High-performance, high-precision floating point library
- Intel Intrinsics Guide, filtered with extensions that exist on most gamers' computers based on the Steam Hardware Survey
- C++ Optimization Strategies and Techniques (slightly outdated)
- Coding Practices which enable the compiler/optimizer to make a faster program (StackOverflow)
- What Every Computer Scientist Should Know About Floating-Point Arithmetic
Likely would be used when sending over networks and such.
| Field | Start | End | Count | Notes |
|---|---|---|---|---|
| Exponent | 79 | 64 | 16 | Bias of 1023 (same as IEEE double precision) |
| Sign | 63 | 63 | 1 | 0 is positive, 1 is negative |
| Mantissa | 62 | 0 | 63 | Technically 64 bits, with a 1 for the MSB |
| Total | 79 | 0 | 80 | This is 10 bytes, or 2 more than a double. |