 35 r3 = pack r5 r3 16 |
| 149 | 36 store32 arg(1) r3 |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 150 | |
| 151 | G8 over A8 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 152 | 9 values: |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 153 | ↑ v0 = splat 3F800000 (1) |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 154 | ↑ v1 = splat 0 (0) |
| 155 | v2 = load8 arg(1) |
| 156 | v3 = to_f32 v2 |
| 157 | v4 = mul_f32 v3 3B808081 (0.0039215689) |
| 158 | v5 = mad_f32 v4 v1 v0 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 159 | v6 = mul_f32 v5 437F0000 (255) |
| 160 | v7 = round v6 |
| 161 | store8 arg(1) v7 |
Mike Klein | 22ea7e9 | 2019-06-10 12:05:48 -0500 | [diff] [blame] | 162 | |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 163 | 3 registers, 9 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 164 | 0 r0 = splat 3F800000 (1) |
| 165 | 1 r1 = splat 0 (0) |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 166 | loop: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 167 | 2 r2 = load8 arg(1) |
| 168 | 3 r2 = to_f32 r2 |
| 169 | 4 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 170 | 5 r2 = mad_f32 r2 r1 r0 |
| 171 | 6 r2 = mul_f32 r2 437F0000 (255) |
| 172 | 7 r2 = round r2 |
| 173 | 8 store8 arg(1) r2 |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 174 | |
| 175 | G8 over G8 |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 176 | 16 values: |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 177 | v0 = load8 arg(0) |
| 178 | v1 = to_f32 v0 |
| 179 | v2 = mul_f32 v1 3B808081 (0.0039215689) |
| 180 | v3 = load8 arg(1) |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 181 | v4 = to_f32 v3 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 182 | v5 = mul_f32 v4 3B808081 (0.0039215689) |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 183 | ↑ v6 = splat 0 (0) |
| 184 | v7 = mad_f32 v5 v6 v2 |
| 185 | ↑ v8 = splat 3E59B3D0 (0.21259999) |
| 186 | ↑ v9 = splat 3F371759 (0.71520001) |
| 187 | v10 = mul_f32 v7 3D93DD98 (0.0722) |
| 188 | v11 = mad_f32 v7 v9 v10 |
| 189 | v12 = mad_f32 v7 v8 v11 |
| 190 | v13 = mul_f32 v12 437F0000 (255) |
| 191 | v14 = round v13 |
| 192 | store8 arg(1) v14 |
Mike Klein | 22ea7e9 | 2019-06-10 12:05:48 -0500 | [diff] [blame] | 193 | |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 194 | 5 registers, 16 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 195 | 0 r0 = splat 0 (0) |
| 196 | 1 r1 = splat 3E59B3D0 (0.21259999) |
| 197 | 2 r2 = splat 3F371759 (0.71520001) |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 198 | loop: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 199 | 3 r3 = load8 arg(0) |
| 200 | 4 r3 = to_f32 r3 |
| 201 | 5 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 202 | 6 r4 = load8 arg(1) |
| 203 | 7 r4 = to_f32 r4 |
| 204 | 8 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 205 | 9 r3 = mad_f32 r4 r0 r3 |
| 206 | 10 r4 = mul_f32 r3 3D93DD98 (0.0722) |
| 207 | 11 r4 = mad_f32 r3 r2 r4 |
| 208 | 12 r4 = mad_f32 r3 r1 r4 |
| 209 | 13 r4 = mul_f32 r4 437F0000 (255) |
| 210 | 14 r4 = round r4 |
| 211 | 15 store8 arg(1) r4 |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 212 | |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 213 | G8 over RGBA_8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 214 | 37 values: |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 215 | v0 = load8 arg(0) |
| 216 | v1 = to_f32 v0 |
| 217 | v2 = mul_f32 v1 3B808081 (0.0039215689) |
| 218 | v3 = load32 arg(1) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 219 | v4 = bit_and v3 FF |
| 220 | v5 = to_f32 v4 |
| 221 | v6 = mul_f32 v5 3B808081 (0.0039215689) |
| 222 | ↑ v7 = splat 0 (0) |
| 223 | v8 = mad_f32 v6 v7 v2 |
| 224 | v9 = mul_f32 v8 437F0000 (255) |
| 225 | v10 = round v9 |
| 226 | v11 = shr_i32 v3 8 |
| 227 | v12 = bit_and v11 FF |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 228 | v13 = to_f32 v12 |
| 229 | v14 = mul_f32 v13 3B808081 (0.0039215689) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 230 | v15 = mad_f32 v14 v7 v2 |
Mike Klein | 1cb0599 | 2019-11-21 11:46:58 -0600 | [diff] [blame] | 231 | v16 = mul_f32 v15 437F0000 (255) |
| 232 | v17 = round v16 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 233 | v18 = pack v10 v17 8 |
| 234 | v19 = shr_i32 v3 16 |
| 235 | v20 = bit_and v19 FF |
| 236 | v21 = to_f32 v20 |
| 237 | v22 = mul_f32 v21 3B808081 (0.0039215689) |
| 238 | v23 = mad_f32 v22 v7 v2 |
| 239 | v24 = mul_f32 v23 437F0000 (255) |
| 240 | v25 = round v24 |
| 241 | ↑ v26 = splat 3F800000 (1) |
| 242 | v27 = shr_i32 v3 24 |
| 243 | v28 = bit_and v27 FF |
| 244 | v29 = to_f32 v28 |
| 245 | v30 = mul_f32 v29 3B808081 (0.0039215689) |
| 246 | v31 = mad_f32 v30 v7 v26 |
| 247 | v32 = mul_f32 v31 437F0000 (255) |
| 248 | v33 = round v32 |
| 249 | v34 = pack v25 v33 8 |
| 250 | v35 = pack v18 v34 16 |
| 251 | store32 arg(1) v35 |
Mike Klein | 22ea7e9 | 2019-06-10 12:05:48 -0500 | [diff] [blame] | 252 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 253 | 6 registers, 37 instructions: |
| 254 | 0 r0 = splat 0 (0) |
| 255 | 1 r1 = splat 3F800000 (1) |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 256 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 257 | 2 r2 = load8 arg(0) |
| 258 | 3 r2 = to_f32 r2 |
| 259 | 4 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 260 | 5 r3 = load32 arg(1) |
| 261 | 6 r4 = bit_and r3 FF |
| 262 | 7 r4 = to_f32 r4 |
| 263 | 8 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 264 | 9 r4 = mad_f32 r4 r0 r2 |
| 265 | 10 r4 = mul_f32 r4 437F0000 (255) |
| 266 | 11 r4 = round r4 |
| 267 | 12 r5 = shr_i32 r3 8 |
| 268 | 13 r5 = bit_and r5 FF |
| 269 | 14 r5 = to_f32 r5 |
| 270 | 15 r5 = mul_f32 r5 3B808081 (0.0039215689) |
| 271 | 16 r5 = mad_f32 r5 r0 r2 |
| 272 | 17 r5 = mul_f32 r5 437F0000 (255) |
| 273 | 18 r5 = round r5 |
| 274 | 19 r5 = pack r4 r5 8 |
| 275 | 20 r4 = shr_i32 r3 16 |
| 276 | 21 r4 = bit_and r4 FF |
| 277 | 22 r4 = to_f32 r4 |
| 278 | 23 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 279 | 24 r2 = mad_f32 r4 r0 r2 |
| 280 | 25 r2 = mul_f32 r2 437F0000 (255) |
| 281 | 26 r2 = round r2 |
| 282 | 27 r3 = shr_i32 r3 24 |
| 283 | 28 r3 = bit_and r3 FF |
| 284 | 29 r3 = to_f32 r3 |
| 285 | 30 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 286 | 31 r3 = mad_f32 r3 r0 r1 |
| 287 | 32 r3 = mul_f32 r3 437F0000 (255) |
| 288 | 33 r3 = round r3 |
| 289 | 34 r3 = pack r2 r3 8 |
| 290 | 35 r3 = pack r5 r3 16 |
| 291 | 36 store32 arg(1) r3 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 292 | |
| 293 | RGBA_8888 over A8 |
| 294 | 14 values: |
| 295 | v0 = load32 arg(0) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 296 | v1 = shr_i32 v0 24 |
| 297 | v2 = bit_and v1 FF |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 298 | v3 = to_f32 v2 |
| 299 | v4 = mul_f32 v3 3B808081 (0.0039215689) |
| 300 | v5 = load8 arg(1) |
| 301 | v6 = to_f32 v5 |
| 302 | v7 = mul_f32 v6 3B808081 (0.0039215689) |
| 303 | ↑ v8 = splat 3F800000 (1) |
| 304 | v9 = sub_f32 v8 v4 |
| 305 | v10 = mad_f32 v7 v9 v4 |
| 306 | v11 = mul_f32 v10 437F0000 (255) |
| 307 | v12 = round v11 |
| 308 | store8 arg(1) v12 |
| 309 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 310 | 4 registers, 14 instructions: |
| 311 | 0 r0 = splat 3F800000 (1) |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 312 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 313 | 1 r1 = load32 arg(0) |
| 314 | 2 r1 = shr_i32 r1 24 |
| 315 | 3 r1 = bit_and r1 FF |
| 316 | 4 r1 = to_f32 r1 |
| 317 | 5 r1 = mul_f32 r1 3B808081 (0.0039215689) |
| 318 | 6 r2 = load8 arg(1) |
| 319 | 7 r2 = to_f32 r2 |
| 320 | 8 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 321 | 9 r3 = sub_f32 r0 r1 |
| 322 | 10 r1 = mad_f32 r2 r3 r1 |
| 323 | 11 r1 = mul_f32 r1 437F0000 (255) |
| 324 | 12 r1 = round r1 |
| 325 | 13 store8 arg(1) r1 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 326 | |
| 327 | RGBA_8888 over G8 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 328 | 32 values: |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 329 | v0 = load32 arg(0) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 330 | v1 = bit_and v0 FF |
| 331 | v2 = to_f32 v1 |
| 332 | v3 = mul_f32 v2 3B808081 (0.0039215689) |
| 333 | v4 = load8 arg(1) |
| 334 | v5 = to_f32 v4 |
| 335 | v6 = mul_f32 v5 3B808081 (0.0039215689) |
| 336 | v7 = shr_i32 v0 24 |
| 337 | v8 = bit_and v7 FF |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 338 | v9 = to_f32 v8 |
| 339 | v10 = mul_f32 v9 3B808081 (0.0039215689) |
| 340 | ↑ v11 = splat 3F800000 (1) |
| 341 | v12 = sub_f32 v11 v10 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 342 | v13 = mad_f32 v6 v12 v3 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 343 | ↑ v14 = splat 3E59B3D0 (0.21259999) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 344 | v15 = shr_i32 v0 8 |
| 345 | v16 = bit_and v15 FF |
| 346 | v17 = to_f32 v16 |
| 347 | v18 = mul_f32 v17 3B808081 (0.0039215689) |
| 348 | v19 = mad_f32 v6 v12 v18 |
| 349 | ↑ v20 = splat 3F371759 (0.71520001) |
| 350 | v21 = shr_i32 v0 16 |
| 351 | v22 = bit_and v21 FF |
| 352 | v23 = to_f32 v22 |
| 353 | v24 = mul_f32 v23 3B808081 (0.0039215689) |
| 354 | v25 = mad_f32 v6 v12 v24 |
| 355 | v26 = mul_f32 v25 3D93DD98 (0.0722) |
| 356 | v27 = mad_f32 v19 v20 v26 |
| 357 | v28 = mad_f32 v13 v14 v27 |
| 358 | v29 = mul_f32 v28 437F0000 (255) |
| 359 | v30 = round v29 |
| 360 | store8 arg(1) v30 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 361 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 362 | 8 registers, 32 instructions: |
| 363 | 0 r0 = splat 3F800000 (1) |
| 364 | 1 r1 = splat 3E59B3D0 (0.21259999) |
| 365 | 2 r2 = splat 3F371759 (0.71520001) |
Mike Klein | a630732 | 2019-06-07 15:44:26 -0500 | [diff] [blame] | 366 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 367 | 3 r3 = load32 arg(0) |
| 368 | 4 r4 = bit_and r3 FF |
| 369 | 5 r4 = to_f32 r4 |
| 370 | 6 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 371 | 7 r5 = load8 arg(1) |
| 372 | 8 r5 = to_f32 r5 |
| 373 | 9 r5 = mul_f32 r5 3B808081 (0.0039215689) |
| 374 | 10 r6 = shr_i32 r3 24 |
| 375 | 11 r6 = bit_and r6 FF |
| 376 | 12 r6 = to_f32 r6 |
| 377 | 13 r6 = mul_f32 r6 3B808081 (0.0039215689) |
| 378 | 14 r6 = sub_f32 r0 r6 |
| 379 | 15 r4 = mad_f32 r5 r6 r4 |
| 380 | 16 r7 = shr_i32 r3 8 |
| 381 | 17 r7 = bit_and r7 FF |
| 382 | 18 r7 = to_f32 r7 |
| 383 | 19 r7 = mul_f32 r7 3B808081 (0.0039215689) |
| 384 | 20 r7 = mad_f32 r5 r6 r7 |
| 385 | 21 r3 = shr_i32 r3 16 |
| 386 | 22 r3 = bit_and r3 FF |
| 387 | 23 r3 = to_f32 r3 |
| 388 | 24 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 389 | 25 r3 = mad_f32 r5 r6 r3 |
| 390 | 26 r3 = mul_f32 r3 3D93DD98 (0.0722) |
| 391 | 27 r3 = mad_f32 r7 r2 r3 |
| 392 | 28 r3 = mad_f32 r4 r1 r3 |
| 393 | 29 r3 = mul_f32 r3 437F0000 (255) |
| 394 | 30 r3 = round r3 |
| 395 | 31 store8 arg(1) r3 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 396 | |
| 397 | RGBA_8888 over RGBA_8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 398 | 50 values: |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 399 | v0 = load32 arg(0) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 400 | v1 = bit_and v0 FF |
| 401 | v2 = to_f32 v1 |
| 402 | v3 = mul_f32 v2 3B808081 (0.0039215689) |
| 403 | v4 = load32 arg(1) |
| 404 | v5 = bit_and v4 FF |
| 405 | v6 = to_f32 v5 |
| 406 | v7 = mul_f32 v6 3B808081 (0.0039215689) |
| 407 | v8 = shr_i32 v0 24 |
| 408 | v9 = bit_and v8 FF |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 409 | v10 = to_f32 v9 |
| 410 | v11 = mul_f32 v10 3B808081 (0.0039215689) |
| 411 | ↑ v12 = splat 3F800000 (1) |
| 412 | v13 = sub_f32 v12 v11 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 413 | v14 = mad_f32 v7 v13 v3 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 414 | v15 = mul_f32 v14 437F0000 (255) |
| 415 | v16 = round v15 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 416 | v17 = shr_i32 v0 8 |
| 417 | v18 = bit_and v17 FF |
| 418 | v19 = to_f32 v18 |
| 419 | v20 = mul_f32 v19 3B808081 (0.0039215689) |
| 420 | v21 = shr_i32 v4 8 |
| 421 | v22 = bit_and v21 FF |
| 422 | v23 = to_f32 v22 |
| 423 | v24 = mul_f32 v23 3B808081 (0.0039215689) |
| 424 | v25 = mad_f32 v24 v13 v20 |
| 425 | v26 = mul_f32 v25 437F0000 (255) |
| 426 | v27 = round v26 |
| 427 | v28 = pack v16 v27 8 |
| 428 | v29 = shr_i32 v0 16 |
| 429 | v30 = bit_and v29 FF |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 430 | v31 = to_f32 v30 |
| 431 | v32 = mul_f32 v31 3B808081 (0.0039215689) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 432 | v33 = shr_i32 v4 16 |
| 433 | v34 = bit_and v33 FF |
| 434 | v35 = to_f32 v34 |
| 435 | v36 = mul_f32 v35 3B808081 (0.0039215689) |
| 436 | v37 = mad_f32 v36 v13 v32 |
| 437 | v38 = mul_f32 v37 437F0000 (255) |
| 438 | v39 = round v38 |
| 439 | v40 = shr_i32 v4 24 |
| 440 | v41 = bit_and v40 FF |
| 441 | v42 = to_f32 v41 |
| 442 | v43 = mul_f32 v42 3B808081 (0.0039215689) |
| 443 | v44 = mad_f32 v43 v13 v11 |
| 444 | v45 = mul_f32 v44 437F0000 (255) |
| 445 | v46 = round v45 |
| 446 | v47 = pack v39 v46 8 |
| 447 | v48 = pack v28 v47 16 |
| 448 | store32 arg(1) v48 |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 449 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 450 | 8 registers, 50 instructions: |
| 451 | 0 r0 = splat 3F800000 (1) |
Mike Klein | 8c1e0ef | 2019-11-12 09:07:23 -0600 | [diff] [blame] | 452 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 453 | 1 r1 = load32 arg(0) |
| 454 | 2 r2 = bit_and r1 FF |
| 455 | 3 r2 = to_f32 r2 |
| 456 | 4 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 457 | 5 r3 = load32 arg(1) |
| 458 | 6 r4 = bit_and r3 FF |
| 459 | 7 r4 = to_f32 r4 |
| 460 | 8 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 461 | 9 r5 = shr_i32 r1 24 |
| 462 | 10 r5 = bit_and r5 FF |
| 463 | 11 r5 = to_f32 r5 |
| 464 | 12 r5 = mul_f32 r5 3B808081 (0.0039215689) |
| 465 | 13 r6 = sub_f32 r0 r5 |
| 466 | 14 r2 = mad_f32 r4 r6 r2 |
| 467 | 15 r2 = mul_f32 r2 437F0000 (255) |
| 468 | 16 r2 = round r2 |
| 469 | 17 r4 = shr_i32 r1 8 |
| 470 | 18 r4 = bit_and r4 FF |
| 471 | 19 r4 = to_f32 r4 |
| 472 | 20 r4 = mul_f32 r4 3B808081 (0.0039215689) |
| 473 | 21 r7 = shr_i32 r3 8 |
| 474 | 22 r7 = bit_and r7 FF |
| 475 | 23 r7 = to_f32 r7 |
| 476 | 24 r7 = mul_f32 r7 3B808081 (0.0039215689) |
| 477 | 25 r4 = mad_f32 r7 r6 r4 |
| 478 | 26 r4 = mul_f32 r4 437F0000 (255) |
| 479 | 27 r4 = round r4 |
| 480 | 28 r4 = pack r2 r4 8 |
| 481 | 29 r1 = shr_i32 r1 16 |
| 482 | 30 r1 = bit_and r1 FF |
| 483 | 31 r1 = to_f32 r1 |
| 484 | 32 r1 = mul_f32 r1 3B808081 (0.0039215689) |
| 485 | 33 r2 = shr_i32 r3 16 |
| 486 | 34 r2 = bit_and r2 FF |
| 487 | 35 r2 = to_f32 r2 |
| 488 | 36 r2 = mul_f32 r2 3B808081 (0.0039215689) |
| 489 | 37 r1 = mad_f32 r2 r6 r1 |
| 490 | 38 r1 = mul_f32 r1 437F0000 (255) |
| 491 | 39 r1 = round r1 |
| 492 | 40 r3 = shr_i32 r3 24 |
| 493 | 41 r3 = bit_and r3 FF |
| 494 | 42 r3 = to_f32 r3 |
| 495 | 43 r3 = mul_f32 r3 3B808081 (0.0039215689) |
| 496 | 44 r5 = mad_f32 r3 r6 r5 |
| 497 | 45 r5 = mul_f32 r5 437F0000 (255) |
| 498 | 46 r5 = round r5 |
| 499 | 47 r5 = pack r1 r5 8 |
| 500 | 48 r5 = pack r4 r5 16 |
| 501 | 49 store32 arg(1) r5 |
Mike Klein | 267f507 | 2019-06-03 16:27:46 -0500 | [diff] [blame] | 502 | |
Mike Klein | 397fc88 | 2019-06-20 11:37:10 -0500 | [diff] [blame] | 503 | I32 (Naive) 8888 over 8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 504 | 34 values: |
Mike Klein | c2fb3b4 | 2019-07-17 12:09:09 -0500 | [diff] [blame] | 505 | v0 = load32 arg(0) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 506 | v1 = bit_and v0 FF |
| 507 | v2 = load32 arg(1) |
| 508 | v3 = bit_and v2 FF |
| 509 | v4 = shr_i32 v0 24 |
| 510 | v5 = bit_and v4 FF |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 511 | ↑ v6 = splat 100 (3.5873241e-43) |
| 512 | v7 = sub_i32 v6 v5 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 513 | v8 = mul_i32 v3 v7 |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 514 | v9 = shr_i32 v8 8 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 515 | v10 = add_i32 v1 v9 |
| 516 | v11 = shr_i32 v0 8 |
| 517 | v12 = bit_and v11 FF |
| 518 | v13 = shr_i32 v2 8 |
| 519 | v14 = bit_and v13 FF |
| 520 | v15 = mul_i32 v14 v7 |
| 521 | v16 = shr_i32 v15 8 |
| 522 | v17 = add_i32 v12 v16 |
| 523 | v18 = pack v10 v17 8 |
| 524 | v19 = shr_i32 v0 16 |
| 525 | v20 = bit_and v19 FF |
| 526 | v21 = shr_i32 v2 16 |
| 527 | v22 = bit_and v21 FF |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 528 | v23 = mul_i32 v22 v7 |
| 529 | v24 = shr_i32 v23 8 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 530 | v25 = add_i32 v20 v24 |
| 531 | v26 = shr_i32 v2 24 |
| 532 | v27 = bit_and v26 FF |
| 533 | v28 = mul_i32 v27 v7 |
| 534 | v29 = shr_i32 v28 8 |
| 535 | v30 = add_i32 v5 v29 |
| 536 | v31 = pack v25 v30 8 |
| 537 | v32 = pack v18 v31 16 |
| 538 | store32 arg(1) v32 |
Mike Klein | aab45b5 | 2019-07-02 15:39:23 -0500 | [diff] [blame] | 539 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 540 | 8 registers, 34 instructions: |
| 541 | 0 r0 = splat 100 (3.5873241e-43) |
Mike Klein | 397fc88 | 2019-06-20 11:37:10 -0500 | [diff] [blame] | 542 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 543 | 1 r1 = load32 arg(0) |
| 544 | 2 r2 = bit_and r1 FF |
| 545 | 3 r3 = load32 arg(1) |
| 546 | 4 r4 = bit_and r3 FF |
| 547 | 5 r5 = shr_i32 r1 24 |
| 548 | 6 r5 = bit_and r5 FF |
| 549 | 7 r6 = sub_i32 r0 r5 |
| 550 | 8 r4 = mul_i32 r4 r6 |
| 551 | 9 r4 = shr_i32 r4 8 |
| 552 | 10 r4 = add_i32 r2 r4 |
| 553 | 11 r2 = shr_i32 r1 8 |
| 554 | 12 r2 = bit_and r2 FF |
| 555 | 13 r7 = shr_i32 r3 8 |
| 556 | 14 r7 = bit_and r7 FF |
| 557 | 15 r7 = mul_i32 r7 r6 |
| 558 | 16 r7 = shr_i32 r7 8 |
| 559 | 17 r7 = add_i32 r2 r7 |
| 560 | 18 r7 = pack r4 r7 8 |
| 561 | 19 r1 = shr_i32 r1 16 |
| 562 | 20 r1 = bit_and r1 FF |
| 563 | 21 r4 = shr_i32 r3 16 |
| 564 | 22 r4 = bit_and r4 FF |
| 565 | 23 r4 = mul_i32 r4 r6 |
| 566 | 24 r4 = shr_i32 r4 8 |
| 567 | 25 r4 = add_i32 r1 r4 |
| 568 | 26 r3 = shr_i32 r3 24 |
| 569 | 27 r3 = bit_and r3 FF |
| 570 | 28 r6 = mul_i32 r3 r6 |
| 571 | 29 r6 = shr_i32 r6 8 |
| 572 | 30 r6 = add_i32 r5 r6 |
| 573 | 31 r6 = pack r4 r6 8 |
| 574 | 32 r6 = pack r7 r6 16 |
| 575 | 33 store32 arg(1) r6 |
Mike Klein | 397fc88 | 2019-06-20 11:37:10 -0500 | [diff] [blame] | 576 | |
Mike Klein | 7b7077c | 2019-06-03 17:10:59 -0500 | [diff] [blame] | 577 | I32 8888 over 8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 578 | 28 values: |
Mike Klein | c2fb3b4 | 2019-07-17 12:09:09 -0500 | [diff] [blame] | 579 | v0 = load32 arg(0) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 580 | v1 = bit_and v0 FF |
| 581 | v2 = load32 arg(1) |
| 582 | v3 = bit_and v2 FF |
| 583 | v4 = shr_i32 v0 24 |
| 584 | ↑ v5 = splat 100 (3.5873241e-43) |
| 585 | v6 = sub_i32 v5 v4 |
| 586 | v7 = mul_i16x2 v3 v6 |
| 587 | v8 = shr_i32 v7 8 |
| 588 | v9 = add_i32 v1 v8 |
| 589 | v10 = bytes v0 2 |
| 590 | v11 = bytes v2 2 |
| 591 | v12 = mul_i16x2 v11 v6 |
| 592 | v13 = shr_i32 v12 8 |
| 593 | v14 = add_i32 v10 v13 |
| 594 | v15 = pack v9 v14 8 |
| 595 | v16 = bytes v0 3 |
| 596 | v17 = bytes v2 3 |
| 597 | v18 = mul_i16x2 v17 v6 |
| 598 | v19 = shr_i32 v18 8 |
| 599 | v20 = add_i32 v16 v19 |
| 600 | v21 = shr_i32 v2 24 |
| 601 | v22 = mul_i16x2 v21 v6 |
| 602 | v23 = shr_i32 v22 8 |
| 603 | v24 = add_i32 v4 v23 |
| 604 | v25 = pack v20 v24 8 |
| 605 | v26 = pack v15 v25 16 |
| 606 | store32 arg(1) v26 |
Mike Klein | aab45b5 | 2019-07-02 15:39:23 -0500 | [diff] [blame] | 607 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 608 | 8 registers, 28 instructions: |
| 609 | 0 r0 = splat 100 (3.5873241e-43) |
Mike Klein | 754bad3 | 2019-06-05 10:47:46 -0500 | [diff] [blame] | 610 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 611 | 1 r1 = load32 arg(0) |
| 612 | 2 r2 = bit_and r1 FF |
| 613 | 3 r3 = load32 arg(1) |
| 614 | 4 r4 = bit_and r3 FF |
| 615 | 5 r5 = shr_i32 r1 24 |
| 616 | 6 r6 = sub_i32 r0 r5 |
| 617 | 7 r4 = mul_i16x2 r4 r6 |
| 618 | 8 r4 = shr_i32 r4 8 |
| 619 | 9 r4 = add_i32 r2 r4 |
| 620 | 10 r2 = bytes r1 2 |
| 621 | 11 r7 = bytes r3 2 |
| 622 | 12 r7 = mul_i16x2 r7 r6 |
| 623 | 13 r7 = shr_i32 r7 8 |
| 624 | 14 r7 = add_i32 r2 r7 |
| 625 | 15 r7 = pack r4 r7 8 |
| 626 | 16 r1 = bytes r1 3 |
| 627 | 17 r4 = bytes r3 3 |
| 628 | 18 r4 = mul_i16x2 r4 r6 |
| 629 | 19 r4 = shr_i32 r4 8 |
| 630 | 20 r4 = add_i32 r1 r4 |
| 631 | 21 r3 = shr_i32 r3 24 |
| 632 | 22 r6 = mul_i16x2 r3 r6 |
| 633 | 23 r6 = shr_i32 r6 8 |
| 634 | 24 r6 = add_i32 r5 r6 |
| 635 | 25 r6 = pack r4 r6 8 |
| 636 | 26 r6 = pack r7 r6 16 |
| 637 | 27 store32 arg(1) r6 |
Mike Klein | 821f5e8 | 2019-06-13 10:56:51 -0500 | [diff] [blame] | 638 | |
| 639 | I32 (SWAR) 8888 over 8888 |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 640 | 14 values: |
Mike Klein | c2fb3b4 | 2019-07-17 12:09:09 -0500 | [diff] [blame] | 641 | v0 = load32 arg(0) |
| 642 | v1 = bytes v0 404 |
Mike Klein | 5e533c9 | 2019-07-22 13:44:54 -0500 | [diff] [blame] | 643 | ↑ v2 = splat 1000100 (2.3510604e-38) |
Mike Klein | c2fb3b4 | 2019-07-17 12:09:09 -0500 | [diff] [blame] | 644 | v3 = sub_i16x2 v2 v1 |
| 645 | v4 = load32 arg(1) |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 646 | v5 = bit_and v4 FF00FF |
| 647 | v6 = mul_i16x2 v5 v3 |
| 648 | v7 = shr_i16x2 v6 8 |
| 649 | v8 = shr_i16x2 v4 8 |
| 650 | v9 = mul_i16x2 v8 v3 |
| 651 | v10 = bit_and v9 FF00FF00 |
| 652 | v11 = bit_or v7 v10 |
| 653 | v12 = add_i32 v0 v11 |
| 654 | store32 arg(1) v12 |
Mike Klein | aab45b5 | 2019-07-02 15:39:23 -0500 | [diff] [blame] | 655 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 656 | 5 registers, 14 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 657 | 0 r0 = splat 1000100 (2.3510604e-38) |
Mike Klein | 821f5e8 | 2019-06-13 10:56:51 -0500 | [diff] [blame] | 658 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 659 | 1 r1 = load32 arg(0) |
| 660 | 2 r2 = bytes r1 404 |
| 661 | 3 r2 = sub_i16x2 r0 r2 |
| 662 | 4 r3 = load32 arg(1) |
| 663 | 5 r4 = bit_and r3 FF00FF |
| 664 | 6 r4 = mul_i16x2 r4 r2 |
| 665 | 7 r4 = shr_i16x2 r4 8 |
| 666 | 8 r3 = shr_i16x2 r3 8 |
| 667 | 9 r2 = mul_i16x2 r3 r2 |
| 668 | 10 r2 = bit_and r2 FF00FF00 |
| 669 | 11 r2 = bit_or r4 r2 |
| 670 | 12 r2 = add_i32 r1 r2 |
| 671 | 13 store32 arg(1) r2 |
Mike Klein | 7b7077c | 2019-06-03 17:10:59 -0500 | [diff] [blame] | 672 | |
Mike Klein | f996311 | 2019-08-08 15:13:25 -0400 | [diff] [blame] | 673 | 6 values: |
Mike Klein | 0f61c12 | 2019-10-16 10:46:01 -0500 | [diff] [blame] | 674 | ↟ v0 = splat 1 (1.4012985e-45) |
| 675 | ↟ v1 = splat 2 (2.8025969e-45) |
Mike Klein | f996311 | 2019-08-08 15:13:25 -0400 | [diff] [blame] | 676 | ↑ v2 = add_i32 v0 v1 |
| 677 | v3 = load32 arg(0) |
| 678 | v4 = mul_i32 v3 v2 |
| 679 | store32 arg(0) v4 |
| 680 | |
| 681 | 2 registers, 6 instructions: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 682 | 0 r0 = splat 1 (1.4012985e-45) |
| 683 | 1 r1 = splat 2 (2.8025969e-45) |
| 684 | 2 r1 = add_i32 r0 r1 |
Mike Klein | f996311 | 2019-08-08 15:13:25 -0400 | [diff] [blame] | 685 | loop: |
Mike Klein | b5c4355 | 2020-01-07 11:39:30 -0600 | [diff] [blame] | 686 | 3 r0 = load32 arg(0) |
| 687 | 4 r0 = mul_i32 r0 r1 |
| 688 | 5 store32 arg(0) r0 |
Mike Klein | f996311 | 2019-08-08 15:13:25 -0400 | [diff] [blame] | 689 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 690 | 24 values: |
| 691 | v0 = load32 arg(0) |
| 692 | v1 = bit_and v0 FF |
| 693 | v2 = load32 arg(1) |
| 694 | v3 = bit_and v2 FF |
| 695 | v4 = add_i32 v1 v3 |
| 696 | v5 = shr_i32 v0 8 |
| 697 | v6 = bit_and v5 FF |
| 698 | v7 = shr_i32 v2 8 |
| 699 | v8 = bit_and v7 FF |
| 700 | v9 = add_i32 v6 v8 |
| 701 | v10 = pack v4 v9 8 |
| 702 | v11 = shr_i32 v0 16 |
| 703 | v12 = bit_and v11 FF |
| 704 | v13 = shr_i32 v2 16 |
| 705 | v14 = bit_and v13 FF |
| 706 | v15 = add_i32 v12 v14 |
| 707 | v16 = shr_i32 v0 24 |
| 708 | v17 = bit_and v16 FF |
| 709 | v18 = shr_i32 v2 24 |
| 710 | v19 = bit_and v18 FF |
| 711 | v20 = add_i32 v17 v19 |
| 712 | v21 = pack v15 v20 8 |
| 713 | v22 = pack v10 v21 16 |
| 714 | store32 arg(1) v22 |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 715 | |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 716 | 5 registers, 24 instructions: |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 717 | loop: |
Mike Klein | a6434a5 | 2020-01-08 14:06:52 -0600 | [diff] [blame^] | 718 | 0 r0 = load32 arg(0) |
| 719 | 1 r1 = bit_and r0 FF |
| 720 | 2 r2 = load32 arg(1) |
| 721 | 3 r3 = bit_and r2 FF |
| 722 | 4 r3 = add_i32 r1 r3 |
| 723 | 5 r1 = shr_i32 r0 8 |
| 724 | 6 r1 = bit_and r1 FF |
| 725 | 7 r4 = shr_i32 r2 8 |
| 726 | 8 r4 = bit_and r4 FF |
| 727 | 9 r4 = add_i32 r1 r4 |
| 728 | 10 r4 = pack r3 r4 8 |
| 729 | 11 r3 = shr_i32 r0 16 |
| 730 | 12 r3 = bit_and r3 FF |
| 731 | 13 r1 = shr_i32 r2 16 |
| 732 | 14 r1 = bit_and r1 FF |
| 733 | 15 r1 = add_i32 r3 r1 |
| 734 | 16 r0 = shr_i32 r0 24 |
| 735 | 17 r0 = bit_and r0 FF |
| 736 | 18 r2 = shr_i32 r2 24 |
| 737 | 19 r2 = bit_and r2 FF |
| 738 | 20 r2 = add_i32 r0 r2 |
| 739 | 21 r2 = pack r1 r2 8 |
| 740 | 22 r2 = pack r4 r2 16 |
| 741 | 23 store32 arg(1) r2 |
Mike Klein | d48488b | 2019-10-22 12:27:58 -0500 | [diff] [blame] | 742 | |