Intel64 Multi-Precision Arithmetic
Eric Bainville - Dec 2006These pages relate my experiments on 64-bit multi-precision arithmetic running on Core 2 Duo processors. See the introduction for further details.
Here are the best timings I could get so far:
| Function | glibc | gmp | this page | Z ← 0 | 1.00 | - | 0.50 | Z ← op Z (not) | - | - | 0.50 | Z ← op Z (neg) | - | - | 1.00 | Z ← X | 1.75 | - | 1.50 | Z ← Z op X (and or xor) | - | - | 1.63 | Z ← Z op X (+ -) | - | 2.63 | 2.53 | Z ← Z op not X (and or xor) | - | - | - | Z ← Z * k | - | 4.00 | - | Z ← Z * k op X (+ -) | - | 4.54 | - | Z ← Z << k | - | 2.40 | - | Z ← Z / k | - | 33.29 | - |
glibc refers to glibc 4.2, and gmp refers to gmp 4.2.1 with Jason W. Martin's patch.
 AMD64 Multi-Precision |  Top of Page |  Intel64 Multi-Precision : Memory Zero |
