Complete maths overhaul
Overhaul the fractal definition to some sort of formula-driven system. This would make it easier to experiment quickly.
That would also be a good time to investigate deeper plots (fixed-point assembler, GMP, possibly other solutions).
Blueprint information
- Status:
- Started
- Approver:
- None
- Priority:
- Medium
- Drafter:
- None
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- New
- Series goal:
- None
- Implementation:
-
Needs Infrastructure
- Milestone target:
- None
- Started by
- Ross Younger
- Completed by
Related branches
Related bugs
| Bug #783077: More fractals | Confirmed |
| Bug #783092: Allow deeper plots | Confirmed |
| Bug #872722: Fractal formula system | Confirmed |
Sprints
Whiteboard
The need for speed is difficult to satisfy in a single process. Some sort of language compiled at runtime is the "obvious" way to do this, but such things can be quite hard.
Python might be a convenient way for experimenting with relatively fast feedback, it'll be interesting to see what speed it gives us (esp. with optimisation turned on). Some JIT-compiled code of some sort (or even assembler) is likely to be much faster.
1. Overhaul underlying maths to allow deeper plots (bug 783092)
2. Formula system (bug 872722)
Graal & Truffle worth looking in to: https:/
NOTE that perturbation theory would allow us to work with much fewer bignum ops than previously, so do that first (and see what the new glitch limit is like) before looking too deeply into a complete maths overhaul.
