| # | User | Rating |
|---|---|---|
| 1 | ecnerwala | 3649 |
| 2 | Benq | 3581 |
| 3 | jiangly | 3578 |
| 4 | orzdevinwang | 3570 |
| 5 | Geothermal | 3569 |
| 5 | cnnfls_csy | 3569 |
| 7 | tourist | 3565 |
| 8 | maroonrk | 3531 |
| 9 | Radewoosh | 3521 |
| 10 | Um_nik | 3482 |
| # | User | Contrib. |
|---|---|---|
| 1 | maomao90 | 174 |
| 2 | awoo | 164 |
| 3 | adamant | 161 |
| 4 | TheScrasse | 159 |
| 5 | nor | 158 |
| 6 | maroonrk | 156 |
| 7 | -is-this-fft- | 152 |
| 8 | SecondThread | 147 |
| 9 | orz | 146 |
| 10 | pajenegod | 145 |
Does anyone know why __int128 seems to use 64-bit operations instead of SIMD instructions?
In this demo, 64-bit instructions are used but switching __int128 with any smaller type (int, long long, etc.) enables vectorization + unrolling.
Is using 64-bit operations just faster than loading into/out of the 128-bit registers?
There is no SIMD instruction for 128-bit addition.
Thanks!