1. Achieves near-perfect balancing on non-pathological data
by using a per-node pivot and calculating.
2. Increases the depth to 5 levels, which seems to be the
sweet spot.
New benchmark vs baseline:
```
Comparing build-reld-palette-cleanup/palette_blending_benchmark to build-reld/palette_blending_benchmark
Benchmark Time CPU Time Old Time New CPU Old CPU New
-----------------------------------------------------------------------------------------------------------------------------------
BM_GenerateBlendedLookupTable_pvalue 0.0002 0.0002 U Test, Repetitions: 10 vs 10
BM_GenerateBlendedLookupTable_mean -0.8768 -0.8767 18432956 2270752 18417846 2270141
BM_GenerateBlendedLookupTable_median -0.8767 -0.8767 18421978 2270802 18417838 2270167
BM_GenerateBlendedLookupTable_stddev -0.9860 -0.8051 33641 473 1222 238
BM_GenerateBlendedLookupTable_cv -0.8860 +0.5815 0 0 0 0
OVERALL_GEOMEAN -0.8768 -0.8767 0 0 0 0
```
Uses a k-d tree to quickly find the best match
for a color when generating the palette blending
lookup table.
https://en.wikipedia.org/wiki/K-d_tree
This is 3x faster than the previous naive approach:
```
Benchmark Time CPU Time Old Time New CPU Old CPU New
-----------------------------------------------------------------------------------------------------------------------------------
BM_GenerateBlendedLookupTable_pvalue 0.0002 0.0002 U Test, Repetitions: 10 vs 10
BM_GenerateBlendedLookupTable_mean -0.7153 -0.7153 18402641 5239051 18399111 5238025
BM_GenerateBlendedLookupTable_median -0.7153 -0.7153 18403261 5239042 18398841 5237497
BM_GenerateBlendedLookupTable_stddev -0.2775 +0.3858 2257 1631 1347 1867
BM_GenerateBlendedLookupTable_cv +1.5379 +3.8677 0 0 0 0
OVERALL_GEOMEAN -0.7153 -0.7153 0 0 0 0
```
The distribution is somewhat poor with just 3 levels, so this can be improved further.
For example, here is the leaf size distribution in the cathedral:
```
r0.g0.b0: 88
r0.g0.b1: 10
r0.g1.b0: 2
r0.g1.b1: 32
r1.g0.b0: 27
r1.g0.b1: 4
r1.g1.b0: 12
r1.g1.b1: 81
```
The previous implementation didn't behave quite like A-* is supposed to.
After trying to figure out what's causing it and giving up,
I've reimplemented it in a straightforward manner.
Now it seems to work a lot better.
Also increases maximum player path length to 100 steps.
We still only store the first 25 steps in the save file for vanilla
compatibility.
In C++, globals initialization order accross translation units is not
defined. Accessing a global via a function ensures that it is initialized.
This will be needed for #7638, which will statically initialize change
handlers after the Options object has been initialized.
1. Moves more assets-related stuff from `init` to `engine/assets`.
2. Removes `SDL_audiolib` dependency from `soundsample.h`.
3. Cleans up some unused/missing includes.
Our implementation has a more modern interface and only
supports the features that we care about.
It always outputs `\n` as newlines and does not output BOM.
The modern interface eliminates awkward `c_str()/data()` conversions.
This implementation preserves comments and the file order of sections
and keys. New keys are written in insertion order.
We now also support modifying and adding default comments,
which may be a useful thing to do for the especially tricky
ini options (this PR doesn't add any but adds the ability to do so).
Sadly, this increases the RG99 binary size by 24 KiB.
I'm guessing this is because the map implementation generates
quite a bit of code.
Note that while it might seem that using `std::string` for every key and
value would do a lot of allocations, most of these strings are
small and thus benefit from Small String Optimization (= no allocations).
Looks like the KallistiOS Dreamcast SDK disables double support
by default: 495e77fd60/environ_dreamcast.sh (L16)
We don't really need doubles in this code.
The one place where we might have needed them is the SMK video
decoder, handled in a separate PR.
This is enough to make it usable as a backing storage for a
`priority_queue`. Had it in one of my branches and figured it might come
in handy in the future.
This gives a very slight FPS boost.
1140 to 1143 FPS on my machine as measured by:
```bash
tools/linux_reduced_cpu_variance_run.sh tools/measure_timedemo_performance.py -n 5 --binary build-rel/devilutionx
```
Gleb Mazovetskiy
staphen
obligaron
Peppersawce
Eric Robinson
Eric Robinson
Yggdrasill
Anders Jenbo