[perf] tags pool experiment

[lifeart]

This PR is create to check GC behaviour on framework benchmark if we start using tags and transactions pools.
Where transaction pool is retainable and tags pool is predefined

Idea here is:

• trying to reuse Tracker classes, including it's Sets and prepare initial pool (reusable)
• Prepare DirtyableTag initial pool (not reusble)
• Prepare UpdatableTag initial pool (not reusable)

In abstract ember app we always need some amount of Tracker, DirtyableTag and UpdatableTag classes.
And it may be a case where we will see GC improvements. And will be able to ship default pool sizes or allow developers to customize it from global variable or code replace functionality or babel plugin.

[lifeart]

locally, on 20 runs: (10k items rendering)

 Benchmark Results Summary    

duration phase estimated improvement -50ms [-64ms to -37ms] OR -1.66% [-2.12% to -1.25%]
renderStart phase estimated improvement -47ms [-61ms to -39ms] OR -2.7% [-3.47% to -2.2%]
glimmer-render-10000-rows-start phase no difference [-6ms to 7ms]
paint phase no difference [-1ms to 1ms]

ci, 100 runs: (10k items rendering)

    Benchmark Results Summary    
duration phase estimated improvement -126ms [-147ms to -108ms] OR -2.15% [-2.5% to -1.85%]
renderStart phase estimated improvement -117ms [-134ms to -101ms] OR -3.59% [-4.09% to -3.1%]
glimmer-render-10000-rows-start phase estimated improvement -12ms [-20ms to -3ms] OR -0.47% [-0.82% to -0.11%]
paint phase estimated regression +1ms [0ms to 2ms] OR +0.74% [0.07% to 1.42%]
[17:53:49] Generating Benchmark Reports [started]
[17:53:52] Generating Benchmark Reports [completed]

---

Full featured bench results (23 Dec 2023)

    Benchmark Results Summary    

duration phase estimated improvement -85ms [-117ms to -54ms] OR -1.43% [-1.96% to -0.91%]
render1000Items1End phase estimated improvement -5ms [-9ms to -1ms] OR -0.97% [-1.84% to -0.18%]
clearItems1End phase estimated improvement -6ms [-6ms to -5ms] OR -20.02% [-20.96% to -19.07%]
render1000Items2End phase estimated improvement -5ms [-8ms to -3ms] OR -1.33% [-1.94% to -0.7%]
clearItems2End phase estimated regression +1ms [1ms to 1ms] OR +4.61% [3.43% to 5.71%]
render10000Items1End phase estimated regression +35ms [14ms to 57ms] OR +0.93% [0.36% to 1.52%]
clearItems3End phase estimated improvement -3ms [-4ms to -2ms] OR -1.43% [-2.07% to -0.77%]
render1000Items3End phase estimated improvement -94ms [-98ms to -90ms] OR -21.91% [-22.73% to -21.05%]
append1000Items1End phase estimated improvement -35ms [-38ms to -31ms] OR -8.38% [-9.16% to -7.43%]
updateEvery10thItem1End phase estimated regression +7ms [7ms to 7ms] OR +42.53% [41.2% to 44.09%]
selectFirstRow1End phase estimated regression +3ms [2ms to 3ms] OR +7.22% [5.72% to 9.09%]
selectSecondRow1End phase estimated improvement -5ms [-5ms to -4ms] OR -38.11% [-41.3% to -34.92%]
removeFirstRow1End phase no difference [0ms to 2ms]
removeSecondRow1End phase estimated improvement -2ms [-3ms to -1ms] OR -9.21% [-14.38% to -4.86%]
swapRows1End phase estimated regression +3ms [1ms to 4ms] OR +4.38% [2.24% to 6.92%]
clearItems4End phase estimated regression +2ms [1ms to 2ms] OR +4.3% [2.66% to 5.95%]
paint phase no difference [0ms to 0ms]

artifact-1.pdf

2x pool size:

    Benchmark Results Summary    

duration phase estimated improvement -151ms [-175ms to -127ms] OR -2.52% [-2.9% to -2.11%]
render1000Items1End phase estimated improvement -12ms [-16ms to -7ms] OR -2.35% [-3.26% to -1.47%]
clearItems1End phase estimated improvement -6ms [-7ms to -6ms] OR -21.54% [-22.72% to -20.49%]
render1000Items2End phase estimated improvement -18ms [-21ms to -16ms] OR -4.6% [-5.2% to -3.98%]
clearItems2End phase estimated improvement -4ms [-4ms to -4ms] OR -15.49% [-16.43% to -14.59%]
render10000Items1End phase estimated improvement -111ms [-133ms to -88ms] OR -2.96% [-3.54% to -2.35%]
clearItems3End phase estimated improvement -2ms [-3ms to -1ms] OR -0.98% [-1.64% to -0.34%]
render1000Items3End phase no difference [-5ms to 4ms]
append1000Items1End phase no difference [-5ms to 1ms]
updateEvery10thItem1End phase estimated regression +2ms [2ms to 3ms] OR +14.41% [13.07% to 15.87%]
selectFirstRow1End phase no difference [0ms to 1ms]
selectSecondRow1End phase estimated improvement -4ms [-5ms to -4ms] OR -34.95% [-37.71% to -32.11%]
removeFirstRow1End phase no difference [-2ms to 0ms]
removeSecondRow1End phase estimated regression +7ms [6ms to 8ms] OR +40.89% [34.57% to 46.51%]
swapRows1End phase estimated regression +2ms [1ms to 3ms] OR +2.97% [1.48% to 4.2%]
clearItems4End phase estimated improvement -3ms [-3ms to -2ms] OR -6.7% [-7.97% to -5.37%]
paint phase no difference [0ms to 0ms]

artifact-1.pdf

[lifeart]

Dec 25, updated benchmark

    Benchmark Results Summary    

duration phase estimated regression +345ms [246ms to 441ms] OR +2.09% [1.49% to 2.67%]
renderEnd phase estimated improvement -3ms [-4ms to -1ms] OR -3.77% [-6.16% to -1.45%]
render1000Items1End phase estimated improvement -35ms [-47ms to -25ms] OR -2.73% [-3.64% to -1.95%]
clearItems1End phase no difference [-2ms to 3ms]
render1000Items2End phase estimated improvement -11ms [-18ms to -1ms] OR -1.03% [-1.77% to -0.13%]
clearItems2End phase estimated improvement -2ms [-4ms to 0ms] OR -3.25% [-6.38% to -0.71%]
render5000Items1End phase no difference [-32ms to 38ms]
clearManyItems1End phase estimated improvement -3ms [-6ms to -1ms] OR -1.32% [-2.43% to -0.21%]
render5000Items2End phase estimated regression +430ms [251ms to 520ms] OR +10.25% [5.97% to 12.4%]
clearManyItems2End phase estimated improvement -7ms [-10ms to -4ms] OR -2.82% [-4.16% to -1.52%]
render1000Items3End phase estimated improvement -19ms [-34ms to -16ms] OR -2.18% [-3.94% to -1.83%]
append1000Items1End phase estimated improvement -12ms [-20ms to 0ms] OR -1.27% [-2.02% to -0.05%]
append1000Items2End phase estimated improvement -17ms [-26ms to -10ms] OR -1.73% [-2.61% to -1%]
updateEvery10thItem1End phase no difference [-6ms to 2ms]
updateEvery10thItem2End phase no difference [-1ms to 0ms]
selectFirstRow1End phase no difference [-1ms to 0ms]
selectSecondRow1End phase estimated regression +6ms [0ms to 10ms] OR +4.48% [0.18% to 7.02%]
removeFirstRow1End phase estimated regression +12ms [8ms to 17ms] OR +4.65% [3.15% to 6.67%]
removeSecondRow1End phase estimated regression +13ms [7ms to 16ms] OR +5.22% [2.94% to 6.39%]
swapRows1End phase no difference [-2ms to 1ms]
swapRows2End phase no difference [0ms to 8ms]
clearItems4End phase estimated regression +13ms [10ms to 16ms] OR +9.25% [7.25% to 11.03%]
paint phase no difference [0ms to 0ms]

artifact-1.pdf

[lifeart]

Closing as not significant