Skip to content

docs: record kDisco PR #52 Kotlin 2.3.21 toolchain upgrade as performance-neutral for fast-sim #761

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Copilot wants to merge 2 commits into
base: toKdisco0.6.1
Choose a base branch
from
+36 −0
Open

docs: record kDisco PR #52 Kotlin 2.3.21 toolchain upgrade as performance-neutral for fast-sim #761

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
1a59808
Initial plan
Copilot Jul 12, 2026
4d69adc
docs: record kDisco PR #52 Kotlin 2.3.21 evaluation as performance-ne…
Copilot Jul 12, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 36 additions & 0 deletions docs/FAST_SIM_BENCHMARK.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -99,3 +99,39 @@ The native `fast-sim` binary delivers substantial improvements for CLI simulatio

These characteristics make the native binary ideal for scripting, CI/CD pipelines, batch
simulation runs, and environments where JVM installation is undesirable.

---

## Supplementary: kDisco Kotlin Toolchain Evaluation (Issue #752)

**Date:** 2026-07-11
**Question:** Does upgrading kDisco's Kotlin `2.1.10 → 2.3.21` (kDisco PR [bedaHovorka/kdisco#52](https://github.com/bedaHovorka/kdisco/pull/52)) speed up `fast-sim`?

### Method

Published kDisco PR #52's branch to `mavenLocal` as `0.6.0-k2321-SNAPSHOT`, pointed a throwaway
interlockSim worktree at it, rebuilt the native `linuxX64` release binary, measured
`example shuntingLoop 300` (median of 7 runs, 5 trains, 211.1 s sim time).

### Result: **performance-neutral**

| kDisco build | Kotlin version | Wall-clock (median of 7) |
|---|---|---|
| 0.6.0 (current at evaluation) | 2.1.10 | **341 ms** |
| 0.6.0-k2321-SNAPSHOT (PR #52) | 2.3.21 | **349 ms** |

The 8 ms gap is **inside run-to-run noise** (±7%; the same binary measured 307–329 ms and
336–348 ms across repeated batches). There is no measurable performance difference in either direction.

### Conclusion

kDisco PR #52 is **neither an optimisation nor a pessimisation** for `fast-sim`. It should be
merged on toolchain-currency grounds rather than performance grounds.

Caveat on scope: this rebuilds only the *kDisco library* with 2.3.21. The dominant hot code
(`Train.Motor.derivatives`, ~1.85M calls/run, see [#750](https://github.com/bedaHovorka/interlockSim/issues/750))
lives in interlockSim and is compiled by interlockSim's own Kotlin version. The RKF45 step-count
problem (#750) dominates by two orders of magnitude; codegen changes are not the lever for that issue.

kDisco CI run for the `to0.6.1` branch (which includes the PR #52 changes):
[bedaHovorka/kdisco/actions/runs/29184671509](https://github.com/bedaHovorka/kdisco/actions/runs/29184671509) — ✅ **passed**
Loading