implemented pair_symmetry that uses a LUT to calculate the distances

2026-05-20 10:29:17 +02:00
parent e37300134b
commit 045dadba5f
3 changed files with 410 additions and 0 deletions
@@ -35,6 +35,7 @@ System Version: IdeaPad Slim 5 14AKP10
 | --- | --------------------------------------- | ---------------- | -------------- | ------------------------- | ------ |
 | 0   | [Baseline](optimization_baseline.md)    | 1'058'400'000.00 | 850'352'900.00 | 1.24                      |        |
 | 1   | [Inlining](optimization_inlining.md)    | 1'058'400'000.00 | 756'088'260.00 | 1.40                      | Elliot |
+| 2   | [Symmetry](optimization_symmetry.md)    |                  |                |                           | Smail  |

 ### ARM (Apple)

@@ -0,0 +1,90 @@
+# Optimization: Inlining
+
+- **Author**: Smail
+- **Predecessor**: variant_inlining_7.cpp
+- **File(s)**: variants/variant_inlining_7_symmetry_1.cpp
+- **Status**: TBA
+
+## What this changes
+
+Instead of calculating the distance and norm vector between two pedestrians everytime,
+we created a LUT that only calculates the distances once and r_a_b_norm (with expensive sqrt flop `dx`, `dy`, `norm`).
+The forces are somewhat symmetric, however it would still require a lot of calculations,
+maybe this second symmetry optimization can be done at a later stage.
+
+When i > j we need the opposite vector which is $curr * -1$.
+
+
+## Code (relevant snippet)
+
+TBD - see `variants/variant_inlining_7_symmetry_1.cpp` for the final version, and the
+individual file headers for the incremental steps.
+
+## Benchmark
+
+Conditions: single run (TODO: switch to min-of-N), N_ITER=1000,
+-O3 -march=native -flto. Predecessor reference for the same config is the
+baseline row.
+
+### x86
+
+**Machine**:
+
+```
+Processors: 16 x AMD Ryzen AI 7 350 w/ Radeon 860M
+Memory: 32 GiB of RAM (30.6 GiB usable)
+Graphics Processor: AMD Radeon 860M Graphics
+Manufacturer: LENOVO
+Product Name: 83HX
+System Version: IdeaPad Slim 5 14AKP10
+```
+
+**Frequency**: 2 GHz
+
+Existing single-point measurement (N unspecified, kept from the previous
+OPTIMIZATIONS.md; W matches roughly N=120 at N_ITER=1000):
+
+| Variant  | W(n) [Flops]     | T(n) [Cycles]  | Performance [Flops/Cycle] |
+| -------- | ---------------- | -------------- | ------------------------- |
+| Baseline | 1'058'400'000.00 | 850'352'900.00 | 1.24                      |
+| Inlining | 1'058'400'000.00 | 756'088'260.00 | 1.40                      |
+| Symmetry |  |  |                       |
+
+Inlining is about 12.9% faster on this point. A full N sweep with min-of-3 is TODO.
+
+Full sweep (TODO):
+
+| N    | W(n) [Flops] | T(n) [Cycles] | Performance [Flops/Cycle] | Diff vs Predecessor |
+| ---- | ------------ | ------------- | ------------------------- | ------------------- |
+| 50   |              |               |                           |                     |
+| 100  |              |               |                           |                     |
+| 200  |              |               |                           |                     |
+| 500  |              |               |                           |                     |
+| 1000 |              |               |                           |                     |
+
+### ARM (Apple)
+
+**Machine**:
+
+```
+TBD
+```
+
+**Frequency**: TBD
+
+| N    | W(n) [Flops] | T(n) [Cycles] | Performance [Flops/Cycle] | Diff vs Predecessor |
+| ---- | ------------ | ------------- | ------------------------- | ------------------- |
+| 50   |              |               |                           |                     |
+| 100  |              |               |                           |                     |
+| 200  |              |               |                           |                     |
+| 500  |              |               |                           |                     |
+| 1000 |              |               |                           |                     |
+
+## Analysis
+
+Speedup likely comes mainly from halving the number of `sqrt` flops.
+
+
+## Takeaway
+
+TBD