RFR: 8349457: [leyden] Workaround Leyden problems with -XX:-TieredCompilation

[Aleksey Shipilev]

On Wed, 5 Feb 2025 12:14:56 GMT, Aleksey Shipilev <shade at openjdk.org> wrote:

> Currently, supplying `-XX:-TieredCompilation` would break with Leyden workflows, because the archive-backed code cache heap would only be created if `SegmentedCodeCache` is enabled. With default compilation policy, this happens normally when tiered compilation is enabled, but not without it.
> 
> There are other minor peculiarities that we need to handle, which are exposed with `runtime/cds` + `-XX:-TieredCompilation` runs.
> 
> This prevents us from testing C2-only Leyden configs and studying compilation dynamics without tiered policy getting in the way.
> 
> Additional testing:
>  - [x] Ad-hoc Leyden experiments now work with `-XX:-TieredCompilation`: the C2-only code is being stored and loaded successfully from the CDS archive
>  - [x] `runtime/cds` now passes with `-XX:-TieredCompilation`
>  - [x] `runtime/cds` still passes by default (i.e. with tiered compilation)

Fun perf results:


$ build/linux-x86_64-server-release/images/jdk/bin/java -Xms64m -Xmx1g -XX:+UseParallelGC \
     -XX:CacheDataStore=JavacBenchApp.cds -cp JavacBenchApp.jar JavacBenchApp 50

# --- Training with C1 and C2 (+Tiered)

# Production with C1 and C2 (+Tiered)
  Time (mean ± σ):     434.0 ms ±   8.5 ms    [User: 1506.7 ms, System: 218.4 ms]
  Range (min … max):   423.4 ms … 449.2 ms    10 runs

# Production with C2 only (-Tiered)
  Time (mean ± σ):     561.2 ms ±  12.8 ms    [User: 3578.0 ms, System: 263.2 ms]
  Range (min … max):   542.3 ms … 589.2 ms    10 runs

# --- Training with C2 only (-Tiered)

# Production with C1 and C2 (+Tiered)
  Time (mean ± σ):     408.0 ms ±   5.4 ms    [User: 1183.2 ms, System: 165.2 ms]
  Range (min … max):   401.3 ms … 421.0 ms    10 runs

# Production with C2 only (-Tiered)
  Time (mean ± σ):     482.0 ms ±  11.6 ms    [User: 1959.1 ms, System: 182.5 ms]
  Range (min … max):   467.2 ms … 508.2 ms    10 runs

-------------

PR Comment: https://git.openjdk.org/leyden/pull/32#issuecomment-2640333261