bench(gradle): harden JMH suite against run-to-run variance

[not-matthias]

1. forceGC = true in the Gradle jmh block — System.gc() between
   iterations so a GC pause can't land mid-measurement.
2. -Xbatch JVM arg — synchronous JIT compilation so the C2 background
   thread can't steal cycles during measurement.
3. Trim Param matrices across DP / Bit / Rle / Regex benchmarks to a
   single representative value each.
4. SortBenchmark trim + allocation fix — drop 5 redundant sort variants,
   keep one size, replace Arrays.copyOf (fresh Integer[] per invocation)
   with System.arraycopy into a pre-allocated working buffer. This is the
   direct fix for the timSort[100] flakiness called out in the ticket.
5. BacktrackingBenchmark trim + allocation hoist — reduce each Param to 2
   values and move Integer[] / ArrayList allocations out of the timed
   region via Setup.
6. Tune Warmup(10) / Measurement(10) / Fork(1) on all 8 benchmark classes.
   Generous warmup reaches JIT steady state; generous measurement iterations
   give tight per-fork confidence intervals. Fork(1) leaves cross-JVM variance
   sampling to the 6-distribution CI matrix.

[codspeed-hq]

Merging this PR will degrade performance by 14.35%

⚡ 12 improved benchmarks
❌ 2 regressed benchmarks
✅ 28 untouched benchmarks
⏩ 92 skipped benchmarks¹

⚠️ Please fix the performance issues or acknowledge them on CodSpeed.

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	generateCombinations[9]	28.6 µs	24 µs	+19.3%
❌	fib[30]	8.4 ms	9.4 ms	-11.22%
⚡	permutations[7]	554.3 µs	451.8 µs	+22.7%
⚡	mergeSort[10000]	2.9 ms	2.2 ms	+35.3%
❌	levenshteinDistance[saturday sunday]	400 ns	467 ns	-14.35%
⚡	dualPivotQuickSort[10000]	4.9 ms	3.3 ms	+48.21%
⚡	quickSort[10000]	2.5 ms	2.2 ms	+13.72%
⚡	generateCombinations[7]	12.6 µs	11 µs	+14.35%
⚡	fibonacciOptimized[30]	31 ns	27 ns	+14.81%
⚡	fibonacciBottomUp[30]	2.7 µs	2.2 µs	+23.65%
⚡	permutations[5]	12.8 µs	11 µs	+16.9%
⚡	multiPatternScan[24]	6.4 ms	5.3 ms	+19.29%
⚡	compileAndMatch[24]	4 ms	3.6 ms	+11.27%
⚡	timSort[10000]	2.4 ms	1.6 ms	+48.37%

---

_{Comparing cod-2519-harden-codspeed-jvm-against-regressions (dacf718) with main (c7c19aa)}

Footnotes

1. 92 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[Copilot]

Pull request overview

This PR hardens the example Gradle JMH benchmark suite to reduce run-to-run variance in CI by adjusting JMH configuration, trimming parameter matrices, and removing allocation noise from timed regions.

Changes:

• Increase warmup/measurement iterations (10/10) and standardize on @Fork(1) across benchmarks.
• Trim benchmark @Param matrices to fewer representative values to reduce CI wall-clock and reduce noise.
• Remove per-invocation allocations from hot benchmark paths (notably SortBenchmark and parts of BacktrackingBenchmark) and add JMH JVM/GC stabilization knobs in Gradle.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 7 comments.

Show a summary per file

File	Description
examples/example-gradle/build.gradle.kts	Enable forceGC and add -Xbatch to reduce GC/JIT jitter during measurement.
examples/example-gradle/src/jmh/java/com/thealgorithms/sorts/SortBenchmark.java	Reduce param sizes, drop redundant sort variants, and replace per-invocation Arrays.copyOf with System.arraycopy into a reusable buffer.
examples/example-gradle/src/jmh/java/bench/SleepBenchmark.java	Increase warmup/measurement iterations for stability.
examples/example-gradle/src/jmh/java/bench/RleBenchmark.java	Increase warmup/measurement iterations and reduce input-size params.
examples/example-gradle/src/jmh/java/bench/RegexBenchmark.java	Increase warmup/measurement iterations and reduce backtracking param matrix.
examples/example-gradle/src/jmh/java/bench/FibBenchmark.java	Increase warmup/measurement iterations.
examples/example-gradle/src/jmh/java/bench/DynamicProgrammingBenchmark.java	Increase warmup/measurement iterations and trim multiple DP benchmark params.
examples/example-gradle/src/jmh/java/bench/BitManipulationBenchmark.java	Increase warmup/measurement iterations and trim bit-value params.
examples/example-gradle/src/jmh/java/bench/BacktrackingBenchmark.java	Increase warmup/measurement iterations, trim params, and hoist allocations into @Setup.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[GuillaumeLagrange]

OLGTM but I have two remarks

1. Do we have stats of the variance before/after changes in order to make sure we're not throwing things around randomly and hope it improves stuff?
2. The warmup and iterations required for our benchmarks to be less noisy should be somewhere in the docs as guidelines, because people ARE going to report unreliable results, as they should, if we encountered them ourselves