RealWorldNativePluginPerformanceTest.groovy

Clone Tools
  • last updated a few seconds ago
Constraints
Constraints: committers
 
Constraints: files
Constraints: dates
Lock in recent performance improvements

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline performance tests

We introduced a ThreadLocal in the project locking which slowed things

down a little bit (10s of ms) in several builds. Rebaselining to prevent

this from affecting all branches.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline performance tests after merging single project lock changes

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline performance tests

The introduction of the "platform" support introduced a couple small regressions,

but also improvements in some cases. This commit re-enables the Java IDE performance

test now that we have a baseline to compare with. Work on improving performance

is going to happen later.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline RealWorldNativePluginPerformanceTest

All scenarios seem to be slower on `master` now. The one with the

biggest regression is

`build on nativeMonolithic with 0 parallel workers`.

That regressed about 3%.

Let's investigate!

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
Improve performance for getRegularFileSnapshot (#6199)

Only intern on demand and revert "Rebaseline build on nativeMonolithicOverlapping".

    • -1
    • +0
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 1 more file in changeset.
Rebaseline build on nativeMonolithicOverlapping

It looks like the tests with 0 workers got slower (around 4% for

`nativeMonolithicOverlapping` and 1% for `nativeMonolithic`).

The builds with 12 parallel workers seem to be as fast as the

baseline.

Rebaselining so I can investigate the regression.

    • -0
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
Refactor duplicate handling to mitigate performance issues

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 4 more files in changeset.
Temporarily rebaseline performance test until we can mitigate the problem

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
Rebaseline all performance tests

To lock in memory usage improvements.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 32 more files in changeset.
Rebase native performance test on latest nightly

There was quite a bit of performance improvement lately, so we are

locking down those improvement for future versions.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 5 more files in changeset.
Lock in some dependency management performance improvements

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline performance tests

There is a known performance regression due to more work done during dependency resolution. Future

commits will attempt to mitigate the regression.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Rebaseline performance tests

To account for a regression across most tests,

which was caused by a bugfix for lazy task creation.

Lazy task creation was only that fast because of

that bug, so this new performance level is the

expected one. We can definitely improve it further,

see https://github.com/gradle/gradle-native/issues/678

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 30 more files in changeset.
Rebaseline RealWorldNativePluginPerformanceTest

This keeps failing but no obvious regression observed. Seems to be a gradual

regression since long time ago.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
Rebaseline performance tests

This change accepts a 1% regression in the

ProjectCreationPerformanceTest."create many empty projects" test.

The original baseline contains a bug that accidentaly made the

execution faster.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 31 more files in changeset.
Unignore some performance tests.

    • -2
    • +0
    ./RealWorldNativePluginPerformanceTest.groovy
Rebaseline performance tests

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Ignore native performance test until it can be re-baselined.

    • -0
    • +2
    ./RealWorldNativePluginPerformanceTest.groovy
Rebaseline performance tests

To account for recent configuration time improvements.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Lock in some performance improvements

There have been several performance improvements both in dependency management and thanks

to software model bridging.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Rebaseline performance tests

- lock in recent native build improvements

- accept a minor regression in dependency resolution that will be fixed soon

- inline native baseline constant. Please use ./gradlew rebaselinePerformanceTests instead

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Re-enable native performance tests with a 4.6 baseline

    • -3
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 5 more files in changeset.
Replace honest profiler with JFR

JFR is available on all systems that have an Oracle JDK, Honest profiler only works on Linux.

JFR profiles CPU, memory, IO, locking and more, Honest profiler only profiles CPU.

Both of them use the same low-overhead async profiling API.

This PR also inlines the code necessary for flamegraph generation.

Users will no longer need any additional setup on their machine.

    • -21
    • +0
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 12 more files in changeset.
Rebaseline all performance tests

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Rebaseline all performance tests

To lock in recent improvements and accept a few

minor regressions.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.
Ignore failing perf tests until network issues are resolved

    • -0
    • +2
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 1 more file in changeset.
Bump minimum version of Gradle used for native perf tests

    • -0
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 1 more file in changeset.
Revert "Let Gradle 4.6 be the new performance baseline"

This reverts commit 9c1e61921d7b875af27d42a6c71918c6505d0902.

    • -1
    • +1
    ./RealWorldNativePluginPerformanceTest.groovy
  1. … 29 more files in changeset.